Vous êtes sur la page 1sur 252

BONDSHIP

project guidelines
2005 Det Norske Veritas. All rights reserved
Title: BONDSHIP project guidelines
Editors: Jan R. Weitzenbck and Dag McGeorge
ISBN 82-515-0305-1
First edition 2005
The copyright of all figures used in this book is owned by the BONDSHIP project partners, except for the
following figures:
o 5-1 to 5-5 and 5-9: Professor Robert Adams, used with permission
o 8-1, 8-4 and 8-5: ASM International, used with permission
o 8-2 and 8-3: Hilger u. Kern GmbH, used with permission
Foreword
This document was produced by the members of the BONDSHIP project. BONDSHIP -
Bonding of lightweight materials for cost effective production of high speed craft and
passenger ships - is a 4.6 M GROWTH project funded by the European Commission under
the 5
th
framework programme. BONDSHIP is a major European initiative to introduce
adhesive bonding into shipbuilding as an industrial process for joining of lightweight and
dissimilar materials and structures.
The project ran from April 2000 to June 2003 and had 13 partners from 7 nations: (1) Det
Norske Veritas AS (project co-ordinator), Norway; (2) Fincantieri - Cantieri Navali Italiani
S.p.A., Italy; (3) Vosper Thornycroft (UK) Limited, United Kingdom; (4) Jos. L. Meyer
GmbH, Germany; (5) Alcan Mass Transportation Systems, Switzerland; (6) Sika Technology
AG, Switzerland; (7) CETENA S.p.A. - Centro per gli Studi di Tecnica Navale, Italy; (8)
Fraunhofer Gesellschaft zur Frderung der angewandten Forschung e.V., IFAM, Germany;
(9) FiReCo AS, Norway; (10) University of Southampton, United Kingdom; (11) Dlgation
Gnrale pour l'Armement - Direction des Centres d'Expertise et d'Essais, France; (12) NDT
Solutions Ltd, United Kingdom; (13) Stena Rederi AB, Sweden.
The guidelines can be considered the main deliverable of the BONDSHIP project. They sum
up all the steps necessary to design, build, inspect and repair all types of bonded joints in
ships. The guidelines describe a general framework for the safe use of bonded joints.
However, they do not provide detailed recipes for the user to follow. Hence additional
information is needed to successfully use adhesive bonding. There are two parts:
1) Code (DNV Report No. 2004-0134)
1
: The objective is to provide general requirements
to ensure the reliability and safety of load-carrying bonded joints in ships.
2) Recommended Practices (DNV Report No. 2004-0193): This document provides
guidance and examples on how to design, produce and inspect an adhesively bonded
joint. Furthermore it shall provide the basis for meeting the general requirements laid
out in the Code document.
Significant progress was made in BONDSHIP. However, most designers, builders and owners
of ships are not aware of the possibilities (and limitations) that adhesive bonding offers. The
BONDSHIP guidelines show how to safely introduce bonded joints, first in less critical areas
and increasingly also in more critical areas as service experience is gained and confidence in
the long term performance is built. By making the BONDSHIP guidelines available to the
public we hope to establish a broad user base for adhesive bonding in marine structures thus
paving the way for establishing adhesive bonding as a standard joining process in
shipbuilding.
Jan R. Weitzenbck Dag McGeorge
DNV Research DNV Structural Integrity & Laboratories
Hvik, Norway,
February 2005
1
The main ideas of this report have been published in: Jan Weitzenbck and Dag McGeorge, The designer's dilemma: How to
deal with the uncertainty about the long-term performance of adhesively bonded joints?, Proceedings of the Institution of
Mechanical Engineers Part M: Journal of Engineering for the Maritime Environment, 2004, vol 218, issue 4, p 273-276

TECHNICAL REPORT
DET NORSKE VERITAS
BONDSHIP GUIDELINES: CODE
REPORT NO. 2004-0134
REVISION NO. 02
DET NORSKE VERITAS
TECHNICAL REPORT
Rev2_Code_DNVRepNo_2004-0134.doc
DET NORSKE VERITAS AS
DNV Research
Veritasveien 1
1322 Hvik
Norway
Tel: +47 67 57 99 00
Fax: +47 67 57 99 11
http://www.dnv.com
Date of first issue: Project No.:
27.02.2004 91010302
Approved by: Organisational unit:
Carl Arne Carlsen
Head of DNV Research
DNV Research
Client: Client ref.:
European Commission G3RD-CT-2000-00101
Summary:
This document provides general requirements to ensure the reliability and safety of load-carrying bonded joints
in ships. This document applies to all types of adhesively bonded joints in ships. The document encompasses
the design, manufacture and use of bonded joints and is based on the following assumptions.
Joint design approval: Numerical analysis cannot reliably predict joint failure without additional large scale
tests. While numerical analysis can give extremely useful insights into the behaviour of bonded joints, the
approval of joint will only utilise representative tests as a cost-effective means of assessing bonded joints.
Long-term performance: This document is based on the assumption that the long-term performance of a
bonded joint cannot be reliably predicted from the results of accelerated ageing tests. Therefore, requirements
to the resistance of the joint are combined with requirements that limit the consequences of failure of the joint
and that it must be possible to repair the joint using an approved repair method.
Production: Adhesive bonding is a complex process with many variables whose interaction is not fully
understood. Furthermore, there are currently no NDT methods to measure reliably the adhesion strength of a
finished joint. Hence, instead of checking the quality of a joint afterward it is imperative to establish a quality
control system for each individual step of the production process to ensure joints can be produced reliably and
consistently.
Report No.: Subject Group:
2004-0134 E6, H1, H8
Indexing terms
Report title: Key words Service Area
Market Sector
BONDSHIP Guidelines: Code Adhesive bonding
Joining methods
Manufacturing
Guidelines
Ship design
Work carried out by:
Jan Weitzenbck, Dag McGeorge
Work verified by:
David Beresford
Date of this revision: Rev. No.: Number of pages:
31.01.2005 02 14
No distribution without permission from the
client or responsible organisational unit
(however, free distribution for internal use
within DNV after 3 years)
No distribution without permission from the
client or responsible organisational unit.
Strictly confidential
Unrestricted distribution
2002 Det Norske Veritas AS
All rights reserved. This publication or parts thereof may not be reproduced or transmitted in any form or by any means, including
photocopying or recording, without the prior written consent of Det Norske Veritas AS.
DET NORSKE VERITAS
Report No: 2004-0134, rev. 0
TECHNICAL REPORT
Page i
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
Table of Content Page
1 INTRODUCTION ....................................................................................................... 1
1.1 Background 1
1.2 Objective 1
1.3 Scope 1
1.4 Basic assumptions and concepts 1
1.5 Definitions 2
2 DESIGN OF ADHESIVELY BONDED JOINTS....................................................... 3
2.1 Outline 3
2.2 Risk reduction 3
2.3 Environment 3
2.4 Representative tests 4
2.5 Safety factors 5
3 FIRE............................................................................................................................. 6
3.1 Fire resistance of bonded joints 6
3.2 Non-load carrying fire divisions or elements 6
3.3 Load carrying fire divisions or elements 7
4 QUALITY AND SAFETY.......................................................................................... 8
4.1 Quality assurance and quality control (QA/QC) 8
4.2 Health and safety 8
5 QUALIFICATION OF BONDING PERSONNEL..................................................... 9
6 IN-SERVICE INSPECTION..................................................................................... 10
7 REFERENCES........................................................................................................... 11
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 1
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
1 INTRODUCTION
1.1 Background
This report is a deliverable of the BONDSHIP project. BONDSHIP - Bonding of lightweight
materials for cost effective production of high speed craft and passenger ships - is a 4.6 M
GROWTH project funded by the European Commission under the 5
th
framework programme.
The project ran from April 2000 to June 2003 and has 13 partners from 7 nations. BONDSHIP is
a major European initiative to introduce adhesive bonding into ship building as an industrial
process for joining of lightweight and dissimilar materials and structures.
The guidelines are a summary of the collective know-how and experience of the project partners.
The guidelines are split into two parts. Part 1 is the Code part (this document) setting safety
relevant requirements. The second part is a collection of recommended practices - how to e.g.
select adhesives, design and analyse joints [1]. The present document is a substantially revised
and extended version of references [2] and [3]. Reference [4] was a useful reference document in
the preparation of these guidelines.
1.2 Objective
The objective of this document is to provide general requirements to ensure the reliability and
safety of load-carrying bonded joints in ships.
1.3 Scope
This document applies to all types of adhesively bonded joints in Ships. The bonded joints can
be either structural or non-structural. The document encompasses the design, manufacture and
use of bonded joints.
1.4 Basic assumptions and concepts
Long-term performance: This document is based on the assumption that the long-term
performance of a bonded joint cannot be reliably predicted from the results of accelerated ageing
tests. Therefore, requirements to the resistance of the joint are combined with requirements that
limit the consequences of failure of the joint. Furthermore, it must be possible to repair the joint
using an approved repair method. This may not be necessary if past service experience
documents adequate long-term performance of the joint.
Joint design approval: Numerical analysis cannot reliably predict joint failure without additional
large scale tests. While numerical analysis can give extremely useful insights into the behaviour
of bonded joints, the approval of joint will only utilise representative tests (as defined in section
2.4) as a cost-effective means of assessing bonded joints. The above applies to new joint designs.
Approval of variations of existing design can probably be done on the basis of a numerical
analysis, especially when there is in-service experience with the old design.
Production: Adhesive bonding is a complex process with many variables whose interaction is not
fully understood. Furthermore, there are currently no NDT methods to measure reliably the
adhesion strength of a finished joint. Hence, instead of checking the quality of a joint afterward it
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 2
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
is imperative to establish a quality control system for each individual step of the production
process to ensure joints can be produced reliably and consistently. NDT methods can play an
important part in the QA/QC of the production processes [6].
1.5 Definitions
A bonded joint is a joint where adherends are bonded either by placing a layer of adhesive or
resin material between the adherends. The primary function of the adhesive is to transfer loads
from one adherend to the other. This distinguishes the adhesive from a sealant.
Quality is defined as the ability of the bond to meet the functional requirements defined by the
designer in the short and long term with the specified level of reliability.
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 3
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
2 DESIGN OF ADHESIVELY BONDED JOINTS
2.1 Outline
Qualification of the design of an adhesively bonded joint follows the following general
approach.
1. The consequences of failure of the bonded joint should be minimised through a
formal risk reduction procedure unless adequate long-term performance can be
documented from past successful service experience (Section 2.2).
2. The intended operating environment must be accounted for (Section 2.3).
3. Adequate resistance is to be documented using representative tests (Section 2.4).
4. Adequate safety factors are used (Section 2.5)
5. The loads and temperatures that the joint may experience in a fire must be accounted
for (Chapter 3).
Guidance Note: It is very difficult to define the terms structural and non-structural joints. In
the context of this document we are not making a distinction between the two. Hence all
requirements apply equally to both. However, as experience is growing, it may be possible to
relax the requirements for less-critical joints.
2.2 Risk reduction
In cases where adequate long-term performance of the bonded joint solution has not been
demonstrated by successful in-service experience, possible consequences of long-term
degradation and possible consequential damage shall be identified through a formal hazard
identification.
The identified hazards shall be reviewed and measures taken to control the risks. Redundancy
shall be provided such that the consequences of such failures are limited to economic losses.
Inspection and repair procedures shall be documented.
To limit the probability of severe long-term degradation and hence minimise risk, measures shall
be taken to ensure that, among available alternatives, those with unfavourable long-term
properties are discarded. This is to be based on accelerated screening ageing tests.
2.3 Environment
The following in-service environmental conditions shall be considered when defining the
characteristic strength of the joint:
x Humidity
x Temperature
x UV-light
x Chemicals (Chemicals may include fuels, cargo, effluents etc.)
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 4
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
x Welding
Guidance Note: It may be necessary to allow welding during construction and/or repair of
the structure. The use of bonded joints may reduce the possible scope for such work. Hence, a
requirement to the joint's ability to accept welding in the vicinity may be a necessary
requirement. This requirement can be treated as a requirement to resistance to a specified
environment condition (elevated temperature outside the range of normal operating
temperature). In addition the effect of thermal contraction or expansion of the surrounding
structure shall be considered.
For each environmental condition its variation over time shall be identified, i.e. whether the
exposure is permanent, variable or intermittent in nature, whether temperature is cycled through
r 0 qC etc.
Unless otherwise agreed all joints shall as a minimum be designed for the following range of
conditions:
x Relative humidity: 0 95 % *
x Temperature: 0qC - 40qC
* Submergence of the joint under water shall be considered if relevant.
Guidance Note: For adhesive selection and production further environmental conditions have
to be specified. The production related environmental factors will be addressed in chapter 4.
The recommended practices [1] present a systematic approach (IFAM tables) to specifying
all relevant conditions and requirements to the joint in a single exercise.
2.4 Representative tests
Tests complying with the following requirements are considered representative.
The fabrication of specimens must be representative of industrial fabrication at the shipyard.
Normally, the specimens should be prepared by the shipyard that shall manufacture the real
structure. Production shall as far as practically possible be carried out according to the
procedures that apply to such industrial fabrication at the shipyard. The following aspects should
be given particular consideration:
x Surface preparation of adherends
x Handling before priming
x Priming of the prepared surface
x Preparation of the primed surface before bonding operation
x Environmental conditions during all manufacturing steps
x Curing conditions
x Differences from normal shipyard productions that cannot practically be avoided when
small test objects are manufactured.
Deviations from the real joints must be documented to be conservative. Specifically, the mode
of loading must be representative of that the joint will experience in the structure when exposed
to the operational loads. Peel stresses must be included as relevant. Stress concentrations must
be at least as severe as in the real application unless fully accounted for by modelling.
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 5
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
The long-term degradation of the joint strength in the intended operational environment shall as
far as possible be represented by the measurements and explicitly accounted for in design.
Reversible effects, such as swelling and plasticization, should be accounted for separately from
permanent physical and chemical degradation of the joint.
Results from idealised tests may be used directly if they can be shown to be conservative.
Modelling may be used to document that the test results are representative for joint geometries
differing from the tested configuration. Furthermore, modelling may be used to extend
application of test results to loading modes not actually tested. In this way, test results may be
applied more widely and the degree of conservatism may be reduced by the aid of modelling.
2.5 Safety factors
A safety factor shall be adopted that is commensurate with the uncertainty in predicted long-term
joint resistance both due to variability in the state of as produced joints and in the effects of
environmental exposure they will receive in service and due to unavoidable differences between
the tested objects and the components in the real ship.
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 6
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
3 FIRE
This section is based on reference [5].
3.1 Fire resistance of bonded joints
The fire resistance of a bonded joint is usually inferior to the fire resistance of the adherend
material. When fire resistance is required the joint has also to be designed against relevant fire
scenarios. The following information shall be specified:
1. A description of the temperature development in the joint during a fire: e.g. maximum
temperature, duration etc. as relevant.
2. Requirements to strength of the joint during a fire
3. Requirements to the residual strength of the joint after the fire
The IMO documents The International Convention for Safety of Life at Sea (SOLAS)
including the Code of safety for High Speed Crafts (HSC-code), is the governing rules for fire
safety on board commercial vessels in international trade. They define load carrying and non-
load carrying fire divisions. These two main classes require different treatment.
Guidance Note: The fire reaction (heat release, smoke and toxicity) properties of the adhesive
do not represent any decrease in fire safety level. This is not only due to the small areas of
adhesive exposed to fire, but also the small amount used compared to other combustible
materials used in a ship (paint, decorative surfaces, deck covering, etc).
Guidance Note: If relevant, the effect of the fire extinguishing media on the properties of the
bonds shall be considered.
3.2 Non-load carrying fire divisions or elements
It must be demonstrated by fire tests or by engineering/layout that the bonded part(s) of the
division will stay smoke and flame tight during the required time. Mechanical fasteners or
arrangements can often replace the need for joint strength in a fire situation.
If the division is relaying on the structural integrity of the adhesive in order to stay gas and
smoke tight, this should be demonstrated by a test. Usually this will require full scale fire
resistance testing, but small scale testing can be evaluated depending on the type of joint.
If the bonded elements and the joint itself is part of a load carrying division, but is not a part of
the load carrying elements in this divisions (e.g. doors, hatches, fill-in elements, non-critical
joints, etc.) these shall be treated as a Non-load carrying division or element. It must be clearly
demonstrated that:
x The division will maintain structural strength without the adhesive joint and /or the
structural strength of the bonded elements for the required time
x The division will remain smoke and flame tight for the required time
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 7
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
3.3 Load carrying fire divisions or elements
The bonded joint is load carrying during the fire, and is critical for the function of the division.
An element that is carrying local load, but is a part of a non-load carrying division, can still be
classified as a load carrying element.
The following must be presented in addition to the information requested in section 3.1:
x Detailed load and stress analysis of the adhesive joint for all relevant load cases.
x The most critical load case or combination of load cases must be used to find worst-case
stresses for the bonded joint.
Two methods for qualification of the bonded joint can be used:
1. Characterise the thermo-mechanical properties of the material, find the allowable
temperature limit and measure temperature during large scale test.
2. Apply load during the full-scale fire test.
Guidance Note: Method 1 is equivalent to the philosophy that was used when aluminium was
allowed in shipbuilding.
Guidance Note: Method 2 is equivalent to the philosophy that was used when composite
structures were allowed in the construction of high-speed crafts.
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 8
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
4 QUALITY AND SAFETY
4.1 Quality assurance and quality control (QA/QC)
The yard or fabricator shall develop and implement procedures and practises controlling all
processes having an effect on the quality of the finished bond.
This section outlines the requirements to the content of such a QA documentation. Detailed
requirements are not given as these will depend on the actual joint configuration, choice of
materials etc., and will have to be worked out in detail for specific applications. More detailed
information will be provided in the RPs [1].
The QA documentation shall address the following topics:
x Joint specification
x Receipt and storage of materials
x Bonding operation including surface preparation
x Inspection and verification
x Quality plan
x Qualification of personnel (see chapter 5)
The QA documentation shall give requirements to all relevant items to a level of detail
commensurate with the complexity of the operations and to their influence on the quality.
4.2 Health and safety
Adhesives, primers and solvents are chemical substances that require careful handling. However,
provided precautions are taken they are safe to use. The following list shows the main safety
precautions to be taken (this list has to be modified to adapt to the local requirements):
x Follow safety data sheets provided by the adhesive or paint suppliers
x Follow regulatory requirement regarding maximum allowable concentration of volatile
gasses
x Provide protection to workers for skin, eyes and inhalation
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 9
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
5 QUALIFICATION OF BONDING PERSONNEL
The bonding operation and quality control shall be carried out by qualified personnel and
suitable equipment. Personnel shall be qualified according to the EWF (= European Federation
for Welding, Joining and Cutting; http://www.ewf.be/asp/) scheme or equivalent. Workers shall
be qualified to European Adhesive Bonder standard while supervisors shall meet the
requirements of the European Adhesive Specialist. Personnel responsible for planning and
implementing the bonding process in a company shall be qualified European Adhesive Engineers
(EAE).
Guidance Note: There are many training facilities in Europe that offer European Adhesive
Bonder and European Adhesive Specialist courses. However, at the time of writing of this
document there is only one institute that offers EAE training (IFAM, Bremen, Germany).
Hence, for a transition period the lack of formal EAE qualification can be compensated with
considerable and relevant experience.
All qualification schemes shall contain the following elements.
x Theoretical training
x Practical training
x Qualification exams, tests etc.
x Certification schemes
x Methods and requirements to maintenance of qualifications:
x Training
x Qualification exams
x Operator joint logbook
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 10
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
6 IN-SERVICE INSPECTION
The in-service inspection of bonded joints will be mainly visual inspection of the visible part of
the joint:
1. Check for corrosion of the adherend along the bondline (can lead to failure of the
joint, but not the adhesive!)
2. Are there (creep) cracks in adhesive?
3. An indirect check for (less critical) joints is whether or not leakage is observed on the
other side of the joint.
NDT methods can be used to investigate critical or suspect areas more thoroughly [6]:
x Is the adhesive still in contact with the adherend?
x Are there any defects at the interfaces and what size are they?
x If one can see visual evidence of damage what is the extent of the damage in the joint
(e.g. after impact damage)?
Any damage needs to be repaired using an approved repair method.
DET NORSKE VERITAS
Report No: 2004-0134, rev. 02
TECHNICAL REPORT
Page 11
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_Code_DNVRepNo_2004-0134.doc
7 REFERENCES
[1] Jan Weitzenbck, Dag McGeorge and Harald Osnes, BONDSHIP Guidelines:
Recommended Practices, 2004, DNV Report No.: 2004-0193
[2] Stefan Palm, Bonded joints in ships - code layout: structural design, BONDSHIP
Report Number: 2-23-W-2000-15-0, 2001-03-12
[3] Stefan Palm, Bonded joints in ships - code layout: QA/QC, BONDSHIP Report
Number: 2-23-W-2000-16-0, 2001-03-12
[4] DVS Merkblatt, Elastisches Dickschichtkleben im Schienenfahrzeugbau (in German:
Elastic Bonding of railway vehicles), Merkblatt DVS 1618, January 2002
[5] Bjrn Hyning, Final report on fire testing, BONDSHIP Report Number: 1-12-D-
2002-02-0, 2003-06-2
[6] Richard Freemantle, Non destructive adhesive bond inspection techniques and
guidelines for the BONDSHIP application cases, BONDSHIP Report Number: 2-24-
D-2001-01-4, 2003-06-18
- o0o -

TECHNICAL REPORT
DET NORSKE VERITAS
BONDSHIP GUIDELINES:
RECOMMENDED PRACTICES
REPORT NO. 2004-0193
REVISION NO. 2
DET NORSKE VERITAS
TECHNICAL REPORT
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS AS
DNV Reserach
Veritasveien 1
1322 Hvik,
Norway
Tel: +47 67 57 99 00
Fax: +47 67 57 99 11
http://www.dnv.com
Date of first issue: Project No.:
14.09.2004 91010302
Approved by: Organisational unit:
Carl Arne Carlsen
Head of DNV Research
DNV Research
Client: Client ref.:
European Commission G3RD-CT-2000-00101
Summary:
This report completements the BONDSHIP Code document (DNV Report No. 2004-0134). It
presents methods and actual examples (including data were possible) for the design, production and
in-service phase of an adhesivly bonded joint. The following topics are discussed:
- Specification of bonded joints
- Materials selection
- Failure criteria and characteristic strength values
- Design and analysis of bonded joints
- Testing of materials and structures
- Fire safety
- Production and repair of bonded joints
- Non-destructive inspection
Furthermore, procedures which are not easily found in the literature are presented in an appendix. The
intention of this report is to enable naval architects and other suitably qualified engineers to design
and produce safe and reliable adhesively bonded joints.
Report No.: Subject Group:
2004-0193 E6, H1, H8
Indexing terms
Report title: Key words Service Area
Market Sector
BONDSHIP Guidelines: Recommended
Practices
Adhesive bonding
Joining methods
Manufacturing
Guidelines
Ship design
Work carried out by:
Jan Weitzenbck, Dag McGeorge, Harald Osnes
Work verified by:
Stefan Palm
Date of this revision: Rev. No.: Number of pages:
31.01.2005 02 232
No distribution without permission from the
client or responsible organisational unit
(however, free distribution for internal use
within DNV after 3 years)
No distribution without permission from the
client or responsible organisational unit.
Strictly confidential
Unrestricted distribution
2002 Det Norske Veritas AS
All rights reserved. This publication or parts thereof may not be reproduced or transmitted in any form or by any means, including
photocopying or recording, without the prior written consent of Det Norske Veritas AS.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page i
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table of Content Page
1 INTRODUCTION ....................................................................................................... 1
1.1 Background 1
1.2 Objective 1
1.3 Scope 1
1.4 Link to code document 3
1.5 Definitions 3
1.6 How to use this document 3
1.7 References 3
2 SPECIFICATION OF BONDED JOINTS.................................................................. 4
2.1 Introduction 4
2.2 Application example: Specification of external casings 6
2.3 References 6
3 MATERIAL SELECTION.......................................................................................... 9
3.1 Introduction 9
3.2 Selection of suitable materials for screening tests 9
3.3 Selection of test methods 9
3.3.1 Introduction 9
3.3.2 Measurement of Tg 10
3.3.3 Measurement of pH value 10
3.3.4 Lap shear test - strength 10
3.3.5 Lap-shear test - strain to failure 10
3.3.6 Boeing wedge test 10
3.3.7 Measurement of electrical resistance 10
3.3.8 Bead test 10
3.4 Criteria for selection of suitable candidate systems 11
3.5 Application example: Screening tests for material selection 11
3.5.1 Introduction 11
3.5.2 Rigid adhesives 12
3.5.2.1 Introduction 12
3.5.2.2 Notes on lap shear test 12
3.5.2.3 Notes on the Boeing wedge test 12
3.5.3 Flexible adhesives 13
3.5.3.1 Introduction 13
3.5.3.2 Remarks on bead peel test 13
3.5.3.3 Remarks on pH measurements 13
3.5.3.4 Remarks on electrical resistance 14
3.5.3.5 Remarks on lap shear - strain to failure 14
3.5.4 Selection of adhesives 15
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page ii
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3.5.4.1 Introduction 15
3.5.4.2 Rigid adhesives 15
3.5.4.3 Flexible adhesives 15
3.5.4.4 Selected adhesives 16
3.5.5 Design values 16
3.5.6 Conclusions 16
3.6 References 17
4 FAILURE CRITERIA AND CHARACTERISTIC STRENGTH VALUES............ 18
4.1 Introduction 18
4.2 Fracture due to a single extreme load 18
4.3 Creep rupture due to sustained static loading 20
4.4 Fatigue fracture due to cyclic loading 20
4.5 Application of limit states to bonded joints with rigid and flexible adhesives 21
4.6 Knock-down factors 21
4.7 References 22
5 DESIGN AND ANALYSIS OF BONDED JOINTS ................................................ 23
5.1 Introduction to design of bonded joints 23
5.1.1 General aspects 23
5.1.2 Examples on joint configurations 24
5.1.3 General rules for methodical and geometrical design of adhesive joints 29
5.1.4 Main types of adhesive materials 32
5.1.4.1 Flexible adhesives 32
5.1.4.2 Rigid adhesives 33
5.1.5 Example of joint analysis using the method by Bigwood and Crocombe 33
5.2 Joint requirements 37
5.3 General design procedure for adhesive joints 37
5.4 Design of flexible adhesive joints 38
5.4.1 Pre-design - possible joint configurations and overlap length (phase 1) 38
5.4.2 Evaluation of the stresses and strains in the assembly (phase 2) 39
5.4.3 Application of failure criteria (phase 3) 41
5.4.3.1 Determination of strength limits and maximum stresses and strains 41
5.4.4 Design procedures for flexible adhesive joints in ships 44
5.5 Design of rigid adhesive joints 45
5.5.1 Pre-design - possible joint configurations and overlap length (phase 1) 45
5.5.2 Evaluation of the Stresses in the Assembly (phase 2) 47
5.5.3 Application of failure criteria (phase 3) 49
5.5.3.1 Failure criteria for the adherends 49
5.5.3.2 Failure criteria for the adhesive layer 49
5.5.3.3 Determination of strength limits and maximum stresses 50
5.5.4 Design procedures for rigid adhesive joints in ships 52
5.5.4.1 Design method of Brede 52
5.5.4.2 Design method based on non-linear finite element analysis. 54
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page iii
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
5.6 Application example: Optimisation of attachments below decks 55
5.7 References 62
6 TESTING OF MATERIALS AND STRUCTURES................................................. 63
6.1 Introduction 63
6.2 Materials characterisation tests 63
6.2.1 Thick adherend shear test (TAST) 63
6.2.2 Simple shear strength test 64
6.2.2.1 Introduction 64
6.2.2.2 Equipment 65
6.2.2.3 Samples 65
6.2.2.4 Definitions 66
6.2.2.5 Simplified method to measure strength for different adhesive systems 67
6.2.3 True stress / true strain behaviour, EModulus, Poissons ratio 67
6.2.3.1 Introduction 67
6.2.3.2 Samples 67
6.2.3.3 Test procedure 68
6.2.3.4 Definitions 68
6.2.4 Compression Tests Rhagava equivalent stress 69
6.2.4.1 Introduction 69
6.2.4.2 Samples and Testing 69
6.2.4.3 Definitions 70
6.2.5 Creep strength 71
6.2.5.1 Introduction 71
6.2.5.2 Requirements 71
6.2.5.3 Surfaces and samples 72
6.2.5.4 Testing 72
6.2.5.5 Stress rupture strength 73
6.3 Testing of full scale joint samples representative of real Application Cases 73
6.3.1 Introduction 73
6.3.2 General recommendations 73
6.3.3 Example 1: Bonded pillar 74
6.3.4 Example 2: Bonded joint in steel sandwich panels 74
6.4 Environmental degradation ageing and fatigue 75
6.4.1 Introduction 75
6.4.2 Major assumptions 76
6.4.3 Environmental factors 76
6.4.4 Representative testing 77
6.5 References 77
7 FIRE SAFETY........................................................................................................... 78
7.1 Introduction 78
7.2 Governing documents and general requirements 78
7.3 Fire engineering terms 79
7.3.1 Fire reaction 79
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page iv
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
7.3.2 Fire resistance 79
7.4 SOLAS requirements 79
7.4.1 General 79
7.4.2 Fire reaction requirements 80
7.4.3 Fire resistance requirements 81
7.4.3.1 General 81
7.4.3.2 A class divisions 81
7.4.3.3 B class divisions 82
7.4.4 Areas of fire resisting divisions 82
7.5 HSC-code requirements 82
7.5.1 Fire reaction requirement, and discussion 82
7.5.2 Fire resistance requirement 83
7.5.2.1 General 83
7.5.2.2 60 and 30 min fire resisting divisions 83
7.5.2.3 Areas of fire resisting divisions 84
7.6 Fire testing 84
7.6.1 Description of furnace 84
7.6.2 Description of loading arrangement 86
7.6.3 Test specimens and fire protection 86
7.7 Application example: aluminium casing bonded to steel deck 87
7.7.1 General 87
7.7.2 Description of fire test specimens 87
7.7.2.1 Materials used for test specimens 88
7.7.2.2 Loading and boundary conditions 89
7.7.2.3 Fire protection, panel -I 89
7.7.2.4 Fire protection, panel-II 90
7.7.3 Test results panel-I 90
7.7.3.1 Observation during test 90
7.7.3.2 Observations after test 90
7.7.3.3 Temperature measurement 91
7.7.3.4 Comments and discussion to the temperature registrations and results: 92
7.7.4 Test results panel-II 93
7.7.4.1 Observation during test 93
7.7.4.2 Observations after test 93
7.7.4.3 Temperature measurement 93
7.7.4.4 Comments and discussion to the temperature registrations and results: 94
7.7.5 Conclusions application example 95
7.8 Overall conclusions from BONDSHIP fire tests 95
7.8.1 Classification and testing of adhesive joints 95
7.8.2 Joint design 96
7.9 References 96
8 PRODUCTION OF BONDED JOINTS.................................................................... 97
8.1 Production procedures 97
8.1.1 Joint specification 97
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page v
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8.1.2 Receipt and storage of materials 97
8.1.2.1 Receipt and handling of materials and consumables 97
8.1.2.2 Storage condition of materials 97
8.1.3 Bonding operation 98
8.1.3.1 Working environment and handling 98
8.1.3.2 Surface preparation 98
8.1.3.3 Application of adhesive 98
8.1.3.4 Fit-up 98
8.1.3.5 Curing 99
8.1.4 Storage of finished bonds 99
8.1.5 Health and safety 99
8.2 Manufacturing technology 99
8.2.1 Introduction 99
8.2.2 Dispensing of adhesives 100
8.2.3 Manual adhesive processing 101
8.2.3.1 Introduction 101
8.2.3.2 Prepare workbench 101
8.2.3.3 Weighing and mixing of the adhesive 101
8.2.3.4 Apply adhesive 101
8.2.3.5 Automatic mixing and dispensing systems 102
8.2.4 Mechanised adhesive processing 103
8.2.4.1 Basic setup of a mixing and dispensing unit 103
8.2.4.2 Pumps suitable for adhesive delivery 103
8.2.4.3 Valves 104
8.2.4.4 Mixer 105
8.2.4.5 Ancillary equipment 105
8.2.5 Cure of adhesive 105
8.3 Application example: Production procedures for aluminium superstructure 106
8.3.1 Casing design 106
8.3.1.1 Bonded joint design 106
8.3.2 Material information 108
8.3.3 Surface preparation 109
8.3.3.1 Steel surfaces 109
8.3.3.2 Aluminium surfaces 109
8.3.4 Bonding procedures 110
8.3.4.1 Selection of bonding procedure 110
8.3.4.2 Selection of adhesive 110
8.3.4.3 Procedure for "spacers" selection 111
8.3.4.4 Description of bonding equipment 111
8.3.4.5 "Due point" 112
8.3.4.6 Bonding operation 112
8.3.4.7 Final surface refinement 113
8.4 Application example: Production procedures for bonding of lashing devices 114
8.4.1 Lashing device in mock-up 114
8.4.2 Lashing device samples for laboratory testing 115
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page vi
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8.5 List of references 116
9 REPAIR OF BONDED JOINTS ............................................................................. 118
9.1 Repair procedures 118
9.2 Application example: repair of aluminium superstructure 118
9.2.1 Background 118
9.2.2 Damages 119
9.2.3 Repairs 120
9.3 List of references 125
10 NON-DESTRUCTIVE INSPECTION.................................................................... 126
10.1 Overview 126
10.2 Adhesive bonding and NDT 126
10.2.1 Definition of bonding parameters 126
10.2.2 Definition of defect parameters 127
10.2.3 Description of available test technologies 128
10.2.3.1 Acoustic methods 128
10.2.3.2 Radiography 128
10.2.3.3 Shearography 129
10.2.3.4 Thermography 129
10.2.3.5 Ultrasonics 129
10.2.3.6 Visual 130
10.3 Procedures, documentation, requirements and application 130
10.3.1 Procedures and documentation 130
10.3.2 Production and in-service requirements 131
10.3.3 NDT application areas for adhesive bond inspection 131
10.3.3.1 Design impacts 131
10.3.3.2 Mechanical testing, materials selection 131
10.3.3.3 Production 131
10.3.3.4 In-service 132
10.4 Ultrasonics equipment and inspections 132
10.4.1 Ultrasound basics 132
10.4.2 Inspection types 134
10.4.2.1 Pulse echo 134
10.4.2.2 Through transmission 135
10.5 Application examples 135
10.5.1 WP 2: Acceptance and qualification of joints task 2.2: Selection of
materials 135
10.5.1.1 Adhesive and adherend characterisation 135
10.5.1.2 Adhesive joint simulations 137
10.5.1.3 Coupon testing 138
10.5.2 WP3: Application case 1 - task 3.4: Performance of joints and critical
defects 139
10.5.2.1 University of Southampton monitoring of joints 140
10.5.2.2 University of Southampton monitoring of joint displacement 141
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page vii
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
10.5.2.3 VT test panel inspection 143
10.5.2.4 VT mock-up inspection 146
10.5.2.5 VT defect test sample inspection 149
10.5.3 WP4: Application case 6 casing 151
10.5.4 WP5: Application cases at Meyer shipyard 153
10.5.4.1 AC1 secondary load bearing wall structure 154
10.5.4.2 AC7 gutter ways 156
10.5.4.3 AC9 balcony structure 158
10.6 Conclusions 161
10.7 References 161
11 DEFINITIONS AND PROCEDURES.................................................................... 163
11.1 Test procedures 163
11.1.1 Measurement of pH value (IFAM test standard WP-AA-60) 164
11.1.2 Lap shear test - strength 165
11.1.2.1 Sika flexible adhesives 165
11.1.2.2 CTA rigid adhesives 165
11.1.3 Lap-shear test - strain to failure (IFAM test standard WP-AA-11) 167
11.1.4 Boeing wedge test (ASTM D 3762-98) 170
11.1.4.1 Principle of the method 170
11.1.4.2 Geometrical quantities recommended 170
11.1.4.3 Preparation of the specimens 171
11.1.4.4 Wedges: 171
11.1.4.5 Calculation of the wedge thickness : 171
11.1.4.6 Calculation of the length of the not bonded area : 172
11.1.5 Measurement of electrical resistance (Sika Test Procedure 316) 173
11.1.6 Bead test (Sika SQP033-0 and Sika SQP034-0) 177
11.2 Analytical analysis methods for joints 182
11.2.1 Parameters for single-lap joints 182
11.2.2 Analysis using nominal stresses and strains 183
11.2.3 Analyses of Volkersen and Goland Reissner 185
11.2.4 Approach of Bigwood Crocombe 187
11.2.5 Analysis of Wiedemann pre-design, overlap length of joints 188
11.3 Numerical analysis methods the finite element method 196
11.3.1 Linear finite element analysis 197
11.3.2 Non-linear finite element analysis 197
11.3.3 Finite element analysis for flexible adhesive joints 199
11.3.4 Finite element analysis for rigid adhesive joints 202
11.4 Non-destructive inspection 206
11.4.1 Glossary 206
11.4.2 Standards 207
11.4.2.1 General standards 207
11.4.2.2 Radiological methods 207
11.4.2.3 Ultrasonic methods 208
11.4.2.4 Visual inspection methods 208
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page viii
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.4.3 Inspections and operators 208
11.4.4 List of equipment manufacturers 210
11.4.4.1 Acoustic methods 210
11.4.4.2 Radiography 210
11.4.4.3 Shearography 210
11.4.4.4 Thermography 210
11.4.4.5 Ultrasonics 210
11.4.4.6 Visual 211
11.4.5 NDT test procedure - rigid adhesive bondline measurement 212
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page ix
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table of Figures Page
Figure 1-1: Outline of the design and production process for bonded joints............................. 2
Figure 2-1 Typical external casings in the upper deck of a cruise ship ..................................... 6
Figure 5-1: Joining of sheet materials in tension; the higher the load rating, the stronger
the joint (from [1]).................................................................................................................... 25
Figure 5-2: Possible T-joints and efficiency of the joint concerning the specified loads.
(from [1]).................................................................................................................................. 26
Figure 5-3: Possible corner joints and efficiency of the joint concerning the specified
load (from [1]).......................................................................................................................... 27
Figure 5-4: Tubular joints for axial and torsional loading (from [1]) ...................................... 28
Figure 5-5: Stiffener attachment and load spreading (from [1]) .............................................. 28
Figure 5-6: Load flux and principle of similar structural strength........................................... 29
Figure 5-7: Principle of corresponding deformation................................................................ 30
Figure 5-8: Symmetric loading reduces secondary load effects .............................................. 30
Figure 5-9: Additional design features to avoid or reduce peel-stress (from [1]).................... 30
Figure 5-10: An example for a self-reinforcing adhesive joint ................................................ 31
Figure 5-11: Unstable design of an adhesive joint ................................................................... 32
Figure 5-12: Input window of the Microsoft Excel spreadsheet .............................................. 34
Figure 5-13: Shear and peel stresses for a rigid adhesive single-lap joint with k=0 and
k=1............................................................................................................................................ 35
Figure 5-14: Stress distributions for steel and aluminium with a rigid adhesive..................... 36
Figure 5-15: Stress distributions for steel and aluminium with a flexible adhesive ................ 36
Figure 5-16 General design procedure for adhesive joints ...................................................... 37
Figure 5-17: Linear-elastic numerical analysis for lap-shear test ............................................ 42
Figure 5-18: Failure criteria from maximum shear and peel stress at the joints ends ............ 43
Figure 5-19: Procedure to determine a failure criterion and accuracy of numerical results .... 43
Figure 5-20: Attachments: Identification of hot spots ............................................................. 55
Figure 5-21: Case study of attachment below deck ................................................................. 60
Figure 5-22: Stress distribution at upper end of adhesive joint, x = const ............................... 61
Figure 5-23: Stress distribution along adhesive joint at w/2.................................................... 61
Figure 6-1: Thick adherend shear test samples used for shear experiments. ........................... 64
Figure 6-2: Simple shear testing equipment............................................................................. 65
Figure 6-3: Sample geometry for simple shear test.................................................................. 66
Figure 6-4: Pure adhesive sample for tensile testing................................................................ 68
Figure 6-5: Compression test sample....................................................................................... 70
Figure 6-6: Form and dimensions of test specimen. ................................................................ 72
Figure 6-7: Bonded pillar sample prepared for testing (left) and sketch of test setup
(right)........................................................................................................................................ 74
Figure 6-8: Bonded joint between steel sandwich panels (left) and four-point bending
test setup (right)........................................................................................................................ 75
Figure 6-9: Schematic of steel sandwich joint tested in tension. ............................................. 75
Figure 7-1: Small-scale furnace at FiReCo. The picture shows a FRP/sandwich panel
during test. ................................................................................................................................ 85
Figure 7-2: Standard fire curves for furnace fire testing (BS 476 curve is identical to
ISO 834/IMO Res 754(18)....................................................................................................... 86
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page x
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 7-3: Aluminium casing bonded to steel deck................................................................ 87
Figure 7-4: Test assembly on furnace, with load shear load applied (to the right).................. 88
Figure 7-5: Fire test specimen with thermocouple positions indicated (TC#) ......................... 88
Figure 7-6: Weld pins for fire insulation mounting ................................................................. 89
Figure 7-7: Insulation of specimen (right), and edge protection of specimen prior to fire
test (left). .................................................................................................................................. 90
Figure 7-8: Sample just after joint failure. The adhesive failed near to the aluminium
panel, but there were still plenty of adhesive left on both the aluminium and steel part. ........ 91
Figure 7-9: Adhesive left on the aluminium part ..................................................................... 91
Figure 7-10: Temperature readings during test of panel-I. ...................................................... 92
Figure 7-11: Temperature readings during test of panel-II ...................................................... 94
Figure 8-1 Dispensing of adhesive on surface: (a) extruded beads, (b) droplets, (c)
spray; (from [3]) ..................................................................................................................... 100
Figure 8-2 Commercial metering system (DOPAG).............................................................. 102
Figure 8-3 Handheld dispensing gun with disposable static mixer (DOPAG) ...................... 102
Figure 8-4 Automated application of adhesive to wind screens, [5]...................................... 103
Figure 8-5 Variable ratio metering and mixing machine, [4] ................................................ 104
Figure 8-6: Drawing of half aluminium casing ceiling.......................................................... 107
Figure 8-7: Bonded joint lay-out: 3D view............................................................................ 107
Figure 8-8: Bonded joint scantlings ....................................................................................... 108
Figure 8-9: Adhesive curing process without "booster" ........................................................ 111
Figure 8-10: one component adhesive pump with an add-on Kit for supplying, dosing
and mixing Sika Booster-Paste to Sikaflex -254................................................................ 112
Figure 8-11: Filling of the joint.............................................................................................. 113
Figure 8-12: Final surface refinement .................................................................................... 113
Figure 8-13: Set-up bonding AC5 mock-up components ...................................................... 114
Figure 8-14: Set-up bonding AC5 IFAM components........................................................... 115
Figure 9-1: Bonded joint configuration.................................................................................. 118
Figure 9-2: Casing mock up - Outside View.......................................................................... 118
Figure 9-3: Removing of the adhesive ................................................................................... 119
Figure 9-4: Bonded line to be replaced .................................................................................. 119
Figure 9-5: Carbonisation of a portion of bonded line ....................................................... 120
Figure 9-6: Bonded line before the repair .............................................................................. 120
Figure 9-7: Filling of Sikaflex 292 Booster ........................................................................... 121
Figure 9-8: stopper-band at the middle of the joint .............................................................. 122
Figure 9-9: Filling of Sikaflex 292......................................................................................... 122
Figure 9-10: Repairing of the joint with Sikaflex 292 ........................................................... 123
Figure 9-11: Adhesive burning .............................................................................................. 123
Figure 9-12: Removing of the carbonised adhesive............................................................... 124
Figure 9-13: Filling of Sikaflex 292....................................................................................... 124
Figure 9-14: Superficially Repair of the joint with Sikaflex 292........................................... 124
Figure 10-1: Basic adhesive joint topology............................................................................ 127
Figure 10-2: Typical defect parameters and location in joint. .............................................. 127
Figure 10-3: Principles of ultrasound wave propagation in liquids and solids. .................... 132
Figure 10-4: Ultrasound propagation across interfaces.......................................................... 133
Figure 10-5: Conversion of electrical pulses into ultrasound pulses. .................................... 134
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page xi
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-6: Pulse echo ultrasound inspection. ..................................................................... 134
Figure 10-7: Through transmission ultrasound inspection..................................................... 135
Figure 10-8: Test arrangement used for adhesive characterisation........................................ 136
Figure 10-9: Repeat compression and shear wave ultrasonic attenuation measurements
for Araldite 420. ..................................................................................................................... 136
Figure 10-10: Comparison of compression wave attenuation data for rigid and flexible
adhesive employed in the BONDSHIP materials selection process. ..................................... 137
Figure 10-11: Joint simulations for a composite to aluminium bonded and disbonded
joint......................................................................................................................................... 138
Figure 10-12: Measured data from an intact coupon and a delaminated bond which had
suffered cohesive failure in the adhesive. .............................................................................. 139
Figure 10-13: Pictures of Vosper Thornycroft application case. ........................................... 139
Figure 10-14: VT test structure showing principal joints. ..................................................... 140
Figure 10-15: Ultrasonic C-scan of DBC sample after DBC test. ......................................... 141
Figure 10-16: Instrumentation and test assembly using through transmission ultrasound
to monitor joint....................................................................................................................... 142
Figure 10-17: Through transmission compression wave data versus time during bend
test. ......................................................................................................................................... 142
Figure 10-18: Test sample sent for laser shearography testing showing primary test
areas........................................................................................................................................ 143
Figure 10-19: LTI 5100 Laser shearography system. ........................................................... 143
Figure 10-20: Shearography image of upper section of centre inter panel joint.................... 144
Figure 10-21: Shearography image of lower section of centre inter-panel joint. .................. 145
Figure 10-22: Shearography image of lower left-hand corner panel. ................................... 145
Figure 10-23: Photograph of VT mock-up during inspection work....................................... 146
Figure 10-24: RapidScan large area ultrasound inspection system........................................ 147
Figure 10-25: RapidScan inspection of inter-module butt-strap joints.................................. 147
Figure 10-26: Repeat ultrasonic scans of butt-strap joint with diagram of scanned area. .... 148
Figure 10-27: Rippling of adhesive which could give rise to observed patchiness. .............. 148
Figure 10-28: Large impact area, ultrasonic size (16 mm), visible size (10 mm) ................. 149
Figure 10-29: Barely visible impact damage (BVID), ultrasonic size (8 mm), visible size
(4 mm) .................................................................................................................................... 149
Figure 10-30: Defect test sample containing a range of simulated defects............................ 150
Figure 10-31: Location of defects in test sample. .................................................................. 150
Figure 10-32: Resultant ultrasonic C-scan of defect test sample. .......................................... 150
Figure 10-33: Fincantieri AC6 - engineering drawings and photograph of test joint. .......... 151
Figure 10-34: Photograph of measurement of aluminium interface and resultant test
data. ........................................................................................................................................ 152
Figure 10-35: Photograph of measurement of steel interface and resultant test data. ........... 152
Figure 10-36: Photograph of measurement of bulk joint and typical test data. ..................... 153
Figure 10-37: Photograph of measurement of bulk adhesive and typical test data................ 153
Figure 10-38: Meyer Werft application 1, engineering drawings and photographs .............. 154
Figure 10-39: Test grid for external wall lip.......................................................................... 155
Figure 10-40: Typical bonded and disbonded data reading from flaw detector. ................... 155
Figure 10-41: Colour map of bonded and disbonded regions................................................ 156
Figure 10-42: Meyer Werft application 7, engineering drawings and photographs .............. 157
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page xii
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-43: Testing of gutter way and close-up of flaw detector display. ......................... 158
Figure 10-44: Adhesive bead width measurement showing extent of bonding on gutter
way. ........................................................................................................................................ 158
Figure 10-45: Meyer Werft application 9, engineering drawings and photographs .............. 159
Figure 10-46: Typical ultrasound data with automated thickness reading to assess
adhesive bondline. .................................................................................................................. 160
Figure 10-47: Diagram showing measurement locations and recorded bondline
thicknesses.............................................................................................................................. 160
Figure 11-1 Parameters to describe a single lap joint ............................................................ 183
Figure 11-2 Single-lap joints and bending moment factor..................................................... 187
Figure 11-3 Adherend-adhesive sandwich with applied moments and forces....................... 187
Figure 11-4 Possible configurations for joints described with the approach of Bigwood
and Crocombe ........................................................................................................................ 188
Figure 11-5 t
max
/o
10
as a function of the parameter , o = 1.................................................. 190
Figure 11-6 Ratio of shear stress distribution to the applied tensile stress according to
Volkersen as a function of the overlap length........................................................................ 190
Figure 11-7: Minimum overlap length as a function of the thickness of the adherends,
d=0.3 mm............................................................................................................................... 191
Figure 11-8: Tensile stress in the adherends vs. thickness of the adherends, d= 0.3 mm...... 191
Figure 11-9: Shear stress distribution at F = 12480 N according to Volkersen..................... 193
Figure 11-10: Shear stress distribution at F = 12480 N according to Bigwood and
Crocombe ............................................................................................................................... 193
Figure 11-11: Shear stress and peel stress distribution at F = 12480 N according to
Bigwood and Crocombe......................................................................................................... 194
Figure 11-12: Decreasing stress peaks at s
1
> 0.8 mm and F=12480 N................................. 195
Figure 11-13: A lap shear sample for local analysis modelled with h-elements (2D) ........... 197
Figure 11-14: A lap shear sample for local analysis modelled with p-elements (3D) ........... 198
Figure 11-15: Nominal and true stress-strain of a flexible adhesive...................................... 200
Figure 11-16: Ogden potential strain energy function to describe hyperelastic, flexible
adhesive Sikaflex 360 HC from first benchmark................................................................... 201
Figure 11-17: Potential strain energy functions to describe hyperelastic, flexible
adhesive Sikaflex 360 HC from first benchmark................................................................... 202
Figure 11-18: True-stress true-strain behaviour of Vantico Araldite 420 at room-
temperature............................................................................................................................. 203
Figure 11-19: Description of the material model as input to the finite element code............ 204
Figure 11-20: Numerical simulation and experimental data of lap shear test........................ 204
Figure 11-21: Von Mises equivalent stress............................................................................ 205
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page xiii
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table of Tables Page
Table 2-1: Main characteristic features for defining requirements for a bonded joint .............. 5
Table 2-2: Application example: specification of external casing ......................................... 7
Table 3-1 Shear strength and glass transition temperature for rigid adhesives with metal
adherends (* = only 72h post-cure; ** = not possible to measure Tg) .................................... 11
Table 3-2 Results from Boeing wedge test on rigid adhesives with metal adherends (see
also section 3.5.2.3).................................................................................................................. 13
Table 3-3 Summary of results for flexible adhesives; grey shading indicates that the
adhesive is below the specified threshold value....................................................................... 14
Table 5-1: Typical mechanical constants for an epoxy adhesive and a rubber like,
flexible adhesive....................................................................................................................... 34
Table 5-2: Recommended analysis methods for rigid adhesive joints..................................... 47
Table 5-3: Knock-down factors for designing adhesive joints ................................................ 54
Table 5-4: Experimental results of lap-shear-tests to determine failure criteria for rigid
and semi-rigid adhesives and used surfaces / adherends.......................................................... 57
Table 5-5: Failure criteria for Vantico Araldite 420 (A) and Plexus MA 550 (P), see also
previous table ........................................................................................................................... 58
Table 5-6: Comparison of failure criterion with numerical results of the case study.............. 59
Table 7-1: Observations during first test run ........................................................................... 90
Table 7-2: Observations during second test run....................................................................... 93
Table 8-1: Main characteristic features to work out a list of requirements............................ 114
Table 8-2: Steps of production procedure AC5 Vantico Araldite 420 mock-up ................... 115
Table 8-3: Bonding parameter AC5 IFAM test specimen ..................................................... 116
Table 8-4: Steps of production procedure AC5 Vantico Araldite 420 IFAM test
specimen................................................................................................................................. 116
Table 10-1: Guidance list of NDT techniques and capabilities. ........................................... 130
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 1
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
1 INTRODUCTION
1.1 Background
This report is a deliverable of the BONDSHIP project. BONDSHIP - Bonding of lightweight
materials for cost effective production of high speed craft and passenger ships - is a 4.6 M
GROWTH project funded by the European Commission under the 5
th
framework programme.
The project ran from April 2000 to June 2003 and has 13 partners
1
from 7 nations. BONDSHIP
is a major European initiative to introduce adhesive bonding into ship building as an industrial
process for joining of lightweight and dissimilar materials and structures.
The guidelines are a summary of the collective know-how and experience of the project partners.
The guidelines are split into two parts. Part 1 is the Code part [1] setting safety relevant
requirements. The second part (this document) is a collection of recommended practices - how to
e.g. select adhesives, design and analyse joints. This document is a substantially revised and
extended version of references [2] and contains some elements from references [3] and [4].
The three authors contributed to different parts of this report. Harald Osnes wrote chapter 5
Design and analysis of bonded joints and sections 4.6 Knock-down factors, 11.2 Analytical
analysis methods for joints and 11.3 Numerical analysis methods the finite element method
while Dag McGeorge authored chapters 4 Failure criteria and characteristic strength values
(except for 4.6) and 6 Testing of materials and structures. Jan Weitzenbck wrote the remaining
chapters and was responsible for overall co-ordination.
1.2 Objective
The objective of this document is to provide guidance and examples on how to design, produce
and inspect an adhesively bonded joint. Furthermore it shall provide the basis for meeting the
general requirements laid out in the Code document [1].
1.3 Scope
This document applies to all types of adhesively bonded joints in Ships. The bonded joints can
be either structural or non-structural. The document encompasses the design, manufacture and
use of bonded joints. Figure 1-1 shows a summary of the total process for new building and in-
1
BONDSHIP project partners:
1) Det Norske Veritas AS (project co-ordinator), Norway;
2) Fincantieri - Cantieri Navali Italiani S.p.A., Italy;
3) Vosper Thornycroft (UK) Limited, United Kingdom;
4) Jos. L. Meyer GmbH, Germany;
5) Alcan Mass Transportation Systems, Switzerland;
6) Sika Technology AG, Switzerland;
7) CETENA S.p.A. - Centro per gli Studi di Tecnica Navale, Italy;
8) Fraunhofer Gesellschaft zur Frderung der angewandten Forschung e.V., IFAM, Germany;
9) FiReCo AS, Norway;
10) University of Southampton, United Kingdom;
11) Dlgation Gnrale pour l'Armement - Direction des Centres d'Expertise et d'Essais, France;
12) NDT Solutions Ltd, United Kingdom;
13) Stena Rederi AB, Sweden.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 2
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
service. This diagram is not meant as a complete description of how to design and build a ship
structure. Its aim is merely to highlight the adhesive-specific aspects of that process.
In-service:
New building:
Specification of
joint
requirements
Select materials
Joint design and
analysis
Material & joint
testing
Production
and QA/QC
Fire testing of
bonded
structure
NDT
(integrated in
production
process)
IFAM tables
(specified
requirements!!)
Compatible
with
specifiation?
Structure is
ready for use!
Repair of
bonded joints
Inspection of
bonded joints
= number of chapter for
further reference
Input from all stake holders:
designers, end users, production
department, approval bodies
NO
YES
Repair
Joint meets all
requirements?
Very important:
requirements set by
class or USCG!
YES
NO
Mistake/
modification?
YES
NO
10
9
6*
9
X
8
2
5
6
7
2
3
* of code
document
Figure 1-1: Outline of the design and production process for bonded joints
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 3
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
1.4 Link to code document
The code part [1], as was discussed in section 1.1, is setting safety relevant requirements. It is
important to point out that the joint approval will initially only utilise representative tests. This is
mainly because numerical analysis cannot reliably predict joint failure without additional large
scale tests. The above applies to new joint designs. Approval of variations of existing designs
can probably be done on the basis of a numerical analysis, especially when there is in-service
experience with the old design.
The following list shows how the code part relates to the chapters in the present document.
- 2. Design of adhesively bonded joint
2.2 Risk reduction: partly in section 5.1 (but not formal hazard identification)
2.3 Environment: chapter 2
2.4 Representative tests: section 6.3
2.5 Safety factors: chapter 4
- 3. Fire: chapter 7
- 4. Quality and safety: section 8.1
- 5. Qualification of bonding personnel: -
- 6. In-service inspection: sections 9.1, 10.3, 10.4 and 10.5
1.5 Definitions
A bonded joint is a joint where adherends are bonded by placing a layer of adhesive or resin
material between the adherends. The primary function of the adhesive is to transfer loads from
one adherend to the other. This distinguishes the adhesive from a sealant.
Quality is defined as the ability of the bond to meet the functional requirements defined by the
designer in the short and long term with the specified level of reliability.
1.6 How to use this document
References used in a chapter are collected and presented at the end of each chapter.
Any data or test results published in this document are only valid for the particular joint
configuration, materials and test conditions used in the experiment. For example other adherend
materials or joint geometries can lead to considerable differences in the results. Hence the
information provided here serves only as example or illustration. It is necessary to repeat the
tests for each new case to obtain valid data relevant for the application case of interest.
1.7 References
[1] Jan Weitzenbck and Dag McGeorge, BONDSHIP Guidelines: Code, DNV Report
No.: 2004-0134, 2004
[2] Philippe Noury, Structure and table of content of guidelines, BONDSHIP Report
Number: 2-23-D-2001-02-0, 2001
[3] Stefan Palm, Bonded joints in ships - code layout: structural design, BONDSHIP
Report Number: 2-23-W-2000-15-0, 2001
[4] Stefan Palm, Bonded joints in ships - code layout: QA/QC, BONDSHIP Report
Number: 2-23-W-2000-16-0, 2001
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 4
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
2 SPECIFICATION OF BONDED JOINTS
2.1 Introduction
This section was adapted from reference [1]. The first step in the design of adhesively bonded
joints is to work out a detailed list of requirements to determine all features that could affect the
behaviour and performance during the lifetime of the joint (G. Pahl, W. Beitz,
Konstruktionslehre, 4. Auflage, Berlin (1997)). Table 2-1 shows a full set of main characteristic
features as a starting point to prepare a detailed list of requirements for any application case. The
list of requirements with its main features is described in Table 2-1. It can be used as a formal
instrument to establish the joint requirements during all phases in the life of the structure from
conceptual development until break-up and scraping or recycling. The definitions of the main
features are not always strictly correct in the sense of their physical meaning, but they have to be
understood in a more general sense with the purpose to cluster the requirements and to gain a full
set of information.
The main features which have to be discussed during the mechanical engineering part of the
design process are in particular:
Geometry, including all geometric data of the structure around the adhesive joint.
Kinematics, including all accelerations and deformations of the structure which might create
forces in the joint.
Forces, all forces and loads at the joint including the stiffness of the surrounding structure.
Energy, influence of temperature which might directly change any material properties or
generate strain due to different thermal expansion factors of materials.
Material, including physical, mechanical and chemical properties and any change of these
properties due to ageing.
Verification, any testing and design methods like numerical and analytical methods,
standards and regulations.
Maintenance, including protection against possible damage and applicable repair methods.
A possible format of a list of requirements is shown in Table 2-2. In this table the main
characteristic features of Table 2-1 are listed and numbered. The particular information has to be
filled in during several iterative steps. The first step should be taken by the responsible designer
of the joint. Later on, the list should be completed during several meetings with responsible
personal from all departments involved in designing, testing and fabrication of the joint. If all
relevant information is collected through the whole design process, and even through lifetime, a
list with valuable know-how will be developed. In other words, this document will become an
excellent tool for knowledge management. While it may be awkward and cumbersome to use the
first time, it will greatly simply future joint designs and modifications of existing ones. It is
important to note that the items presented in Table 2-1 present an ideal list; for a specific joint
one has to select a subset of factors that are relevant to the bonded joint under consideration.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 5
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 2-1: Main characteristic features for defining requirements for a bonded joint
Main characteristic
feature
Examples
1 Geometry
Size, height, width, length, diameter, space requirements,
arrangement, completion and extension,...
2 Kinematics Kinematic quantities, kind of motion, moving direction, rate of
motion, speed, acceleration,...
3 Forces Quantity of force, direction of force, frequency of force, weight,
load, deformation, stiffness, spring properties, stability, resonance,...
4 Energy Power, efficiency, loss, friction, storage, pressure, temperature,
resilience, transformation of energy,...
5 Material Physical, mechanical and chemical properties, prescribed materials
and process materials,...
6 Signal Input and output signal, kind of signal,...
7 Safety Safety engineering, protection systems, operating safety, protection
of labour, environmental protection,...
8 Ergonomics Man-machine-relationship: operation, control, operating facility,
clearness, illumination, design,...
9 Production Restrictions due to: manufacturing place, preferred manufacturing
method, dimension limits, manufacturing facilities, possible quality
and tolerance,...
10 Verification Measuring and testing methods, special regulations, numerical
modelling, calculation and verification,...
11 Assembly Special regulations for assembly, assembly, handling,...
12 Transport Restrictions due to necessary transport (size, weight, ...), kind and
conditions of forwarding,...
13 Application With little noise, rate of wear, use, range of application, place of
application, operating temperature, other operating environmental
conditions (i.e. humidity),...
14 In-service:
Maintenance, repair
Maintenance-free or easy to maintain, maintenance accessory,
maintenance rate and cost, replacement, corrective maintenance,
coat of paint, cleaning,...
15 Recycling Reutilization, repeated application, waste disposal, final waste
disposal, elimination,...
16 Costs Admissible production costs, tooling costs, investment,
amortisation,...
17 Deliverables Time schedules, dead line date of development, intermediate steps,
time of delivery,...
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 6
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
2.2 Application example: Specification of external casings
The following example was taken from reference [2]. An external aluminium or composite
casings will be bonded to a steel or aluminium superstructure on the upper deck (see Figure 2-1).
Figure 2-1 Typical external casings in the upper deck of a cruise ship
2.3 References
[1] Markus Brede, Final report on analytical and FE modelling of joints and verification
of easy-to-use design rules, BONDSHIP Report Number: 1-11-D-2002-01-0; 2003
[2] Jan Weitzenbck, Basic specification of application cases, BONDSHIP Report
Number: 2-21-D-2000-01-0; 2001
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 7
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 2-2: Application example: specification of external casing
List of Requirements M=must have
D=Desire
Project-No. Product Worked out by
399042mb AC 6: External casings
Requirements
M No. Designation Quantities, Data Responsible
D Comments
1.1 Geometry: Aluminium alloy or composite GRP casings bonded to
steel or aluminium superstructure.
See picture P1-AC6 Finc
D 1.2 Geometry: The joint must take into account deck's curvatures. Finc
M 1.3 Geometry: Deck's plates may undergo local deflection (out of
planarity) due to welding process.
Finc
2.1 Kinematics: Longitudinal acceleration of cruise ship.
Depending on ship dimensions and on the position of the casing.
1.8 m/s
2
Finc
DNV
2.2 Kinematics: Transversal acceleration at upper deck of cruise ship.
Depending on ship dimensions and on the position of the casing.
2.9 m/s
2
at
amidships
4.5 m/s
2
at hull ends
Finc
DNV
2.3 Kinematics: Vertical acceleration at upper deck of cruise ship.
Depending on ship dimensions and on the position of the casing.
2.3 m/s
2
at
amidships
7.5 m/s
2
at hull ends
Finc
DNV
M 2.4 Kinematics: The above value of acceleration (points 2.1, 2.2, 2.3)
are relevant to a large cruise ship.
L=260 m
2.5 Kinematics: Longitudinal acceleration of mono-hull fast ferry.
Depending on ship dimensions and on the position of the casing.
3.5 m/s
2
Finc
DNV
2.6 Kinematics: Transversal acceleration at upper deck of fast ferry.
Depending on ship dimensions and on the position of the casing.
6.6 m/s
2
at
amidships
9.8 m/s
2
at hull ends
Finc
DNV
2.7 Kinematics: Vertical acceleration at upper deck of fast ferry.
Depending on ship dimensions and on the position of the casing.
5.3 m/s
2
at
amidships
9.8 m/s
2
at hull ends
Finc
DNV
2.8 Kinematics: The above value of acceleration (points 2.5, 2.6, 2.7)
are relevant to a mono-hull fast ferry.
L=140 m V=40 knots
2.9 Kinematics: Roll angle. 20 Finc, DNV
2.10 Kinematics: Vibration induced by equipments. Frequency range up
to 80 Hz 4.0 mm/s amplitude velocity.
(ISO 6954) Finc,
CET
3.1 Forces: Self weight and inertia loads.
This should be considered in conjunction to transverse heeling
angle.
Due to gravity and to
above accelerations
Finc
CET
3.2 Forces: Weight of equipments.
This should be considered in conjunction to the transverse heeling
angle.
Due to gravity and to
above accelerations
Finc
CET
3.3 Forces: Wind pressure. 0.52.4 KN/m
2
Finc
M 4.1 Energy: Deformations due to temperature variation.
Max. and minimum values and excursion.
-20/60 C (if placed
in external decks)
Finc
5.1 Materials: Ship superstructures in steel or aluminium.
For steel: FE510 / Shot-blasting surface (SA 2.5) / With
CERABOND primer.
For aluminium: plates in alloy 5083 or 5383 H321 or H111 skin
passed, Epoxy primer for external surfaces.
Finc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 8
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
5.2 Materials: for casings: steel, aluminium.
As above reported.
Finc
5.3 Materials: for casings: composites
Fibre ROVIMAT 1200
Resin Polyester isophtalic unsaturated
SIRESTER 150868/T / SYNOLITE 0288-T-1
Catalysts: TRIGONOX 61 / Accel.+NL49P
Tixotropic: NEOTIX or SIRTIX SP9
Gel coat NEOGEL NPG A
Catalysts: BUTANOX M50
All the possible
combinations of the
materials reported in
points 5.1, 5.2, 5.3,
must be taken into
consideration.
Finc
6.1 Signals: NOT RELEVANT
M 7.1 Safety: Fire, smoke emission, toxicity. To be specified DNV, FiReCo
7.2 Safety: Protection of joint toward environmental exposure,
If needed.
Adhesives
suppliers
7.3 Safety: Additional safety equipments in case of failure of bonded
joints.
Depending on
ageing behaviour
Finc
7.4 Safety: General considerations. Existing Regulation
and Rules
Finc,
DNV
7.5 Safety: Safety of personnel during fitting up operations.
** The additional safety equipments relevant point 7.3 can be
studied in order to make the assembling operations easier.
Mechanical
equipment for
assembly.
Emission from
adhesives
Finc
Adhesives
suppliers
8.1 Ergonomics: aesthetic functions. Owner
requirements.
Finc
9.1 Production: Bonding processes suitable to be employed in
shipyard conditions.
Outdoor conditions,
whether seasons.
Finc
9.2 Production: Bonding processes should be conditioned as less as
possible from shipyard conditions.
Dust, oils, silicones. Finc
10.1 Controlling: Assessment of the good condition of the substrates. Indication by
Adhesives suppliers
Finc
10.2 Controlling: Set up of procedures to control the quality of joints
during production in the shipyard.
Specify of
parameters to be
controlled.
Adhesives
suppliers
10.3 Controlling: Set up of procedures to control the quality of joints
during the ship life.
Specify of NDT
Tests
Adhesives
suppliers,
NDT
11.1 Assembly: Geometric tolerances. Stating of values Finc
12.1 Transportation: Protection of treated surfaces. Finc
13.1 Employment: World wide on different types of ship. Finc
14.1 Maintenance: good accessibility for controlling and repairing. Finc
15.1 Recycling: Definition of procedures for dismantling of ships. Adhesives
suppliers
16.1 Cost: Convenient referred to conventional joints. Finc
Agreed Date Casing - Table 2
16 October 2000 Revision 2
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 9
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3 MATERIAL SELECTION
3.1 Introduction
The main instrument for selecting the materials used in the design of bonded joints is the
screening test (this chapter has been adapted from reference [1]). Screening tests are usually
carried out at the beginning of the design process. The screening test programme is used to
reduce the vast number of possible combinations of adhesives, primers, paints and other surface
preparations available for different adherend materials such as steel, aluminium and composites.
Tests are chosen to obtain relevant test data for selecting materials in a cost and time efficient
manner. Common to all tests is that the specimens are simple and cheap to produce. The test
programme is usually divided according to the type of adhesive: (i) rigid and (ii) flexible
adhesives. The main reason for this is that many tests only work with a certain type of adhesive
due to their specific mechanical properties. For example the bead test is only suitable for flexible
adhesives as it requires that adhesive beads are lifted and pulled off the surface with a pair of
pliers.
3.2 Selection of suitable materials for screening tests
The substrates or adherend and also in some cases the surface finishes (paints, primers,
sandblasted,) are predetermined by the application case. For example aluminium is used in the
superstructure to save weight. These materials and surfaces can be determined using the table of
requirements presented in chapter 2. The results of this survey can also be used to compile a list
of requirements for selecting the adhesives. Based on these requirements adhesive suppliers may
be asked to recommend suitable adhesives. The list of requirements should include:
- Materials / surfaces of joint
- Geometry and load of joint (e.g. minimum bondline thickness and variation)
- Temperature in-service and fire safety requirements
- Environment in production and in service
- Application of adhesive (e.g. how quickly, minimum open time)
- Curing conditions
- Required lifetime of joint
- Special requirements to repairability
It is quite likely that at this stage relatively little is known about the strength requirements of the
joint as one has no or only a vague idea about the detailed joint geometry and loading. Hence the
adhesive selection is mainly based on general factors such as the anticipated service environment
and manufacturing requirements.
3.3 Selection of test methods
3.3.1 Introduction
The methods described in this section are commonly used for screening purposes. Many of the
test procedures are reproduced in section 11.1, including deviations from the standards.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 10
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3.3.2 Measurement of Tg
Many room temperature curing (rigid) adhesives have a glass transition temperature (Tg) of
around 50
o
C to 70
o
C. It is therefore important to establish the precise Tg in order to select a
suitable temperature for accelerated ageing tests. These measurements are not applicable to
flexible adhesive (e.g. PUR) as their Tg is well below 0
o
C. The applicable test is the torsion-
pendulum test according to ASTM E 1356 or ISO 6721.
3.3.3 Measurement of pH value
The measurement of the pH value of all adhesives is important since experience has shown that
some adhesive can be quite acidic or alkaline under the long-term influence of water. This can
create corrosion problems in the joint. It is therefore important to select adhesives with a
"neutral" pH value. The test procedure follows the IFAM test standard WP-AA-60.
3.3.4 Lap shear test - strength
This is the most common test used for all types of adhesives. Lap shear strength is measured
before and after ageing to assess the ability of the adhesives to withstand adverse environments.
The relevant test standards are: ASTM D 1002, DIN EN 1465, ISO 4587.
3.3.5 Lap-shear test - strain to failure
Flexible adhesives show creep when loaded. This test is a slightly modified standard lap-shear
tests where a constant displacement (strain) is applied. The details of the tests are documented in
the IFAM test standard WP-AA-11. The aim of this test is to asses the ability of the adhesive to
sustain strain.
3.3.6 Boeing wedge test
The Boeing wedge test (ASTM D 3762-98) was selected to assess the durability of the bonding
system - the complete surface preparation and coating and the adhesive. This test is only used for
rigid adhesives.
3.3.7 Measurement of electrical resistance
The measurement of electrical resistance of all adhesives is important to make sure that all
selected adhesives have sufficient specific resistance. This will ensure that the adhesive layer
acts as electrical insulator and prevents electrochemical corrosion not only at the joint but
anywhere in the structure. The applicable test procedure is the Sika Test Procedure 316
"Determination of electrical specific volume resistance".
3.3.8 Bead test
The bead test (Sika SQP033-0, Sika SQP034-0) was selected to assess the durability of the
bonding system - the complete surface preparation and coating and the adhesive. This test is only
used for flexible adhesives.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 11
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3.4 Criteria for selection of suitable candidate systems
Criteria for selecting adhesives are derived from the requirements obtained from the survey using
the table compiled in chapter 2. Furthermore they have to be specific to the selected test(s). In all
cases one is looking for good (or the "best") performance of an adhesive, surface or adhesive
system in an aggressive environment. It is difficult in most cases to define an absolute minimum
performance limit. It is therefore up to the designer to make the final selection.
3.5 Application example: Screening tests for material selection
3.5.1 Introduction
The example shown in this section was taken from reference [1]. It is important to note that the
test results are given as illustration only as discussed in section 1.6.
Table 3-1 Shear strength and glass transition temperature for rigid adhesives with metal
adherends (* = only 72h post-cure; ** = not possible to measure Tg)
Lap shear strength (in MPa) and type of failure (A = adhesive failure, C =
cohesive failure, Comp = failure in composite)
glass transition
temperature (in
o
C)
Room temperature, not aged
aged at
70C
aged at 70C and 95% RH
Steel Alu 5083 Alu 6082 GRP Alu 6082 Steel Alu 5083 Alu 6082 GRP
Initial
After 624 h
at 70C
1,7 3 3,4 4,3 - 1,7 3,2 2 5 Teromix
6700
70% A 50% A A C - 50% A. 50% A 50% A C
57 53 *
12,1 14 5,2 8,9 - 9,6 7,4 1,8 9,1 2
K

-

P
o
l
y
u
r
e
t
h
a
n
e
Sika MG
1045.1 A A A A - 70% A.* A A A
61,5 < Tg
< 73
78 *
9,6 9,7 10 10,5 11,1 11,3 10,5 11,9 11 Plexus
MA550
C C C C C C C C C
~ 121
135 < Tg <
144
9,2 8,2 11 10,4 7,8 8,6 8,4 7,8 9,8
Lord 660
C A 70% A. 80% C 80% A C A A 80% C.
>112 >136
17 17,8 17,5 12,4 21,7 14,8 15,5 14,6 12,7
Lord 410
C C C C A C C C Comp
106 < Tg <
117
122
17,6 19,1 19,2 13 21,2 13,4 14,8 15,3 13,4 Permabond
6050
C C C C C C * 80% C C C
128 >145
21 21,4 23,6 8,4 25,9 16,4 13,6 19,7 11,6
Loctite 341
C 80% A 60% A. Comp A A A A Comp
141 ------ *
24,5 26 25,9 6,5 23,2 15,7 18,2 18,8 11,9
A
c
r
y
l
i
c

(
m
o
s
t

a
r
e

m
o
d
i
f
i
e
d

-

n
o
t

L
o
c
t
i
t
e

3
4
1
)

Loctite
3295
50 to
80% A
A A Comp A A A A Comp
>143
134,5 < Tg
< 145
14,7 10,4 12,3 11,7 11,8 - 5,3 6 12,2
WRA 4303
C A A Comp A C A A Comp
88 98
20,2 19,4 19,2 7,5 22,4 9,3 15 13,7 8,4 Araldite
2015
C C A A A C A A Comp
83 < Tg <
97,5
114 < Tg <
122
23,7 20,8 22,3 7,5 22,2 16,3 16,7 17,1 11,6
3M 1838
A A 70% A Comp A A A A Comp
75 < Tg <
81,5
88,5
26 26,4 22,8 8,1 28,5 12,6 17,8 19 13,4 Permabond
32
C C 80% A Comp A 50% A A A Comp
72 < Tg <
7 6,5
100
20,3 23,2 23 10,6 21,1 14,3 14,7 13,3 12,4
3M 490
C 60% C 50% A. Comp C A A A Comp
78 < Tg <
92
95
25,4 25,4 24,9 9,4 26,4 22 24,3 20,2 9,3 Araldite
420
A A A Comp A A A A Comp
~72 < 111,5
31,5 25,8 31,8 10,1 - 25,7 - 23,8 11,2
E
p
o
x
y

(
W
R
A

i
s

m
o
d
i
f
i
e
d
)

Hysol 9395
80% C 50% A A Comp A A A A Comp
** 192
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 12
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3.5.2 Rigid adhesives
3.5.2.1 Introduction
The lap shear and Boeing wedge test were used to study the rigid adhesives. In addition the glass
transition temperature was studied. All the results have been summarised in Table 3-2 and Table
3-1. The substrates were steel, aluminium (5083 and 6082) and FRP. The surfaces were either
galvanised (steel), primed with weld through primer or painted with epoxy, polyurethane or
acrylic topcoat and the appropriate primer. The cleaning methods used were acetone wipe and
sand or grit blasting or a combination thereof. Details about the results obtained with the
individual tests can be found in the following sections.
3.5.2.2 Notes on lap shear test
Basically all adhesives fail adhesively on aluminium substrates when exposed to 70
o
C and 95%
relative humidity for a period of 624 hours. The only exceptions are the following (acrylic)
adhesives: Plexus MA 550, Lord 410 and Permabond 6050. In case of the Lord 410 adhesive
failure was observed for high temperature ageing but not when it was combined with humidity.
Composite lap shear joints failed almost always in the composite. It was observed that low
strength adhesives fail in the adhesive layer while high strength adhesives fail in the composite,
usually the first ply. The strength of the joints increases after ageing for those adhesives which
fail in the composite. This effect is attributed to post curing of the vinyl ester matrix resin of the
composite.
All acrylic and epoxy adhesives have a Tg of more than 70
o
C, in most cases it is more than
100
o
C as shown in Table 3-1. Only the two polyurethane adhesives Sika MG 1045,1 and
Teroson Teromix 6700 have a Tg of less than 70
o
C. This could help to explain the poor ageing
performance of the two adhesives.
3.5.2.3 Notes on the Boeing wedge test
The wedge test allows the assessment of the complete adhesive - surface system. It was therefore
disappointing to note that not a single sample failed cohesively. Failure occurred either at the
adhesive-adherend interface, the primer/paint-adherend interface or within the primer or paint. It
should be noted, however, that the thickness of the steel (5 mm) and aluminium (6 mm) was
thicker than the usual thickness used for the wedge test (3 mm). To compensate for the higher
adherend thickness thinner wedges were used to maintain the same fracture energy (see also
section 11.1.4 on page 170).
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 13
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 3-2 Results from Boeing wedge test on rigid adhesives with metal adherends (see also
section 3.5.2.3)
Summary of test results
Araldite 420 only 50% of the specimens stay intact until the end of tests - exhibit consistent
crack growth
Steel: intact specimens for washed or painted surfaces (VIII, XI, XIII). Failure
for "Black" and painted (I, II, XII).
Alu: grit blasted and painted failed (III, VI, XVIII), painted surfaces intact: XV,
XXI, XXII
Lord 410 fairly consistent crack growth for painted steel surfaces (I, II, XIII) - in one case
specimen failed after one day (XI).
only one result for painted alu (XXI) same crack length as steel
Permabond 6050 2 out of the 3 specimens fail before the end of the test
Alu: grit blasted and painted surface fail (III, XVIII), grit blasted 6082 intact
after 72h (IV)
no results for steel
Loctite 3295 results are not consistent
alu: 2 samples fell apart within 48 hours (XV, XXII), 4 surfaces display crack
growth (sand blasted and painted surfaces; III, IV, XVIII, XXI)
steel: 3 surfaces failed within 24 hours (I, II, XI), 3 had crack lengths
comparable to aluminium (VIII, XII, XIII)
3.5.3 Flexible adhesives
3.5.3.1 Introduction
This section summarises the results for the flexible adhesives shown in Table 3-3. The following
tests were used for evaluating the adhesives:
- bead peel test
- pH measurements
- electrical resistance
- lap shear - strain to failure
- lap shear - strength
3.5.3.2 Remarks on bead peel test
The following criteria had to be passed by all adhesives listed in Table 3-3.
- need to have at least 75% cohesive failure
- the results obtained by the two test houses had to agree
The bead peel test is only suitable for flexible adhesives. Hence both Plexus adhesives were not
tested because of their (comparatively) high strength and stiffness.
3.5.3.3 Remarks on pH measurements
As discussed in section 11.1.1 the pH value of the adhesives should be within a certain range to
avoid corrosion of the adherends:
- steel: always to be protected against the impact of water
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 14
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
- aluminium: pH 6 to pH 8
- polycarbonate: pH 4 to pH 10
3.5.3.4 Remarks on electrical resistance
An adhesive should have a specific electrical volume resistance of at least 10 MOcm to avoid
electrochemical corrosion (see also section 11.1.5). This value is approximately the average of
the resistance of an insulator and a conducting material. Only the Sikaflex 265 did not reach this
value. The Terostat-8590 gave inconsistent results.
3.5.3.5 Remarks on lap shear - strain to failure
The criterion for passing this test was, that adhesive failure occurred in less than 5% of the
bonding area. Samples were tested at room temperature. Some were tested after being stored for
1 day at 30C and 1 day storage at room temperature. All samples were subjected to constant
tensile strains at 10% or 30% for a duration of 3 weeks. Of the 4 adhesive which did not qualify,
two adhesives, Betamate VP71570 and Sikaflex 292 only just failed the 30% strain after ageing
test.
Table 3-3 Summary of results for flexible adhesives; grey shading indicates that the
adhesive is below the specified threshold value
Bead peel (suitable
surfaces, see also
[1])
pH
value
electrical
resistance
[MOcm]
lap shear -
strain to
failure
lap shear - strength
[MPa] and standard
deviation [MPa]
1. EXP Betamate
VP71570
II, III, IV, IX, XII,
XIII, XV, XVI,
XVIII, XXVI
5,5
SA
> 10 n 1,55 0,37
2. P-Bond III, XI
5,5
SA
> 10 n 1,61 0,22
3. Sikaflex-292 III - XXII, XXIV
6,0
S
> 10 n 1,08 0,17
4. Sikaflex 222 UV see Sikaflex 295 UV
5. Sikaflex-295 UV III - IX, XI - XV,
XVII - XXII
6,0
S
> 10 y 1,26 0,23
6. Sikaflex-265 III - V, IX, XII -
XXII
5,0
SA
1-2 - -
7. Sikaflex 552 VP III, IV, VIII, XIII -
XXII
5,5
SA
> 10 y 1,28 0,20
8. Sikaflex-291 I - III, XI - XXII
6,0
S
> 10 y 1,03 0,07
9. Terostat-8590
UHV/M
I - III, IX - XXII,
XXIV
6,0
S
> 10 &
0,2
y 3,63 1,31
10. Terostat-MS 9380
no suitable surface 7,5
S
> 10 n 1,69 0,25
11. Plexus MA 420 -
3,5
SAF
> 10 - 6,38 1,57
9,29 2,88
12. Plexus MA 550 -
3,5
SAF
> 10 - 9,40 0,42
8,88 1,36
Notes:
- pH value: S = corrosion protection required for steel, A = c. p. required for aluminium, F =
c. p. required for FRP;
- strain to failure: n = did not meet criterion described in section 3.5.3.5, y = met criterion
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 15
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3.5.4 Selection of adhesives
3.5.4.1 Introduction
The screening experiments yielded a huge amount of data which can make interpretation of the
results difficult. To start the selection one should use the specification of the ACs (e.g.
deliverable 2.1 D1 Basic specification of applications cases) and decide on the most important
criteria and select the substrates which need to be bonded. Next adhesives and surface
preparations are identified which are suitable for a given substrate. The choice whether flexible
or rigid adhesives are to be used is based e.g. on the expected build tolerances, predicted loads
and deformation of the joint, own know-how and previous experience and other factors. The
selection is further refined by comparing the performance of the adhesives in the various ageing
tests (Table 3-2, Table 3-1 and Table 3-3) with the specification of the ACs. Additional factors
which need to be considered include costs and manufacturing requirements such as pot life and
viscosity.
3.5.4.2 Rigid adhesives
Table 3-1 shows the lap shear strength with and without ageing and the glass transition
temperature. The aluminium and steel surfaces were sandblasted. Only the three acrylic
adhesives, Plexus MA550, Lord 410 and Permabond 6050, exhibit cohesive failure for all ageing
conditions. The other adhesives failed adhesively in one or more instances after ageing,
especially for the aluminium surfaces. For steel there were two epoxy adhesives (Araldite 2015
and WRA 4303) which gave cohesive failure after ageing. The other adhesives failed in a
predominantly adhesive failure mode.
The wedge test results are summarised in Table 3-2. All painted surfaces failed in the paint or
primer layer. Metallic surfaces (unpainted, either degreased or grit blasted) failed all adhesively.
There are at least two possible ways of interpreting the results: (i) The results suggest that for a
strong and durable joint primers and paints have to be removed from metallic adherends. In
addition there is a need for a proper surface preparation since degreased and grit blasted surfaces
all failed adhesively. (ii) The observed failure mode is not due to an inferior material
performance but can be attributed to modifications of the specimen geometry (see section
3.5.2.3) - hence it is the test specimen which is causing the poor results rather than the selected
materials. It is not possible to say which of these two views are correct here.
3.5.4.3 Flexible adhesives
All relevant results for the flexible adhesives can be found in Table 3-3. It was concluded that
the two Plexus adhesives (550 and 420) were not suitable for the bead peel and strain to failure
test. Their behaviour was much more similar to that of the rigid adhesives. For example their lap
shear strength is considerably higher than the other adhesives investigated here. Furthermore it
was observed that these two adhesives pull the paint or primer off the substrate surface. Another
4 adhesives were found not to be suitable for the BONDSHIP ACs: Betamate VP71570, P-Bond,
Sikaflex-265 and Terostat-MS 9380.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 16
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3.5.4.4 Selected adhesives
The following adhesives were selected for further study in the project upon completion of the
screening test programme and following discussion with the project partners. The selection was
partly based on the partner's previous experience with some of the adhesives, the need to have at
least one adhesive for each AC and of course the performance in the screening tests:
- Flexible adhesives
Sikaflex 292
Sikaflex 552
Terostat 8590 UHV/M
- Rigid adhesives
Araldite 420
Lord 410
- Other
Plexus 550
The selected flexible adhesives work on all the substrates/ surfaces specified for the ACs. For
rigid adhesives it was noted in reference [2] that not all the selected adhesives are suited for all
the application cases under consideration. For example the Araldite adhesive has been found to
be suitable only for 2 ACs.
3.5.5 Design values
This screening experiment was meant for selecting suitable adhesive - surface/substrate
combination for use in the ACs of the BONDSHIP project. Hence, most of the results are only of
a qualitative nature to be able to rank different candidate materials. Only the lap shear - strain to
failure results can be used for joint design. The other values are not directly applicable.
However, it would be possible, for example, to use the static lap shear test results to calibrate FE
simulations and back-calculate a critical shear strength of that adhesive. However, this was not
the aim of this study.
3.5.6 Conclusions
From this screening test programme it is possible to draw some general conclusions. These
conclusions are only valid for this particular test programme and the chosen adhesives,
surfaces (treatments) and substrate materials:
- The strength of the adhesives was not used as selection criterion as the different ACs have
very different strength requirements. It can be noted, however, that rigid adhesives achieve
much higher strength than flexible adhesives even if they fail in an adhesive failure mode.
- Flexible adhesives can bond most surfaces used in the screening test programme including
painted surfaces when used with the correct surface preparation.
- For high strength (rigid) adhesives the joint strength was usually limited by the adhesion
strength of the primer or paint. Hence to utilise the full potential of these adhesives it seems
that weld through primers and paints need to be removed first.
- There are questions about the current surface preparation methods for aluminium and steel
joints made with rigid adhesives. The wedge test results seem to suggest that these are not
sufficient to ensure a reliable long-term performance. However, there is also the possibility
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 17
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
that this effect is caused by the non-standard geometry of the specimens. This needs to be
addressed in the remainder of the BONDSHIP project.
- This example compares only mechanical performance of adhesives and does not consider
manufacturing related issues such as viscosity, open time or pot life and cure conditions.
These factors must be taken into account before making the final selection for each adhesive.
3.6 References
[1] Jan Weitzenbck, Selection of adhesives and substrate materials (workhorse
materials), BONDSHIP Report Number: 2-22-D-2001-02-1; 2002
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 18
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
4 FAILURE CRITERIA AND CHARACTERISTIC STRENGTH VALUES
4.1 Introduction
The basis for qualifying an adhesive joint is a set of failure criteria (limit state equations) that
represent all the critical failure mechanisms. These failure criteria contain at least one
characteristic strength parameter, other material properties, load effects and safety factors. The
use of these failure criteria requires that the following items be consistently addressed:
1. The load effects occurring in the joint must be established from the loads that act on
the structure by a suitable load-response analysis of the structure that contains the
bonded joint. Some tests to obtain the material properties required for such modelling
can be found in Section 6.2.
2. The characteristic strength parameters that enter the failure criteria must be reliably
obtained from representative tests
2
. Some suitable test methods can be found in
Section 6.2.
3. The load effects and characteristic strength parameters are combined with appropriate
safety factors in the failure criterion in order to establish whether the joint is reliable.
General requirements for how to identify the critical failure mechanisms and corresponding
failure criteria (limit state equations) can be found in the reference [1]. In the following sections,
three of the most relevant failure criteria for bonded joints are presented:
1. Fracture due to a single extreme load
2. Creep rupture due to sustained static loading
3. Fatigue fracture due to cyclic loading.
In each section, recommendations are provided for how to establish the characteristic load effect
and strength parameters. Appropriate values for safety factors are indicated and compared to
those commonly encountered. Furthermore, a discussion is provided of how these criteria apply
to typical adhesives.
4.2 Fracture due to a single extreme load
The loading to be considered should represent an extreme load covering all phases of the
structure and the intended lifetime of the structure.
2
Definition of the characteristic value according to DNV Offshore standard OS-C501 (Section 4):
A 602 The characteristic value is a nominal value to characterise a stochastic variable. The characteristic
value of a mechanical property is usually a value, which has a small probability of not being exceeded in a
hypothetically unlimited test series.
A 603 The characteristic value of a strength property is defined in this standard as a low 2.5% quantile in
the distribution of the arbitrary strength. This is equivalent to the 97.5% tolerance. For more details see B400 and
C1100.
A 604 The characteristic value for stiffness shall be taken as the mean value in the distribution of the
arbitrary value of the stiffness property.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 19
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
In this section, it is assumed that the failure criterion can be expressed in terms of a single
characteristic strength value. For joints with a tough rigid adhesive, this parameter can typically
be a critical fracture strain. One may conservatively assume that the plastic strains are small and
hence use a critical elastic stress together with a linear elastic model of the joint. This would
normally be quite appropriate for bonded joints with elastic adhesives. Alternatively, one may
establish the joint strength directly from tests of full scale joint samples. This provides three
variants of the failure criterion where, in all cases,
m
is a partial material factor:
o In terms of strains (here shear strain):

fc
/
m
>
max

where
fc
is a characteristic value of the measured strain at fracture and
max
is the
maximum predicted strain in the joint in response to environmental actions.
o In terms of stresses (here shear stress):
t
fc
/
m
> t
max

where t
fc
is a characteristic value of the measured stress at fracture and t
max
is the
maximum predicted stress in the joint in response to environmental actions.
o In terms of fracture load of joint samples:
P
fc
/
m
> P
max
where P
fc
is a characteristic value of the measured load at fracture and P
max
is the
maximum predicted loading of the joint in response to environmental actions.
The maximum load effects (
max
, t
max
or P
max
) must be calculated taking due account of any
partial factors that represent uncertainties in environmental loads and model uncertainties.
In the latter case, the load effect P
max
is direct output of an analysis of the response of the
structure to the extreme loading events. If an elastic adhesive is used, it may be required to take
account of the flexibility of the joint in the analysis. Apart from that, a conventional structural
response analysis would suffice. The value of P
fc
is obtained from the measured joint strengths
as described below.
When fracture prediction is based on stresses or strains occurring within the bonded joint, the
stress or strain distribution in the joint must be established by an appropriate theoretical model.
Models for this purpose are described in chapter 5. Test methods for obtaining material inputs
for such models are discussed in section 6.2. Characteristic strength values must be established
from representative tests.
The characteristic strength values (
fc
, t
fc
, and P
fc
) may depend on the intended service
environment (e.g. temperature, humidity etc.). The representative mechanical tests to obtain
these values should include such effects.
From the representative mechanical tests, the sample mean x and standard deviation o should
be computed. The characteristic value
C
x is be obtained by subtracting k
m
times the sample
standard deviation from the sample mean. The factor k
m
depends on the number of samples
tested and would be 4.9 for five samples reducing to 3.0 for 15 samples. Values for this factor
for other numbers of test samples can be found in reference [1] (Sec. 4, B 400).
eq. 4-1
o
m C
k x x =
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 20
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The partial material factor
m
can now be specified if:
o Separate partial safety factors take account of load and model uncertainties.
o All environmental degradation effects are included in the characteristic material
properties by representative material characterisation tests.
o A robust manufacturing procedure is established that provides a small and stable
variability in the strength of the as-produced joints.
o All simplifying assumptions are conservative.
If all these conditions are met and the standard deviation in the measured strength of the bonded
joints is less than 10% of the mean value, a safety factor of
m
= 1.62 is recommended. If the
strength variability is greater, appropriate values for this factor can be found in reference [1]
(Sec. 8 and App. E).
4.3 Creep rupture due to sustained static loading
This section provides a simple failure criterion for creep rupture due to sustained static loading.
As in the previous section, several variants of the failure criterion may be considered. For
simplicity, only the variant expressed in terms of stress will be considered. The characteristic
time to failure t is expressed by the stress rupture curve:
eq. 4-2
log(t)
c
= log(t
0
) k log t x o
c
where t is the applied stress, o
c
is the sample standard deviation calculated from the creep test
results. This curve is established by linear regression to creep test data. The factor x is 4.0 for
15 tests and 3.0 for 50 tests. Values for this factor for other numbers of test samples can be
found in reference [1]. Characteristic strength values representative for the intended lifetime
may be calculated from the above equation by setting time t equal to the lifetime of the joint. A
safety factor of
m
= 1.62 is recommended used with this characteristic strength according to
reference [1]. This characteristic curve can be used to extrapolate to failure times 100 times as
long as the longest test duration.
4.4 Fatigue fracture due to cyclic loading
This section provides a simple failure criterion for fatigue fracture due to cyclic loading. As in
the previous section, several variants of the failure criterion may be considered. For simplicity,
only the variant expressed in terms of stress will be considered. The characteristic number of
cycles to failure N is expressed by the fatigue curve:
eq. 4-3
log(N)
c
= log(N
0
) k log t x o
c
where t is the applied stress, o
c
is the sample standard deviation calculated from the fatigue test
results. This curve is established by linear regression to fatigue test data. The factor x is defined
as for creep rupture above.
This characteristic curve can be used to extrapolate to cycle numbers 100 times larger than that
of the tests. A safety factor on number of cycles of 50 is recommended used for fatigue life
assessment [1].
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 21
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
4.5 Application of limit states to bonded joints with rigid and flexible
adhesives
In principle, all critical failure mechanisms shall be considered, and hence all the three limit
states above needs to be used. However, in some cases discussed below, simplifications may be
appropriate.
A bonded joint with a rigid adhesive that primarily carries load by shear stresses in the adhesive
bondline should be designed such that an appreciable portion of the joint remains elastic when
loaded up till failure. This is achieved if the overlap length is sufficiently long. For such joints,
any creep deformation that occurs in the highly loaded region of the adhesive (towards the ends
of the overlaps) will be resisted by the region of the adhesive bondline that still remains elastic.
This would limit creep deformations and hence prevent creep rupture in the joint. It should be
noted that this does not apply in general, but only for well designed joints. In such cases, the
creep criterion may not need to be considered. Furthermore, bonded joints tend to show a
fatigue SN curve with an appreciably smaller slope (i.e. larger k-factor in the limit state
equation) than for metals. Hence, fatigue is less of a concern for such joints than for welded
structures. This may justify not considering the fatigue limit state. In such cases, only the
extreme load limit state remains to be considered. This reflects current practice for such joints.
It should be noted that the safety factor of 1.62 indicated above for creep and extreme loads is
considerably smaller than those normally applied. The following issues explain this difference:
o The low safety factor accounts for uncertainty in the resistance only and assumes other
sources of uncertainty (model uncertainty, uncertainty in loads etc.) to be accounted for
by separate partial factors.
o The characteristic strength is defined in such a way that it accounts for much of the
uncertainty in the resistance.
o It is assumed that long term degradation is included in the characteristic strength value.
If there is uncertainty in how well the characteristic values from the tests represent the
long term properties in service, an additional safety margin needs to be included.
For bonded joints with typical elastic (flexible) adhesives, the creep behaviour is such that, if
designed according to the creep rupture limit state, the sustained stresses in the bonded joint will
be only a very small fraction of the ultimate stress. Hence, both the extreme load and the fatigue
limit state may never become governing. In such cases, only the creep rupture limit state needs
to be considered. However, for joints subjected to variable loads with a mean that is very small
compared to the extreme values, one may need to consider also the other limit states. Some
authors recommend safety factors of 20 or more to be used for flexible adhesives. This contrasts
with the factor of 1.62 recommended above. The explanation for this is that the high safety
factor is intended to be applied to a characteristic static strength and accounts for the difference
between the static strength and the creep strength, whereas the low safety factor is intended used
with a characteristic creep strength value.
4.6 Knock-down factors
A knock-down or reduction factor is a factor (typically less than unity) by which the strength is
multiplied to take account of one or more specific effects that are known (or expected) to reduce
the strength but are not taken into account by other means. Examples are factors applied to the
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 22
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
buckling strength of structural components to account for the effects of geometrical
imperfections (out-of-straightness, eccentricities of loading, etc.) possibly combined with
material yielding. Here the knock-down factor would be applied to the elastic critical load or
stress that has been calculated assuming a geometrically perfect, elastic structure. Knock-down
factors, which are typically evaluated from coupon tests, may also be used to take account of
degradation of properties due to, e.g., temperature, humidity, cyclic loads, long-term static loads
and ageing.
4.7 References
[1] DNV Offshore Standard DNV-OS-C501, COMPOSITE COMPONENTS, January 2003
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 23
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
5 DESIGN AND ANALYSIS OF BONDED JOINTS
5.1 Introduction to design of bonded joints
5.1.1 General aspects
The design process for adhesive joints has to include many aspects from the manufacturing of
joints through lifetime of the joints to disassembling of the structure. The objective of this
chapter is to provide a guideline for designing adhesive joints in the shipbuilding industry,
starting from collecting information for the design process, using analytical approaches and finite
element analysis for the design. The main focus is set on obtaining procedures that are
practicable, pragmatic and safe from the point of view of the shipbuilding industry. Thus, in
addition to being conservative the methods should be easy-to-use. On the other hand academic
completeness is not the most important factor. The different approaches are discussed with their
limitations due to mathematical assumptions or the grade of implementation in standard
commercial finite element codes. Most of the proposed design methods have been thoroughly
verified through comparison with experiments performed during the BONDSHIP project.
As will be clear from the present chapter we will not recommend the same design procedure for
all kinds of adhesive joints. For certain simple joints of low utilisation of loads, methods based
on nominal stresses and strains may be applied. Early analytical solutions, like the approach of
Volkersen or Goland and Reissner, may be used for some further aspects: Here, stress
concentrations at the ends of the joints are taken into account, and approximations on the peel
stresses may be obtained. These methods are suitable to get a first estimate of the overlap-length
of adhesively bonded joints. A more recent and complicated analytical solution method, derived
by Bigwood and Crocombe, has also been thoroughly investigated during the project work. The
relatively involved analytical expressions arising from this method have been implemented in an
Excel worksheet. The method works well for certain application cases. However, procedures
based on linear and non-linear finite element calculations might be the most suitable methods for
more general and complex joint design problems. Thus, it turns out that the term easy-to-use-
design-rule is not necessarily identical to simple analytical approach. In many cases linear or
non-linear finite element analysis may be the most efficient and suitable way to approach a
design of a bonded joint in a structure.
Adhesive bonding is especially suited for joining long or large areas and for the transfer of shear
or compressive loads for limited times (only flexible adhesives) and for shock loading. Unlike
other joining techniques, for example riveting and bolting, bonding techniques enable relatively
uniform stress distribution, and avoid the introduction of pre-stresses in the substrates.
Depending on the type of adhesive used, one or more of the additional advantages listed below
may be obtained
- Good tolerance compensation and gap-filling.
- Joining of dissimilar materials, considerable freedom of design.
- Good insulation against sound and vibration.
- Uncoupling of buckling and vibration modes.
- Compensation of differences in thermal expansion or stiffness
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 24
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The successful design of bonded joints requires profound knowledge of the behaviour of
adhesives under loading and selection of suitable materials and surface preparation. The key to
success lies in a dialogue between the product designer, the adhesive manufacturer and the
production department so that the best way to configure assemblies and joints for bonding can be
chosen.
5.1.2 Examples on joint configurations
As pointed out in the previous subsection bonded joints are to be designed to transfer shear or
compressive loads. Excessive peel stresses and other kinds of tensile loading in adhesives should
be avoided. The figures on the following pages show good and bad examples of bonded joint
configurations. The figures have been reproduced with permission from reference [1]. Figure 5-1
illustrates a number of possible designs for joining of sheet materials in tension. The range spans
from extremely simple joints to rather complicated geometries. Some quantitative measures on
the load carrying capacity and the manufacturing cost of the joints are also included. Figure 5-2
and Figure 5-3 offer a qualitative assessment of several T-joints and corner-joints, respectively.
Finally, in Figure 5-4 and Figure 5-5 examples of tubular joints and stiffener attachments are
shown.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 25
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-1: Joining of sheet materials in tension; the higher the load rating, the stronger
the joint (from [1])
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 26
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-2: Possible T-joints and efficiency of the joint concerning the specified loads.
(from [1])
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 27
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-3: Possible corner joints and efficiency of the joint concerning the specified load
(from [1])
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 28
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-4: Tubular joints for axial and torsional loading (from [1])
Figure 5-5: Stiffener attachment and load spreading (from [1])
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 29
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
5.1.3 General rules for methodical and geometrical design of adhesive joints
In the following some general aspects (from [2]) according to the design of adhesive joints are
given. The design of an adhesive joint should be clear, simple and safe.
Clear means that all loading conditions and the load flux through the joint, which will appear
during lifetime, should be known and well described. Furthermore, manufacturing, use,
maintenance, repair and disassembly should be possible.
Simple means, that the number of components which are used in the joint are as low as
possible. The relationships between components and mechanical quantities, for instance
stress and strain should be obtainable with low effort. If analytical solutions are applied the
geometry of the joint must be defined according to the assumptions of the mathematical
model. Possible symmetries for temperature distributions and mechanical loads during
manufacturing and lifetime should be used. Assembling, maintenance, repair and
disassembling should be as simple as possible.
Safe means that the design of the joint should satisfy the requirements of a safe-life, fail-safe
or redundant design philosophy.
Several principles about load flux, structural strength, introduction of loading into the joint, and
deformations should be applied: A good design avoids sudden changes of the load flux through
the joint which appear at sharp corners or sharp changes of diameters of structural components
(Figure 5-6). The local structural strength of the adherends should be of the same order of
magnitude. This can be achieved by careful selection of materials, local thicknesses and the
shape of the adherends. The level of utilisation of local structural strength should be equal
throughout the structure (the joint) and should not change during the lifetime. The introduction
of loading into the joint should avoid any bending moment.
Deformation of the adherends should correspond to the deformation of the adhesive in the joint
(Figure 5-7). This reduces stress concentrations at the overlap ends. With additional features and
symmetrical loading conditions, detrimental effects such as moments, secondary loads and peel
loads can be avoided or reduced (Figure 5-8 and Figure 5-9).
Figure 5-6: Load flux and principle of similar structural strength
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 30
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-7: Principle of corresponding deformation
Figure 5-8: Symmetric loading reduces secondary load effects
Figure 5-9: Additional design features to avoid or reduce peel-stress (from [1])
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 31
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The adhesive joint should be designed such that it becomes self-reinforcing, self-equalising and
self-protecting:
- Self-reinforcing stands for smart joint design. Harmful tensile and peel loads are converted
into pressure and shear loads. Figure 5-10 shows an example where this design principle has
been applied.
- Self-protecting: This refers to the case were a joint is overloaded. A self-protecting design
does not fail but introduces a new load transfer mechanism e.g. some kind of stopper (bolt,
built-in grove or ridge) that limits excessive shear deformation but is not loaded in normal
operation.
- Self-equalising: see Figure 5-7 as an example of a self-equalising design. The idea is to
balance the load transfer in the joint.
The design of the joint should also yield a stable solution, which is not fulfilled by the design
shown in Figure 5-11.
Self-reinforcing: Self-damaging:
Figure 5-10: An example for a self-reinforcing adhesive joint
The basic principles mentioned above are general design principles for structures. These
principles have to be applied also to the design of adhesive joints. As a consequence, in many
cases where adhesive joints shall be used, the surrounding structure has to be designed for
adhesive bonding. Sometimes this may require significant changes to structural geometries that
were originally designed for, e.g., welding.
Adhesives are polymeric materials with a strength that is one ore two orders of magnitude lower
than the strength of metallic adherends. The strength of such metallic, structural joints can only
be utilised if the size of the bonding area is sufficiently large. In most cases this can only be
reached, if the joint is loaded in shear. According to the general design rules and the mechanical
strength of adhesives any load should be avoided (or reduced as much as possible) which yields
peel.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 32
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-11: Unstable design of an adhesive joint
5.1.4 Main types of adhesive materials
A variety of adhesives are available for the structural joints. The first choice has to be made
between rigid and flexible adhesives according to the joint requirements. Not only mechanical
properties of adhesive materials, such as stiffness and strength, but also the production aspects
such as opening time, gap filling possibility and curing time have to be taken into account in the
selection of adhesives. See chapter 3 (pp. 9) for more information about the material selection
process.
5.1.4.1 Flexible adhesives
The stiffness and strength of flexible adhesives are relatively low compared with rigid adhesives.
On the other hand, the allowable strain of the flexible adhesive is high, while the bond thickness
must be relatively large. More precisely, flexible adhesives may typically be characterised by
- (e.g.) Polyurethane based
- Humidity curing, single component adhesives used to install windows, panels etc.,
- Good sealing properties
- Low strength: t = 1.55 MPa (soft joint)
- Typical adhesive thickness: 315 mm
- Temperature range: -40C+70 (+90) C (for temperature > 70C check with
adhesive manufacture)
- Important: When one of the elements is transparent (glass, colourless GRP etc.) the surfaces
to be bonded with PU adhesives must be protected from sunlight. Any contact with silicone
is strictly forbidden.
Typical advantages of flexible adhesives are as follows:
- Enable uniform stress distribution in the joints
- Good tolerance to bondline thickness variation and gap-filling
- Good resistance against impact loading
- Suitable for joining of dissimilar materials
- Excellent compensation for mismatch in thermal expansion coefficient
- Freedom of design.
Typical applications of flexible adhesive joints are:
- Windows
- Superstructure modules
o Front module
o Side wall module
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 33
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
o Roof module
- Casing, fairing structures.
5.1.4.2 Rigid adhesives
Rigid adhesives provide much higher stiffness, and strength values compared with flexible
adhesives. The failure strain is considerably smaller, and the tolerance capacity on the bondline
thickness is very limited.
Characteristic properties of rigid adhesives are as follows:
- Epoxy-Resin basis (EP)
- One or two component systems for thin adhesive layer bonding of (two different) materials.
- Careful application required
- High strength: t > 15 MPa
- Typical adhesive thickness: 0.22 mm
- Temperature range: -50C+100 C
Furthermore, rigid adhesive joints typically offer features like:
- Good load carrying capacity
- Close surface tolerance, limited gap filling property
- Pretension due to thermal elongation
Typical applications for rigid adhesives are:
- Sandwich cladding layer bonded to frame structure
- Inserts in Sandwich
- Local Attachments
- Structures with relatively small tolerance requirements.
Due to the different nature of flexible and rigid adhesives certain elements in the design
procedure for adhesively bonded joints depend on the adhesive type used. Nevertheless, the main
design steps are the same. The general part of the design method is presented in section 5.3,
while the sections 5.4 and 5.5 cover the particular elements connected to flexible and rigid
adhesive joints, respectively.
5.1.5 Example of joint analysis using the method by Bigwood and Crocombe
The analysis is a key issue when designing adhesively bonded joints. It is of great importance to
gain insight into the behaviour of the joints. The present subsection offers an example on a
general analysis of the stress distribution within the adhesive layer of single-lap joints, as
illustrated at the top of Figure 5-6. The analysis is based on the spreadsheet solution of Bigwood
and Crocombe, which has been implemented and investigated during the BONDSHIP project,
see the section 11.2.4. The use of the spreadsheet will be shown in the context of discussing the
influence of various parameters on the adhesive stress distributions for single-lap joints. The
main behaviour of both rigid and flexible adhesive joints will be investigated. The parameters
that are used as input to the calculations might be classified into four main groups, see Figure
5-12, which shows the input window of the spreadsheet, as follows:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 34
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Material properties; Youngs modulus of the adhesive and adherends, the shear modulus of
the adhesive and Poissons ratio of the adherends and adhesive (the latter is usually
calculated from Youngs modulus and the shear modulus of the adhesive).
The type of 2D analysis performed; either a plane strain or a plane stress analysis (put either
a 1 or a 0). Usually, structural joints are analysed using the plane strain assumption.
Geometrical parameters of the joint such as the overlap length and the thickness of the
adhesive layer and the adherends.
The elements of reduction of the loading conditions (shear forces, axial (tensile or
compressive) forces and bending moments).
Typical material parameters for a rigid, two components epoxy adhesive and a flexible
polyurethane adhesive are provided by Table 5-1. The applied axial load of the adherends is
chosen to be 50 N/mm, and the overlap length is assumed to be 10 mm. This means that the
nominal adhesive shear stress is 5 MPa. With a sample width of 25 mm the applied force
becomes F= 1250 N. The thickness of the adherends is assumed to be 1 mm, while the thickness
of the adhesive layer is 0.3 mm. Assuming, for simplicity, the bending moment factor to be
given by k = 1, this yields a bending moment per unit width M/b = (F/b) * 0.5 (t
adherends
+ t
adhesive
)
= 1250/25* 0.5 * (1 + 0.3) = 32.5 N. The adherends are made of steel with material properties
E= 215 GPa and v = 0.33.
Table 5-1: Typical mechanical constants for an epoxy adhesive and a rubber like, flexible
adhesive
Adhesive Youngs modulus
E [MPa]
Shear Modulus
G [MPa]
Poissons ratio
Epoxy 1800 667 0.35
Flexible polyurethane
adhesive
2 0.68 0.47
Figure 5-12: Input window of the Microsoft Excel spreadsheet
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 35
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Firstly, the single-lap joint with the rigid epoxy adhesive will be discussed. Figure 5-13 shows
the distributions of the shear stresses and the peel stresses for two different values of the bending
moment factor (for a definition of bending moment factor see eq. 11-14 on page 186); k = 1 and
k = 0. The case k = 0 can never been reached for real joints, but it is included here to illustrate
the importance of the bending moment. For k = 0 no peel stress occurs. Thus, the maximum peel
stress for k = 1, which is observed at the ends of the overlap, is due to the applied bending
moment. Also the distribution of the shear stress is influenced by the bending moment. The mean
shear stress, which corresponds to the nominal shear stress, is 5 MPa. For this joint with k = 1 it
is seen that the maximum peel stress is 2.5 times larger than the nominal shear stress, while the
maximum shear stress is less than twice the mean value. Therefore, the maximum values of all
the stress components should be considered for the design of rigid adhesive joints.
Figure 5-13: Shear and peel stresses for a rigid adhesive single-lap joint with k=0 and k=1
Using eq. 11-16 for the case considered here, we obtain k = 0.80, which changes the applied
moment to M/b = 26 N. For a similar joint with adherends made of aluminium (instead of steel)
with E= 73 GPa the bending moment factor and the moment per unit width become k = 0.70 and
M/b = 22.8 N, respectively. Figure 5-14 shows the stress distribution within the adhesive layer
for the single lap joints of aluminium and steel with k values calculated from eq. 11-16.
Comparing the results of the steel joint for k = 1 (Figure 5-13) with the corresponding
distribution for k = 0.80 it is seen that the maximum peel stress is reduced from about 12.5 MPa
to 10 MPa. However, the consequence of reducing the adherend modulus from that of steel to
aluminium is more significant: Now, the maximum peel stress increases from 10 MPa for steel to
15 MPa for aluminium, using k values calculated from eq. 11-16.
Changing to the flexible adhesive the peel stress approaches zero, while the shear stress
distribution is almost uniform and corresponds to the nominal shear stress (Figure 5-15). The
ratio between the modulus of a metallic adherend and the modulus of a typical rigid adhesive is
73000 / 1800 ~ 40 for aluminium and 215000 / 1800 ~ 120 for steel. With the flexible adhesives
introduced herein this ratio becomes approximately three orders of magnitude larger. It is seen
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 36
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
that the stress peaks at the ends of the overlap decrease with increasing ratio of the modulus of
the adherend and the modulus of the adhesive. According to Figure 5-15 the differences in the
stress distributions between the steel and aluminium joints are negligible for flexible adhesive,
while there are considerable differences when using rigid adhesives. For these reasons the design
of adhesive joints with flexible and rigid adhesives has to be treated differently. While the
nominal stress approach often yields a good approximation for the stress distribution within
flexible adhesives, the stress distribution is usually highly non-uniform when rigid adhesives are
applied.
Figure 5-14: Stress distributions for steel and aluminium with a rigid adhesive
Figure 5-15: Stress distributions for steel and aluminium with a flexible adhesive
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 37
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
5.2 Joint requirements
The method presented in chapter 2 to specify the requirements for bonded joints also provides
the designer with essential information such as environmental conditions and the type of material
used. The material screening process discussed in chapter 3 identifies suitable combinations of
adherends and adhesives as well as preparation/coating of surfaces. In order to get realistic
values of loads on the joints, global structural analyses may need to be performed in addition.
5.3 General design procedure for adhesive joints
Although the nature, the load carrying capacity and the application areas of flexible adhesive
joints might differ considerably from those of rigid adhesive joints, the main elements in the
design procedure for both types of joints are similar. One typically starts with determining a
basic configuration of the joint (pre-design). Then, it is time to perform analyses of stresses and
strains in the joint when subjected to the design loads specified. Finally, the stresses and/or
strains are inserted into the failure criteria applied for the joint, and the locations of (possible)
critical points are then detected. The procedure is illustrated graphically in Figure 5-16.
Figure 5-16 General design procedure for adhesive joints
A first configuration of the joint might imitate some previous experience of joint design or at
least be based on the drawings provided within the sub-section 5.1. More advices on the pre-
design (phase 1) of adhesive joints, including rough estimates on the overlap length, the load
capacity and/or the adhesive thickness are provided by the sections 5.4.1 and 5.5.1, for flexible
and rigid adhesive joints, respectively. In any case, the geometrical design of the assembly
should always favour shear and compressive loads in the adhesive. On the opposite, peel loads
and cleavage loads are the greatest enemy of the designers of bonded joints.
The stress/strain repartition (phase 2) might be evaluated quantitatively from the result of a
calculation, for example by easy-to-use analytical methods or numerical finite element analyses.
Evaluation of the
stress/strain repartition in
this assembly
Phase 2
Application of failure
criteria
Phase 3
Localisation of the points,
where problems might
appear (ex: stress
singularities)
Phase 4
Configuration of the
bonded assembly
Phase 1
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 38
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
This topic is covered by the sub-sections 5.4.2 (flexible adhesives) and 5.5.2 (rigid adhesives). A
thorough presentation of analytical and numerical calculation methods is provided by the
appendices 11.2 and 11.3.
Then, the failure criteria introduced for the joint of consideration are applied (phase 3). The
failure criteria may be based on a single stress or strain component, or a criterion using a stress
measure accounting for several components may be applied. In all cases, the stresses and strains
to be compared with the characteristic values of the measured parameters in the failure criteria
are obtained from the analysis performed in phase 2. Important aspects related to failure criteria
are discussed in sections 5.4.3 and 5.5.3 for joints using flexible and rigid adhesives,
respectively. Additionally, a general discussion of failure criteria and characteristic and
maximum (predicted) values to be applied in the failure criteria is given in sections 4.2 - 4.5.
Finally, from the application of failure criteria, critical regions (or zones in which failure is
predicted) might be detected (phase 4). If such points are observed, the design of the joint must
be modified (return to phase 1). Thus, by a trial and error process, the designer progressively
modifies the design in order to obtain a convenient and safe joint.
5.4 Design of flexible adhesive joints
5.4.1 Pre-design - possible joint configurations and overlap length (phase 1)
When a new adhesively bonded joint is to be designed, the first thing to do is to establish a rough
pre-design of the assembly. This pre-design should be based on general considerations and
drawings of suitable joints (see section 5.1), as well as knowledge gained from previous
experience with similar assemblies and loading conditions. In any case, the geometrical design of
the joint should always favour shear and compressive loads in the adhesive, while excessive peel
and cleavage loads should be avoided.
In the context of adhesive joints the surfaces to be bonded should be parallel. Joint dimensions
should be designed so that the solvents can evade through the joint. For that reason both ends
should be exposed, and adhesive bead depths should be within the range 3mm t 20mm. In
order to avoid peeling action under tension or bending, mechanical securing by use of a rim or
edge profiles is inevitable.
Flexible adhesive joints usually do not show considerable stress peaks at the overlap ends.
Therefore, nominal quantities, which are introduced in the appendix 11.2.2, can be used for a
first design approach of such joints. Since flexible joints are often exposed to relatively large
deformations, strains should be considered, in addition to stresses, during the design process.
In many cases flexible adhesives are used to compensate different thermal expansions of the
adherends to be bonded. The difference in thermal expansion (in the width direction) of the
adherends may be expressed by
eq. 5-1
T b b A o A = A ,
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 39
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Where b is the original width or length of the joint, respectively, Ao is the difference in the
coefficients of thermal expansion of the adherends and AT is the difference in temperature
between the curing temperature and the temperature of interest. The maximum thermal shear
strain observed in the adhesive, which is on the order of Ab/d, where d is the thickness of the
adhesive layer, must not exceed the allowable strain. This puts a lower limit on the adhesive
thickness.
5.4.2 Evaluation of the stresses and strains in the assembly (phase 2)
During the work in phase 1 the preliminary design of the joint is specified. This includes the
selection of the joint materials as well as the determination of the geometrical parameters. In
addition, the design loads or the maximum loads applied to the joint have been defined (see for
example [3]). It is then time to perform an analysis of the load-carrying capacity of the joint. A
crucial question is: Is the joint able to carry the design load specified? In order answer this
question, the assembly must be analysed with respect to critical stress and strain components.
These results will in turn be applied in the failure criteria introduced for the joints (see the sub-
section 5.4.3).
Flexible adhesive lap-shear joints may be applied in a large number of contexts. One of the most
important application areas for this fastening technique is the transport area, where it is widely
used in the assembly of road and rail vehicles. For example, the modern bus industry now
depends on flexible adhesives for a wide range of fastening applications including roof
assemblies, window glass, side walls, front and rear ends, floor pans and countless smaller
assemblies. In some applications flexible adhesive joints might be subjected to relatively high
load levels, while the external loads are small in other cases. Then the load-carrying capacity is
of minor concern, which means that there is no need to apply complicated analysis methods that
offer accurate resolution of the stress and strain distributions. On the contrary, estimates based
on nominal stresses and strains might be sufficient. For joints exposed to high level of loads,
however, accurate resolution of stresses and strains is of great importance. Thus, the choice of a
proper analysis method depends strongly on the joint type of consideration. In order to being
able to propose suitable methods for a wide variety of applications, four main joint categories are
introduced in Table 5-1. Along with the definition of the categories, the table contains
suggestions on analysis method for each category. Remark that most of the proposed analysis
methods are described in the appendices 11.2 and 11.3.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 40
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 5-1: Recommended analysis methods for flexible adhesive joints
Joint geometry
Utilisation of load
Simple Complex
Low
Analytical, nominal stresses
and strains (constant
stress/strain distribution)
(Category 1)
Numerical, linear FEA
(Category 2)
High
Numerical, materially and
geometrically non-linear FEA
(Category 3)
Numerical, materially and
geometrically non-linear FEA
(Category 4)
In order to be able to perform the stress/strain analysis of the assembly, the following parameters
must as a minimum be specified:
- Material parameters:
Young's modulus E
1
, E
2
(or E if E
1
= E
2
) of the adherends.
The shear modulus G of the adhesive as well as the Youngs modulus E
a
or the Poissons
ratio v.
- Geometrical parameter:
The thickness s
1
, s
2
(or s if s
1
= s
2
) of the adherends
The thickness d of the adhesive
The overlap length l and the width b of the joint
The parameters listed above are sufficient for the analytical solution method using nominal
stresses and strains. However, for linear finite element calculations (using solid elements)
additional parameters are required; i.e. the shear modulus or the Poisson ratio of the adherends.
Moreover, when using materially non-linear analysis methods, for example a hyper-elastic model
for the flexible adhesive, material parameters describing this kind of behaviour are also needed
(see e.g. section 11.3.3).
Let us conclude this sub-section by including a brief discussion on the content of Table 5-1. For
joints of low utilisation of load, the load-carrying capacity is of no major concern. If, in addition,
the geometry of the joint is simple (category 1) analyses based on nominal stresses and strains
are sufficient. However, in order to obtain conservative results, one should be aware of the
importance of choosing a reasonable effective stiffness for the adhesive layer, see appendix
11.2.2. If the joint geometry is complex (category 2), it is recommended to apply linear finite
element calculations to account for possible effects due to the non-trivial geometry.
When turning to flexible adhesive joints subjected to a relatively high level of applied loads, the
situation changes significantly. Now, the load-carrying capacity of the joint is of great interest.
Therefore it is important to select analysis methods that are capable of predicting accurate stress
and strain distributions. The behaviour of flexible adhesives is well described by non-linear,
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 41
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
hyperelastic material models. Moreover, the deformations and strains might be large. Thus, finite
element analyses, taking both material and geometrical non-linearity into account (see the
appendix 11.3.3), should be applied regardless of the complexity of the joint geometry
(categories 3 and 4).
5.4.3 Application of failure criteria (phase 3)
In the context of flexible adhesive joints the strength of the adherends are typically much higher
than the adhesive strength. Therefore, there is usually no need to introducing failure criteria for
the adherends when using flexible adhesives. On the other hand, it is of great importance to
define applicable failure criteria for the adhesive. Flexible adhesive joints may often undergo
relatively large deformations. Therefore, failure criteria based on maximum strains should be
applied in addition to the criteria for the stress components. Typical criteria for shear loading,
which contain the maximum predicted shear stress and shear strain are introduced in section 4.2.
If some kind of tensile or peel loading is of concern, corresponding failure criteria for axial
stresses and/ or strains must be introduced.
The materially non-linear behaviour of flexible adhesives is usually well described by a
hyperelastic material model. Such models are typically introduced into finite element analyses
through one of the strain energy potential functions that are (often) implemented into the
computer code. Thus, a failure criterion based on the elastic strain energy density can also be
introduced.
5.4.3.1 Determination of strength limits and maximum stresses and strains
The failure criteria discussed in section 5.4.3 above rely on maximum allowable stress and strain
levels. Often, such values are provided by the manufacturer of the materials of consideration.
However, the strength of real joints does not solely depend on the strength limits of the adhesive
and adherends applied. The entire adhesive system, including adhesive primer, cleaner and
adhesive, has to be taken into account. In addition the adherends, including the material, the
surface preparation, the surface pre-treatment, painting etc., are as important as the adhesive
system. Thus, the strength of real joints may be quite different from indications based on pure
adhesive and adherend strength parameters.
Ideally, the design process and strength considerations of bonded joints should include extensive
experimental testing of full scale joints with the same materials, geometries, surface treatments
and environmental conditions as applied when producing the joint for the real application.
However, such a procedure is not always possible. The costs related to extensive testing of full
scale joints can soon become prohibitive.
During the BONDSHIP project a pragmatic approach has been developed. The approach uses
experimental data, gained from shipyard conditions in combination with theoretical analyses.
The experimental basis of the approach is a simple lap-shear or tensile test with adherends
(including the thickness), surface treatments and adhesive system used on the yard. The lap-shear
or tensile samples, although simple, should represent the conditions, including the main load
transfer mechanisms, for the application as close as possible. This has been done during the
coupon test programme [4]. For controlling the manufacturing procedure the samples to be used
in the determination of the stress and strain limits in the failure criteria should even be
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 42
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
manufactured on the yard by the same people who will do the adhesive bonding in the particular
application.
The process of determining the strength parameters to be used in the failure criteria (typically
characteristic values of the measured stresses and/or strains at fracture) is defined as follows:
1. The representative lap-shear samples are loaded until failure.
2. The lap-shear or tensile sample is analysed theoretically by, e.g., analytical or numerical
calculation procedures (depending on the joint category, see Table 5-1). This analysis is
carried out at the failure load obtained in the experiments. An example from [5] is shown
in Figure 5-17.
3. The characteristic strength values for stresses and strains to be used in the failure criteria
are based on the maximum (or nominal for joints in category 1) values of the
corresponding components obtained from the theoretical analysis of the test sample joint.
In the same example the shear stress (T
xy
) and the peel stress (S
y
) is plotted along a
straight line at a distance of d/10 from one adherend (Figure 5-18). The characteristic
strength value (used here as a failure criterion) is determined as shown in Figure 5-19
from the average shear stress (T
xy
) and peel stress (S
y
) measured along a line of the
length, d, equal to the thickness of the adhesive layer, at a distance of d/10 at the end of
the joint where the maximum stress appears.
As can be seen, the determination process is based on a combination of experimental and
theoretical analyses. Different kinds of theoretical approaches may be applied, cf. Table 5-1.
However, it should be remarked that the theoretical method adopted in the process of
determining the failure limits must be applied in the analysis of the real joint as well. This is due
to the fact that predictions of stresses and strains may depend considerably on the analysis
method used. Similarly, the results (in particular near singularities) obtained by finite element
analysis are often significantly influenced by the mesh size. Therefore, the same level of
discretization should be applied in the analyses of the test sample and the joint to be designed.
Figure 5-17: Linear-elastic numerical analysis for lap-shear test
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 43
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-18: Failure criteria from maximum shear and peel stress at the joints ends
Figure 5-19: Procedure to determine a failure criterion and accuracy of numerical results
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 44
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
5.4.4 Design procedures for flexible adhesive joints in ships
Based on the discussions in the sections 5.3 and 5.4.1-5.4.3, it is now time to summarize this
section by specifying a design method applicable for flexible adhesive joints in ships. The
method is defined as follows:
1. For each application case, a list of requirements should be worked out.
2. The pre-design of the joint should be carried out using the advices in section 5.4.1.
3. The shear strength of lap-shear samples and/or tensile strength of samples loaded in tension
(typically in the through-thickness direction) with adherends, adhesives, surfaces and surface
conditions after pre-treatment and at shipyard-conditions have to be determined with respect
to the particular application case. A characteristic strength (i.e. external load) value is
determined from these experiments.
4. Knock-down factors (see Table 5-3) for all possible impacts during lifetime have to be
determined experimentally. Adherends, adhesives, surfaces and manufacturing conditions for
the samples have to be chosen according to the requirements of the application case. For
example, if adhesive bonds shall be realised on painted surfaces all experiments, including
necessary fatigue tests etc., have to be done with these surfaces. Appropriate safety factors
must also be included in the design process. In addition to the measurements of strength, the
test sample should be loaded with constant strain. The maximum strain, for which no creep
fracture occurs, has to be found [6]. If creep and fatigue conditions appear simultaneously,
the lower of the knock-down factors should be used for a first approach. For qualification of
this particular application, experimental proof has to be accomplished with fatigue tests
where the mean load corresponds to the static load.
5. The strength limits for the failure criteria have to be developed according to the sub-section
5.4.3.1 with respect to the experimental results of the lap-shear or tensile samples described
in the item 3 above and the knock-down and safety factors from item 4.
6. The joint to be designed is now investigated by a suitable analysis method, see the sub-
section 5.4.2 (the analysis method applied in 5 should be used here), using the design load
and other requirements specified in 1 as well as the failure criteria obtained in 5.
7. The geometry of the joint, including, for example, the adhesive thickness, has to be changed
until the maximum stresses and strains are below the allowable values.
8. Component tests might be useful for any qualification procedures and verifications.
9. The adhesive joint should be designed to avoid crack initiation during the lifetime of the
joint.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 45
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
5.5 Design of rigid adhesive joints
5.5.1 Pre-design - possible joint configurations and overlap length (phase 1)
When a new adhesively bonded joint is to be designed, the first thing to do is to establish a rough
pre-design of the assembly. This pre-design should be based on general considerations and
drawings of suitable joints (see section 5.1), as well as knowledge gained from previous
experience with similar assemblies and loading conditions. In any case, the geometrical design of
the joint should always favour shear and compressive loads in the adhesive, while excessive peel
and cleavage loads should be avoided.
To dimension rigid adhesive joints the main issue is to keep the stresses below the characteristic
strength values. For rigid adhesive joints, the strength of the adhesive bond is not only
determined by the characteristics of the adhesive, but the adherends are of the same level of
importance. Besides the selection of an appropriate type of adhesive, the main design parameters
are the overlap length of the joint and the thickness of the adherends. Based on certain
simplifying assumptions (see section 11.2.3) Wiedemann has derived easy-to-use analytical
formulas for these parameters. The main conclusions of the analysis of Wiedemann are included
below, while a more detailed discussion of the theory is presented in the appendix 11.2.5.
Due to the specific stress distribution in the joint there is a maximum overlap length, above
which no further load capacity can be reached. The design value of the overlap length for a
single-lap joint can be estimated by
eq. 5-2
( ) o +

>
1
5 2
1 1
G
d s E
l ,
where o is the adherend stiffness ratio defined by
eq. 5-3
2 2
1 1
s E
s E

= o .
In eq. 5-2 the effects of peel stresses and the additional stresses due to the bending moment are
taken into account by introducing a safety factor of 2. The indices of the adherends have to be
selected in a way to ensure that os1. The following equation estimates a shear stress factor,
which relates the maximum shear stress to the mean shear stress as an easy-to-use formula at the
overlap length with the lowest stress levels:
eq. 5-4
o t
t
+
=
1
5
max
m
.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 46
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
In the previous equation t
max
is the maximum shear stress at the ends of the overlap length and
t
m
is the mean or nominal shear stress applied to the joint, respectively. The shear strength of the
adhesive, t
F
, has be larger than the maximum shear stress, t
max
:
max F
t t > or
max F
1 1
t t
s .
The best way to measure t
F
is to use two bonded cylinders and to perform a torsional shear test.
A conservative approach is to test the lap shear strength according to DIN EN 1465.
With the applied tensile stress, o
10
, at adherend 1 one obtains the following equation:
eq. 5-5
( )
1 max
10
5
1
s
l
F
F
F
F

+ = s o
t
o
t
o
t
o
,
where
eq. 5-6
o
F
s
1
= t
F
l .
Hence
eq. 5-7
( )
G s
d E
F

+
s
1
1
10
1
2
o
t o
is obtained as an easy-to-use rule. o
10
/t
F
is called the load capacity of the joint.
The tensile stress on either side of the bond has to be smaller than the ultimate strength of both
adherends. With respect to an easy to use design rule o
10
should be smaller than the yield
strength R
p0.2
of the adherends. The procedure of dimensioning a rigid adhesive joint is to
calculate the overlap length l and to ensure that the applied load o
10
is smaller than given by the
latest equation. As already mentioned, t
F
might be determined with a lap shear test. In order to
optimise the joint, the thickness of the adherends might have to be changed.
According to the basic equations of Volkersen and Goland and Reissner the stresses in the
adhesive layer decrease with increasing thickness of the adhesive layer. However, the
experimentally measured lap-shear strength decreases with increasing thickness of the adhesive
layer, which is apparently in contradiction to what should be expected from the analytical
equations. Therefore, if the thickness of the adhesive layer exceeds 0.5 mm, experimentally
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 47
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
based knock-down factors accounting for the fact that the load capacity of adhesively bonded
joints decreases with increasing thickness of the adhesive layer must be included.
5.5.2 Evaluation of the Stresses in the Assembly (phase 2)
During the work in phase 1 the preliminary design of the joint is specified. This includes the
selection of the joint materials as well as the determination of the geometrical parameters. In
addition, the design loads or the maximum loads applied to the joint have been determined (see
for example [3]). It is then time to perform an analysis of the load-carrying capacity of the joint.
A crucial question is: Is the joint able to carry the design load specified, or will it fail? In order
answer these questions, the assembly must be analysed with respect to, e.g., critical stress
components. These results will in turn be applied in the failure criteria introduced for the joints
(see the sub-section 5.5.3).
Adhesively bonded lap-shear joints may be applied in a large number of contexts. Such joints
may roughly be divided into non-structural and structural joints. In the former case, the applied
loads are small, and the load-carrying capacity is of minor concern. Then, there is no need to
apply complicated analysis methods that offer accurate resolution of the stress distribution. On
the contrary, rough estimates based on nominal stresses might be sufficient. For structural joints,
however, which might be subjected to excessive loads, accurate resolution of stresses is of great
importance. Thus, the choice of a proper analysis method depends strongly on the joint type of
consideration. In order to being able to propose suitable methods for a wide variety of
applications, six main joint categories are introduced in Table 5-2. Along with the definition of
the categories, the table contains suggestions on analysis method for each category. Remark that
most of the proposed analysis methods are described in the appendices 11.2 and 11.3.
Table 5-2: Recommended analysis methods for rigid adhesive joints
Joint geometry
Utilisation of load
Simple Complex
Low
Analytical, nominal
stresses/strains (constant
stress/strain distribution)
(Category 1)
Numerical, linear FEA
(Bigwood and Crocombe may
sometimes be applied)
(Category 2)
Medium
Analytical, linear methods
(e.g. Volkersen, Goland and
Reissner, Bigwood and
Crocombe)
(Category 3)
Numerical, linear FEA
(Bigwood and Crocombe may
sometimes be applied)
(Category 4)
High
Numerical, materially and
geometrically non-linear FEA
(Conservative methods based
on linear FEA may also be
applied)
(Category 5)
Numerical, materially and
geometrically non-linear FEA
(Conservative methods based
on linear FEA may also be
applied)
(Category 6)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 48
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
In order to being able to perform the stress/strain analysis of the assembly, the following
parameters must (as a minimum) be specified:
- Material parameters:
Young's modulus E
1
, E
2
(or E if E
1
= E
2
) of the adherends.
The shear modulus G of the adhesive.
- Geometrical parameter:
The thickness s
1
, s
2
(or s if s
1
= s
2
) of the adherends
The thickness d of the adhesive
The overlap length l
The parameters listed above are sufficient for most closed-form analytical solution methods.
However, for linear finite element calculations (using solid elements) additional parameters are
required; i.e. the shear modulus or the Poisson ratio of the adherends and Youngs modulus or
the Poisson ratio of the adhesive. Moreover, when using materially non-linear analysis methods
material parameters describing this kind of behaviour are also needed.
Let us conclude this sub-section by including a brief discussion on the content of Table 5-2. For
joints of low utilisation of load, the load-carrying capacity is of no major concern. If, in addition,
the geometry of the joint is simple (category 1) crude analyses based on nominal stresses (and
strains) are usually sufficient. However, if the joint geometry is complex (category 2), it is
recommended to apply linear finite element calculations to account for possible effects due to the
non-trivial geometry.
Turning now to rigid adhesive joints subjected to a somewhat higher level of applied loads, the
situation changes significantly. Although the joints are not supposed to be loaded up to their
upper limits, the presence of non-uniform stress distributions, which are well-known for such
joints, must be taken into account. Therefore, analysis procedures using nominal stresses are not
applicable. For standard assemblies (category 3) like the single-lap (or double-lap) joint the
analytical solution methods of, e.g., Volkersen, Goland and Reissner, and Bigwood and
Crocombe are examples on easy-to-use analysis methods of sufficient accuracy. The validity of
these methods is restricted to joints of simple geometries. Therefore, for complex joints
(category 4) linear finite element analyses should be applied. However, the analytical method of
Bigwood and Crocombe may be applied for certain non-standard joint geometries, see the
section 11.2.4.
Finally, rigid adhesive joints of high load utilisation will be discussed briefly. For such joints the
load-carrying capacity is of great interest. Therefore it is important to select analysis methods
that are capable of predicting accurate stress distributions. When the load level approaches the
strength limit of the joint, the adhesive layer is subjected to plastic yield. Thus, the maximum
adhesive stresses in the joint are considerably smaller than predicted by linear-elastic finite
element calculations. Therefore, analysis methods accounting for materially non-linear effects
should be applied. In addition, geometrically non-linear effects are observed for most joints of
practical interest when loaded extensively. Even the simple single-lap joint is significantly
affected by such effects for a wide range of applied loads. Therefore, highly loaded joints should
be analysed by numerical finite element procedures accounting for both geometrically and
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 49
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
materially non-linear effects, regardless of the complexity of the joint geometry (categories 5 and
6).
During the BONDSHIP project a pragmatic design approach based on linear finite element
analysis has been developed [5]. Through comprehensive experimental testing and comparison
with real application cases the method has proven to be conservative in most cases. In fact, the
strength was overestimated in only very few application cases, while the method offered
conservative strength predictions for more than 80% of the test cases. Design methods taking
non-linear effects into account offer higher accuracy in the resolution of stresses and strains
within joints than linear design methods. However, linear methods that have proven to yield
conservative results may be applied as a design tool for structural joints (categories 5 and 6).
5.5.3 Application of failure criteria (phase 3)
Typical failure criteria to be applied to adhesively bonded joints were introduced in section 4.2.
In the present section criteria relevant for bonded joints with rigid adhesives will be discussed.
Finally aspects related to the determination of the strength values in the criteria will be covered.
5.5.3.1 Failure criteria for the adherends
When considering failure criteria for the adherends of adhesive joints a distinction between
isotropic adherends (like aluminium or steel) and orthotropic materials (like composite materials)
has to be made due to the fact that these materials tend to fail differently. For isotropic
adherends, a good design criterion is to avoid yield. Thus a failure criterion (in the form
introduced in section 4.2) for the von Mises equivalent stress should be applied. Orthotropic
adherends, on the other hand, may typically fail due to delamination caused by excessive peel
stresses. A failure criterion based on peel stresses should therefore be introduced. It should be
emphasised that accurate modelling of the adherend peel stresses is a challenging task.
Moreover, it may be extremely difficult to obtain general and reliable estimates on the peel
strength. The peel strength might not be uniquely defined by the materials and the fibre volume
or weight fractions applied. Also the orientation of fibres and laminae, the curing time and
temperature and other aspects related to the environmental conditions in the production process
play a crucial role. Sometimes the peel strength is provided by the manufacturer of the
adherends. Otherwise, it should be determined by experimental testing in the design process.
5.5.3.2 Failure criteria for the adhesive layer
Rigid adhesives usually possess elastic-plastic behaviour. Nevertheless, some rigid adhesives
show a relatively brittle nature, which means that the maximum plastic strain is small. In these
cases the designer should avoid any yielding of the adhesive. Then a failure criterion based on
the yield stress may be applied. Such a criterion is usually expressed with respect to the von
Mises equivalent stress, which is implemented into most commercial finite element codes.
Generally, the von Mises stress depends on all stress components, but if the stress state in the
(critical parts of the) joint is dominated by the peel and shear stresses, denoted o
peel
and
t, respectively, the von Mises equivalent stress may be simplified to
2 2
_
3 ) ( t o o + =
peel Mises von
.
The criterion discussed above might be applied for rigid adhesives. However, it has been shown
that it does not offer an accurate prediction of the yield behaviour of typical polymeric adhesives
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 50
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
that are frequently applied in rigid adhesive joints. The von Mises approach only considers the
deviatoric stresses, while it is known that also the hydrostatic stresses play a crucial role on the
yield behaviour of polymeric adhesives. Thus, a more accurate failure criterion for polymeric
adhesives may be based on the equivalent stress introduced by Raghava et.al. (see [7] or [8]).
Unfortunately, the latter failure criterion is typically not available in finite element codes applied
in ship yards, and it is often very complicated to implement additional criteria into such
commercial packages.
As we have seen, most rigid adhesives behave in an elastic-plastic manner, allowing a
considerably amount of plastic yield before failure. In such cases, local yield in the adhesive
layer should be accepted. Then a failure criterion based on the maximum plastic shear strain
e
or
the total shear strain (
e
+
p
;
p
is the plastic shear strain at adhesive failure) may be introduced
for the lap-shear joint of consideration. When using such a failure criterion one needs reliable
estimates of
e
and
p
. In addition, the behaviour of the adhesive in the elastic and plastic states,
as well as in the transition zone, must be known. Such data may often be provided by the
adhesive manufacturer. Otherwise, the most important parameters might be estimated through
comparison of non-linear analyses and results from experimental tensile tests of simple lap-shear
joints, with a relatively short overlap region, using the adhesive of consideration.
A key issue in the BONDSHIP project is to develop and investigate easy-to-use procedures for
the design of adhesively bonded joints. The two yield criteria based on the von Mises and
Raghava equivalent stresses depend on all stress components. When applied in a finite element
context, this is of no concern (note: the criterion using the Raghava stress equivalent is usually
not implemented into commercial codes) since all stress components are automatically predicted.
However, the situation is quite different when using analytical analysis methods. In such cases,
only the distribution of the adhesive shear stress is calculated (and for some methods also the
peel stresses may be estimated). Then, the failure criteria defined with respect to equivalent
stress measures are not applicable. In the BONDSHIP project failure criteria based on a single
stress component have, therefore, been introduced. For example, failure criteria based on the
shear and peel stresses of the adhesive may be applied. These failure criteria are particularly
suitable when the stress state in the adhesive layer of the joint is clearly dominated by shear or
peel stresses, respectively. The strength values included in these criteria may be estimated by
experimental testing.
5.5.3.3 Determination of strength limits and maximum stresses
The failure criteria discussed in chapter 4 rely on local maximum stress and strain levels.
However, the strength of real joints does not solely depend on the strength limits of the pure
adhesive and adherends applied. The entire adhesive system, including adhesive primer, cleaner
and adhesive, has to be taken into account. In addition the adherends, including the material, the
surface preparation, the surface pre-treatment, painting etc., are as important as the adhesive
system. Thus, the strength of real joints may be quite different from indications based on pure
adhesive and adherend strength parameters.
Ideally, the design process and strength considerations of bonded joints should include extensive
experimental testing of full scale joints with the same materials, geometries, surface treatments
and environmental conditions as applied when producing the joint for the real application.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 51
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
However, such a procedure is not always possible. The costs related to extensive testing of full
scale joints can soon become prohibitive.
During the BONDSHIP project a pragmatic approach has been developed, which has been
applied to certain application cases and verified by component tests. The approach uses
experimental data, gained from shipyard conditions, and a local finite element analysis which
can be included into a global analysis. The experimental basis of the approach is a simple lap-
shear test with adherends (including the thickness), surface treatments and adhesive system used
at the yard. The lap-hear samples, although simple, should represent the conditions, including the
main load transfer mechanisms, for the application as close as possible. This has been done
during the coupon test programme [4]. For controlling the manufacturing procedure the samples
to be used in the determination of the stress and strain limits in the failure criteria should even be
manufactured on the yard by the same people who will do the adhesive bonding in the particular
application.
The process of determining the strength parameters (typically characteristic values of the
measured stress and/or strains at failure) to be used in the failure criteria is defined as follows:
1. The representative lap-shear samples are loaded until failure.
2. The lap-shear sample is analysed theoretically (by, e.g., analytical or numerical
calculation procedures) subjected to the failure load obtained in the experimental analysis
above.
3. The upper limits of stresses and/or strains to be used in the failure criteria are based on
the (possibly, slightly modified) maximum values of the corresponding components
obtained from the theoretical analysis.
As can be seen, the determination process is based on a combination of experimental and
theoretical analyses. Different kinds of theoretical approaches may be applied, cf. Table 5-2.
However, it should be remarked that the theoretical method adopted in the process of
determining the failure limits must be applied in the analysis of the real joint to be designed as
well. This is due to the fact that predictions of stresses and strains, in particular at critical
locations, depend strongly on the analysis method used.
Determination based on linear FEA the method of Brede
Let us now describe and discuss aspects of the determination method introduced by Brede [5].
As an easy-to-use approach this method is based on linear elastic finite element calculations.
That is, linear finite element analysis of the test sample is applied to predict the strength limits
for the failure criteria. However, as pointed out by Brede, when lap-shear joints are loaded up to
failure local adhesive yield, which serves to reduce stresses, will occur. Such effects are not
taken into account by the linear finite element analysis. Therefore, the predicted stress values
might be considerably larger than the values presented in real joints. Thus, the maximum failure
limits obtained by this method, are not generally applicable. On the contrary, these strength
limits should only be applied in conjunction with the same kind of linear analysis method.
It is also well-known that results predicted by finite element analyses depend on the underlying
mesh. Increasing the number of elements serves to increase the local stress values at critical or
singular points. To overcome this problem in a practical way, Brede does not apply the predicted
maximum stresses as failure limits. On the other hand, to reduce the influence of these artificial,
local, numerical effects, Brede adopts the mean shear and peel (as well as the von Mises and
principal) stresses at a line of length d (where d is the adhesive thickness), located d/10 from the
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 52
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
overlap end of maximum stresses, as strength values in the failure criteria. In addition, to
introduce conservatism into the design approach, the maximum stresses (rather than some kinds
of mean values) along the line described above are applied in the analysis of the component to be
designed.
Determination based on non-linear FEA
Now, a more accurate determination method based on fully non-linear finite element calculations
will be described and discussed briefly. While the main steps in the method follow the ideas
introduced by Brede, which are presented above, the crucial difference is connected to the
underlying finite element analysis, which is fully non-linear in the present method.
Firstly, the characteristic maximum load for the representative, but simple lap-shear joint is
determined experimentally from a number of (identical) specimens. Then, the test joint is
analysed numerically by the non-linear finite element method, when subjected to the
characteristic external load. The main material parameters describing the elastic as well as the
plastic behaviour for all the constituents in the joint should be known prior to the numerical
analysis. Such parameters are often provided by the manufacturers. However, sometimes the
plastic behaviour (of the adhesive) is not completely described. Nevertheless, by choosing a
relatively short overlap length of the specimens and using a proper instrumentation, reasonable
estimates on the plastic material parameters may be obtained by comparing and matching the
results from the experimental and numerical analyses. The characteristic strength values to be
applied in the failure criteria such as the plastic strain and the peel stresses at fracture are
obtained in the same manner. Due to the fact that the fully non-linear finite element method,
which represents the joint behaviour well, offers realistic values for stresses and strains, the
maximum (characteristic) values, rather than some kinds of average stresses and strains, are used
directly. However, although non-linear methods offer more accurate predictions of stresses and
strains than linear procedures, the results might be influenced by significant mesh effects, in
particular, near singular points. Therefore, in order to obtain the same level of accuracy in the
analyses of the simple test specimen and the real joint to be designed, the same level of
discretization, i.e. the number of elements through the adhesive thickness, should be applied in
both cases. Moreover, it should be mentioned that various sources of uncertainties are taken into
account through safety factors and knock-down factors as described in section 4.2.
5.5.4 Design procedures for rigid adhesive joints in ships
Based on the discussions in the sections 5.3 and 5.5.1-5.5.3, it is now time to summarize this
section by specifying two important design methods for structural, rigid adhesive joints in ships.
Firstly, the pragmatic method developed in the BONDSHIP project by Brede will be presented.
Thereafter, a design method based on non-linear finite element analysis will be shown.
5.5.4.1 Design method of Brede
The linearly based design method proposed and investigated by Brede is defined as follows:
1. For each application case, a list of requirements should be worked out.
2. The pre-design of the joint should be carried out using the advices in section 5.5.1.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 53
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3. The shear strength of lap-shear samples with adherends, adhesives, surfaces and surface
conditions after pre-treatment and at shipyard-conditions have to be determined with respect
to the particular application case. A characteristic strength (i.e. external load) value is
determined from these experiments.
4. Knock-down factors (see Table 5-3) for all possible impacts during lifetime have to be
determined experimentally. Adherends, adhesives, surfaces and manufacturing conditions for
the samples have to be chosen according to the requirements of the application case. For
example, if adhesive bonds shall be realised on painted surfaces all experiments, including
necessary fatigue tests etc., have to be done with these surfaces. Appropriate safety factors
must also be included in the design process.
5. The strength limits for the failure criteria have to be developed according to the subsection
5.5.3.3 with respect to the experimental results of the lap-shear samples described in the item
3 above and the knock-down and safety factors from item 4. The maximum limits to be
applied in the failure criteria are determined from lap-shear tests using a linear elastic finite
element analysis of the specimens subjected to the characteristic external load. For rigid
adhesives the average peel and shear stresses along a line of the length of the adhesive
thickness, d, at a distance d/10 from the adherend with the highest stress level should be
taken as maximum allowable local stresses.
6. The joint to be designed is now investigated by linear finite element analysis using the design
load and other requirements specified in 1 as well as the failure criteria obtained in 5. The
stress values to be compared with the allowable stresses in the failure criteria, are the
maximum values obtained at the line lying a distance d/10 from the adherend with the
highest stress level. In certain cases, where the stress values at the ends of the line described
above are affected by artificial, numerical effects, the values at the very end of the line (for
example the last two or three nodal points) should be disregarded.
7. The geometry of the joint, including, for example, the adhesive thickness, has to be changed
until the maximum stresses are below the allowable values.
8. Component tests might be useful for any qualification procedures and verifications.
9. The adhesive joint should be designed to avoid crack initiation during the lifetime of the
joint.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 54
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 5-3: Knock-down factors for designing adhesive joints
f
T
Temperature
f
Df
dynamic load, fatigue
f
Di
dynamic load, impact
f
M
humidity, moisture
f
W
Water
f
L
creep, static load
f
p_P3.1
paint, for example, shop-
primer P3.1
f
S
surface preparation and pre-
treatment
f
d
thickness of adhesive layer
f
defect
defects from manufacturing
etc.
5.5.4.2 Design method based on non-linear finite element analysis.
In the following a design method based on non-linear finite element analysis will be specified.
The main parts of the method are identical with the procedure presented in section 5.5.4.1.
However, crucial modifications are introduced in the items 5 and 6 below. The method is defined
as follows:
1. For each application case, a list of requirements should be worked out.
2. The pre-design of the joint should be carried out using the advices in section 5.5.1.
3. The shear strength of lap-shear samples with adherends, adhesives, surfaces and surface
conditions after pre-treatment and at shipyard-conditions have to be determined with
respect to the particular application case. A characteristic strength (i.e. external load)
value is determined from these experiments.
4. Knock-down factors (see Table 5-3) for all possible impacts during lifetime have to be
determined experimentally. Adherends, adhesives, surfaces and manufacturing conditions
for the samples have to be chosen according to the requirements of the application case.
For example, if adhesive bonds shall be realised on painted surfaces all experiments,
including necessary fatigue tests etc., have to be done with these surfaces. Appropriate
safety factors must also be included in the design process.
5. The strength limits for the failure criteria have to be developed according to the
subsection 5.5.3.3 with respect to the experimental results of the lap-shear samples
described in the item 3 above and the knock-down and safety factors from item 4. The
maximum allowable values to be applied in the failure criteria are determined from lap-
shear tests using a fully non-linear finite element analysis of the specimens subjected to
the characteristic external load. For rigid adhesives the maximum values of, e.g., the peel
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 55
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
stress and plastic shear strain should be applied as the strength limits in the failure
criteria.
6. The joint to be designed is now investigated by fully non-linear finite element analysis
using the design load and other requirements specified in 1 as well as the failure criteria
obtained in 5. The stress and strain values to be compared with the allowable values in
the failure criteria are the maximum stress/strain values obtained in the joint.
7. The geometry of the joint, including, for example, the adhesive thickness, has to be
changed until the maximum stresses are below the allowable values.
8. Component tests might be useful for any qualification procedures and verifications.
9. The adhesive joint should be designed to avoid crack initiation during the lifetime of the
joint.
5.6 Application example: Optimisation of attachments below decks
This section was taken from reference [5]. In this section the entire design process for a joint will
be presented for the example of an attachment below deck. From the list of requirements the
joint has to carry a static load of 1000 N. However, to account for possible accelerations of the
ship, this is increased to 1500 N. The material of both adherends is shipbuilding steel, grade A
(St34k, 1.0441). The geometry of one adherend is determined by the profiles which are used
below deck and cannot be changed (Figure 5-20, left adherend defined as adherend 2). Free
design parameters are the overlap length, the width and the thickness of the adherend on the right
hand side of Figure 5-20 (defined as adherend 1). Araldite 420 is selected as adhesive system.
Figure 5-20: Attachments: Identification of hot spots
The surface of both adherends has to be shot blasted with corundum. The failure criterion has to
be determined from the results shown in row 4 of Table 5-4 and Table 5-5. The strength of lap-
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 56
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
shear joints for these surfaces is 28.6 0.38 MPa for three samples yielding a characteristic
value according to o o = o k
m C
(o
c
= characteristic value; k
m
is a constant depending on the
number of samples and the level of confidence which is multiplied with the standard deviation
estimated from the measurement) with k
m
= 7.66 (DIN 55303, part 2, Table 3) of 25.7 MPa.
Table 5-4 and Table 5-5 contain failure criteria which were determined in the same combined
experimental and numerical way for various surfaces and two adhesive systems (Plexus MA 550
and Araldite 420). The procedure has the advantage to be able to include any surface treatment
or surface coating which is possible at the yard with relatively low experimental and numerical
effort. It should be mentioned that the stresses which are evaluated as failure criterion will never
be reached in reality, because of plastic flow in the adhesive.
D
E
T

N
O
R
S
K
E

V
E
R
I
T
A
S
R
e
p
o
r
t

N
o
:

2
0
0
4
-
0
1
9
3
,

r
e
v
.

2

T
E
C
H
N
I
C
A
L

R
E
P
O
R
T
P
a
g
e

5
7

R
e
f
e
r
e
n
c
e

t
o

p
a
r
t

o
f

t
h
i
s

r
e
p
o
r
t

w
h
i
c
h

m
a
y

l
e
a
d

t
o

m
i
s
i
n
t
e
r
p
r
e
t
a
t
i
o
n

i
s

n
o
t

p
e
r
m
i
s
s
i
b
l
e
.

R
e
v
2
_
R
P
s
_
D
N
V
R
e
p
N
o
_
2
0
0
4
-
0
1
9
3
.
d
o
c
T
a
b
l
e

5
-
4
:

E
x
p
e
r
i
m
e
n
t
a
l

r
e
s
u
l
t
s

o
f

l
a
p
-
s
h
e
a
r
-
t
e
s
t
s

t
o

d
e
t
e
r
m
i
n
e

f
a
i
l
u
r
e

c
r
i
t
e
r
i
a

f
o
r

r
i
g
i
d

a
n
d

s
e
m
i
-
r
i
g
i
d

a
d
h
e
s
i
v
e
s

a
n
d

u
s
e
d

s
u
r
f
a
c
e
s

/

a
d
h
e
r
e
n
d
s

s
u
r
f
a
c
e

/

a
d
h
e
r
e
n
d
s

l
a
p
-
s
h
e
a
r

t
e
s
t

o
v
e
r
l
a
p
-
l
e
n
g
t
h

l

[
m
m
]

w
i
d
t
h

b

[
m
m
]

t
h
i
c
k
n
e
s
s

o
f

a
d
h
e
r
e
n
d
s

s

[
m
m
]

t
h
i
c
k
n
e
s
s

o
f

a
d
h
e
s
i
v
e

d

[
m
m
]

e
x
p
e
r
i
m
e
n
t
a
l

l
a
p

s
h
e
a
r

s
t
r
e
n
g
t
h

(
n
o

o
f

s
a
m
p
l
e
s
)

[
M
P
a
]

l
o
a
d

[
N
]

a
d
h
e
s
i
v
e


1

s
t
e
e
l

Q
S
t
E
3
8
0

T
M
,

s
u
r
f
a
c
e

V
I
I
I
,

s
h
o
t

b
l
a
s
t
e
d

w
i
t
h

c
o
r
u
n
d
u
m
,

(
E
=
2
5
5

G
P
a
,

v

=

0
.
3
3
)

T
a
b
l
e

6
.
3
-
1

o
f

1
-
1
2
-
W
-
2
0
0
1
-
0
5
-
0

T
a
b
l
e

3
.
1
.
2
-
1

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

1
3
.
1


0
.
3
6

2
5
.
0


0
.
0
1

2
.
5
3

0
.
3
5


0
.
1
2

3
8
.
4


0
.
7
5

(
3
)

1
2
5
7
6

V
a
n
t
i
c
o

A
r
a
l
d
i
t
e

4
2
0

E

=

1
8
9
5


2
7
.
7

M
P
a
,

v

=

0
.
3
9
4


0
.
0
4
4
5
,


G

=

6
7
9
.
3

M
P
a
,

G

w
a
s

c
a
l
c
u
l
a
t
e
d

f
r
o
m

E

a
n
d

v
2

a
l
u
m
i
n
i
u
m

5
0
8
3
,

s
u
r
f
a
c
e

I
I
I
,

g
r
o
u
n
d

w
i
t
h

1
2
0

g
r
a
i
n
,

a
d
h
e
s
i
v
e

p
r
i
m
e
r

P
C
1
2
0
,

E
=
7
1

G
P
a
,

v
=

0
.
3
3

T
a
b
l
e

6
.
1
-
2

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

T
a
b
l
e

3
.
1
.
1
-
2

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

1
3
.
2


0
.
5
4

2
4
.
7


0
.
3
9

6
.
0

0
.
9
2


0
.
1
0

1
5
.
0


0
.
9
8

(
5
)

4
8
9
0

P
l
e
x
u
s

M
A

5
5
0

E

=

3
1
0


4
2

M
P
a
,

v

=

0
.
4
8
0


0
.
1
4
1

G

=

1
0
4
.
8

M
P
a
,

G

w
a
s

c
a
l
c
u
l
a
t
e
d

f
r
o
m

E

a
n
d

v
3

s
t
e
e
l

g
r
a
d
e

A
,

1
.
0
4
4
1
,

S
t
3
4
k
,

s
u
r
f
a
c
e

I
I
,

w
i
t
h

P
3
.
1
,


E
=
2
1
8

G
P
a
,

v
=

0
.
3
3

T
a
b
l
e

6
.
1
-
1

o
f

1
-
1
2
-
W
-
2
0
0
1
-
0
5
-
0

T
a
b
l
e

3
.
1
.
1
-
1

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

1
2
.
6


0
.
1
0

2
4
.
9


0
.
2
6

6
.
0

2
.
1
2


0
.
2
1

1
0
.
5


0
.
3
4

(
3
)

3
2
9
4

s
e
e

2

4

s
t
e
e
l
,

g
r
a
d
e

A
,

1
.
0
4
4
1
,

S
t
3
4
k
,

s
u
r
f
a
c
e

I
,

s
h
o
t

b
l
a
s
t
e
d

w
i
t
h

c
o
r
u
n
d
u
m
,

a
f
t
e
r

a
g
e
i
n
g

w
i
t
h

p
r
o
t
e
c
t
i
o
n

a
g
a
i
n
s
t

c
o
r
r
o
s
i
o
n
E
=
2
1
8

G
P
a
,

v
=

0
.
3
3

T
a
b
l
e

6
.
2
-
1

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

T
a
b
l
e

3
.
1
.
2

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

1
3
.
3


0
.
5
1

2
5
.
0


0
.
0
7

6
.
0

0
.
7


0
.
1
1

2
8
.
6


0
.
3
8

(
3
)

9
5
0
9

s
e
e

1

5

s
t
e
e
l
,

g
r
a
d
e

A
,

1
.
0
4
4
1
,

S
t
3
4
k
,


s
u
r
f
a
c
e

X
I
I
,

w
h
i
t
e

t
o
p
c
o
a
t

C
4
.
1

b
e
f
o
r
e

a
n
d

a
f
t
e
r

a
g
e
i
n
g

o
r

s
h
o
p

p
r
i
m
e
r

P
3
.
1

E
=
2
1
8

G
P
a
,

v
=

0
.
3
3

T
a
b
l
e

6
.
3
-
1

o
f

1
-
1
2
-
W
-
2
0
0
1
-
0
5
-
0

T
a
b
l
e

3
.
1
.
2
-
1

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

T
a
b
l
e

6
.
2
-
1

o
f

1
-
1
2
-
D
-
2
0
0
2
-
0
1
-
0

1
3
.
4


0
.
1
5

2
4
.
9


0
.
1
9

6
.
0

0
.
7


0
.
3
9

1
6
.
5


0
.
4
8

(
3
)

5
5
0
5

s
e
e

1

DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 58
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 5-5: Failure criteria for Vantico Araldite 420 (A) and Plexus MA 550 (P), see also
previous table
[MPa] Sy (mean)
(peel
stress)
Txy
(mean)
(shear
stress)
S1 (mean)
(max.
principal
stress)
Seq (mean)
(Mises eq.
stress)
Sy (max)
(peel
stress)
Txy (max)
(shear
stress)
S1 (max)
(max.
principal
stress)
Seq (max)
(Mises eq.
stress)
1 (A) 85 58 131 114 224 64 234 172
2 (P) 12 15 25 28 48 16 49 40
3 (P) 6 10 15 19 26 11 26 25
4 (A) 32 32 58 59 106 33 108 85
5 (A) 18 18 33 33 62 19 63 50
The maximum service temperature will be 40C yielding a knock-down factor, f
T
, of 0.7 for the
strength and 0.6 for the elastic moduli E and G [4] of the adhesive. The knock down-factor, f
Df
,
for fatigue loading with thick adherends is 0.5 for dry conditions and
2 . 0 4 . 0 5 . 0 f f f
Df T , M T , M , Df
= = =
for humid conditions ([9], Table 3). For the surface preparation at the shipyard a knock-down
factor, f
S
, has to be applied compared to the conditions of surface preparation for the test sample:
f
S
= 0.7 [10]. The knock-down factor for creep conditions at static load, f
L
, would be 0.7 at 45C
[4]. The knock-down factors for the selected adhesives have been measured for a stress ratio of R
= 0.1. This means that a mean load of 0.45 o
a
is allowed up to 10
6
cycles, which partly considers
the creep which might appear. Because of the remaining uncertainties, we include an additional
safety factor S = 2 ([9], section 3.7.2).
The safety factor, S
LE
= 1.5, which takes account for the uncertainties of the numerical approach
has to be included only for the numerical design approach. According to ( )
j i c a
S , f f o = o (where
f
i
= knock-down factor, S
j
= safety factor, o
c
= characteristic value, o
a
= allowable stress; from
reference [5]) we obtain:
1. for dry conditions:
, MPa 1 . 3
S
f f f
S Df T
c a
=

o = o
2. for humid conditions:
, MPa 8 . 1
S
f f
S T , M , Df
c a
=

o = o
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 59
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3. for dry conditions and linear elastic numerical approach:
, MPa 0 . 2
S S
f f f
LE
S Df T
c a
=


o = o
4. for humid conditions and linear elastic numerical approach:
. MPa 2 . 1
S S
f f
LE
S T , M , Df
c a
=

o = o
Table 5-6: Comparison of failure criterion with numerical results of the case study
[MPa] S
y
(mean)
T
xy
(mean)
S
1
(mean)
S
eq
(mean)
S
y
(max) T
xy
(max) S
1
(max) S
eq
(max)
3 1.8 2.0 3.6 3.8 6.7 2.2 6.8 5.4
40C, dry conditions, 1500 N, adherend 1: width
8.8 mm, thickness 4 mm, overlap length 95 mm
3.5 5.6 5.5 9.8
maximum value / failure criterion 3.5/1.8 = 2 5.6/2.0 = 2.8 5.5/3.6 = 1.5 9.8/3.8 = 2.6
design: maximum value / failure criterion s 1 at
1500 N/2.8 = 535 N

4 1.1 1.2 2.1 2.3 4.0 1.3 4.1 3.2
40C, humid conditions, 1500 N, adherend 1:
width 8.8 mm, thickness 4 mm, overlap length
95 mm
3.5 5.6 5.5 9.8
maximum value / failure criterion 3.5/1.1 = 3.2 5.6/1.2 = 4.6 5.5/2.1 = 2.6 9.8/2.3 = 4.3
design: maximum value / failure criterion s 1 at
1500 N/4.6 = 326 N

4 1.1 1.2 2.1 2.3 4.0 1.3 4.1 3.2
40C, humid conditions, 1500 N, adherend 1:
width 60 mm, thickness 2 mm, , overlap length
95 mm
1.1 1.2 2.1 2.2
maximum value / failure criterion 1.1/1.1 = 1 1.2 / 1.2 = 1 2.1/2.1 = 1 2.2/2.3 = 1
design: maximum value / failure criterion s 1 at
1500 N

The initial thickness of adherend 1 (the attachment) is 4 mm. Since we have mainly a shear
loading of the joint we can take eq. 11-19 on page 189 (section 11.2.5) to determine the overlap
length with s
1
= 4 mm, s
2
= 24 mm at the lower end and s
2
= 7.4 mm at the other side of the joint,
G
adhesive
= 680 0.6 MPa, E
steel
= 215 MPa, d = 0.2 mm. For s
2
= 4 mm we obtain l
*
= 83 mm and
for s
2
= 24 mm the overlap length should be more than 95 mm (remind o s 1). 95 mm overlap
length is selected now for further discussion. The width of the joint can first be estimated by
keeping the nominal stress below the allowed limits. For dry conditions we obtain a minimum
width b = 5.1 mm with an overlap length of 95 mm and an applied load of 1500 N. The tensile
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 60
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
stress in adherend 1 would be 75 MPa. For humid conditions the minimum width would be 8.8
mm. Therefore, we take 8.8 mm as initial width of adherend 1.
In the next step, the method described in section 5.4.3.1 has to be used to determine the failure
criterion for the numerical approach with applied loads as in rows designated 3 and 4 in Table
5-6 and the conditions of row 4 in Table 5-4. The numerical analysis shows that the required
load of 1500 N cannot be carried by the adhesive joint at the conditions which occur (Table 5-6).
The required load is for 40C and humid conditions about 5 times as high as the load capacity of
the joint.
The left hand side of Figure 5-21 shows the displacement u
y
at the selected boundary conditions.
Obviously, the upper end of the joint is loaded in peel while at the lower end of the joint
compression stresses occur. Therefore, for further optimisation of the design, the upper end of
the joint has to be considered. The stresses will be evaluated along two lines, one parallel to the
x-axis and one parallel to the y-axis (Figure 5-21) at d/10 from the profile. Since, the upper side
of the joint is the critical side, the higher stresses will appear close to the profile.
Figure 5-21: Case study of attachment below deck
Since the geometry and the material of the profile (adherend 2) cannot be changed, one way to
reduce the stresses, namely, the reduction of the thickness of the adherend 2 or changing the
material, cannot be followed. We can expect that increasing the width and reducing the thickness
of adherend 1 will reduce stress concentrations at the ends of the joint. The last four rows of
Table 5-6 show that the requirement are reached with a 60 mm width and 2 mm thickness of
adherend 1 and an overlap length of 95 mm. The last optimisation step requires a finite element
analysis.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 61
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-22: Stress distribution at upper end of adhesive joint, x = const
In Figure 5-22 the stress distribution at the upper end of the adhesive joint (x = 25 mm) is
plotted. Similar to the stress distribution in shear direction maximum stresses occur and both
ends of the adherend. The shear stress might appear negative, depending on the coordinate
system of the model. For the shear stress always the absolute values have to be considered.
Figure 5-23: Stress distribution along adhesive joint at w/2
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 62
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 5-23 shows the stress distribution along the joint at z = 50 mm. The peel stress at the
lower end of the joint (x = 120 mm) is negative, which means it is compressive as we have
expected. The shear stress is approximately the same as on the upper side of the joint. The
components of stress s
y
and S
1
are oscillating between 40 and 100 mm. This is typical for the p-
element finite element code if stresses are close to 0.
5.7 References
[1] R.D. Adams, J. Comyn and W.C. Wake, Structural Adhesive Joints in Engineering,
Chapman and Hall, London, ISDN 0-412-70920-1, 1997
[2] Course notes European Adhesive Engineer (week 6), IFAM, Bremen, 2002 (in
German)
[3] DNV rules for Classification of Ships or Classification of High Speed, Light Craft
[4] Markus Brede, Final report on coupon tests for characterisation of adhesives and
surface coatings, BONDSHIP Report Number: 1-12-W-2001-05-0, 2002
[5] Markus Brede, Final report on analytical and FE modelling of joints and verification
of easy-to-use design rules, BONDSHIP Report Number: 1-11-D-2002-01-0; 2003
[6] Brede et al., Screening test results from IFAM and Sika, BONDSHIP Report Number:
2-22-W-2001-01-0
[7] Raghava et al., The macroscopic yield behaviour of polymers, Journal of Materials
Science, 8 (1973), 225-232,
[8] Schleicher, Der Spannungszustand an der Fliegrenze, Ztschr. f. angew. Math. und
Mech., Bd. 6, Heft 3, 1926
[9] Markus Brede, Final report on ageing and fatigue tests, BONDSHIP Report Number:
1-12-D-2002-01-0, 2003
[10] Markus Brede, Test Results from Testing the Prototype, BONDSHIP Report Number:
5-52-D-2002-03-0, 2003
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 63
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
6 TESTING OF MATERIALS AND STRUCTURES
6.1 Introduction
Experimental testing of bonded joints and their materials is carried out for various purposes
during design and qualification of the joints:
o Screening tests are carried out in order to establish the basis for selecting suitable
materials and manufacturing procedures. Such tests are covered separately in chapter 3.
o Material characterisation tests are carried out in order to characterise the properties of the
materials present in the joint for input to theoretical models. Some material
characterisation tests of particular relevance for bonded joints are covered in Section 6.2.
o For joint strength prediction, tailor made tests are sometimes used together with
theoretical models to establish critical strength parameters. Such tests are described
together with the appropriate theoretical models in chapter 5.
o For qualification of the joints, dedicated qualification tests of the full-scale joints are
sometimes required. Recommendations for such tests are provided in section 6.3.
o Ageing and fatigue tests are addressed separately in section 6.4.
6.2 Materials characterisation tests
Material characterisation tests are carried out to establish the material inputs required to make
use of theoretical models of joint behaviour and joint resistance. Finite element models usually
require a large amount of material data whereas simplified analytic models often only require the
most critical ones.
Some of the material parameters describe the response of the material to external actions up till
failure. These include for example the shear modulus (G), the Poissons ratio (v) and, for a
tough rigid adhesive, the yield stress (t
y
). Other material factors govern the final fracture of the
adhesive. These are the parameters that enter the failure criterion (limit state equation) and are
normally harder to define and quantify reliably than the other material properties. Dedicated
representative tests may need to be defined to obtain them. In the following, these parameters
are denoted characteristic strength parameters.
Some standard tests are reviewed in the following, and recommendations are provided for how to
use the test results to obtain inputs to theoretical models.
6.2.1 Thick adherend shear test (TAST)
The thick adherend shear test geometry is described in the standard DIN 54 451 (equivalently in
standard NF 76 141) and also in Figure 6-1. The attraction of this test is that the rotation of
adherences is reduced due to the increased adherend thickness. Higher strength values are
obtained than e.g. with a standard lap-shear test. Samples are machined into bonded plates of
dimensions 120 x 150 mm. These plates are obtained by bonding two aluminium plates (2024T3
series) which are the substrates. The substrates are first grit blasted with fresh alumina grit (size
150m, pressure 7 bars) then degreased with acetone before bonding. During fabrication, the
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 64
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
thickness of the joint has to be homogeneous and is calibrated to the thickness of 0.2mm. After
curing and machining of the samples, the thickness of the joint is measured by an optical
microscope.
Figure 6-1: Thick adherend shear test samples used for shear experiments.
During the tests, conducted at a velocity of 1 mm/min, the relative displacement of the two
substrates is measured by placing a specific extensometer which is based on the use of two
inductive LVDT sensors placed on each side of the sample. Before each series of testing, a
calibration of this measurement is made by testing a blank sample. A blank sample has the same
geometry as a thick adherend shear test sample but there is no adhesive layer between the two
substrates: the transmission of the load between the left and right part of the sample is made by
the metal itself.
6.2.2 Simple shear strength test
6.2.2.1 Introduction
The test described in this section is intended to establish the shear strength and stiffness of an
elastic adhesive. If used to test rigid adhesives, fracture should be expected to be influenced by
stress concentrations and the strain energy in the test setup that is quite different from that of a
bonded joint.
The shear strength of bonded joints depends on the thickness of the adhesive layer. In
consequence the dependence of the shear strength on the thickness of the adhesive layer is
needed. For the lap shear samples of normally used, the thickness changes during testing. This is
especially true for thick adhesive layers. In order to determine the influence of the thickness of
the adhesive layer to lap shear strength IFAM has developed a testing method to keep the
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 65
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
thickness of an adhesive layer constant during testing. This test method is described in the
following paragraphs.
6.2.2.2 Equipment
The testing methods refers to the DIN EN 1465. In addition constraints are needed to ensure
constant thickness of the adhesive layer during testing. Figure 6-2 shows the test equipment for
this test.
Figure 6-2: Simple shear testing equipment.
Figure 6-2 shows the special testing machine without testing samples with constraint no. 2 in an
upper position. The tested sample is fixed in constraints 1 and 2. The resulting shear strength
when moving the constraints in opposite directions is measured. One constraint is in a fixed
position and the other constraint is moving up or down.
6.2.2.3 Samples
The following figure shows a sketch of the adherends fitted to the constraints and their
dimensions:
1
2
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 66
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
30,00 mm 10,00 mm
50,00 mm
20,00 mm
A
d
h
e
s
i
v
e

L
a
y
e
r
20,00 mm
20,00 mm
Figure 6-3: Sample geometry for simple shear test.
The adherends are made of Aluminium (Al Mg Si 0.5).
Preparation of bonding area (grey area) before application of adhesive is the following:
1. cleaning with MEK
2. shot blasting
3. cleaning with special adhesive cleaner or MEK
4. Primer
The thickness of the adhesive layer can be varied and should reflect the adhesive thicknesses of
the joints to be modelled, which normally depends on the flexibility of the tested adhesive. The
more flexible an adhesive is, the thicker the adhesive layers may need to be tested.
It is very important that the adhesives are fully cured. To achieve this, the samples may need to
be cured in a climate chamber.
Three samples of each thickness should be tested.
6.2.2.4 Definitions
The nominal shear stress t
n
is defined as:
eq. 6-1
0
A
F
n
= t
where F is the applied load and A
0
is the initial bonding area before the specimen is loaded.
The nominal shear strain is defined as:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 67
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
eq. 6-2
a
t
s A
= tan
with As is the translational displacement and t
a
is the thickness of the adhesive layer.
With G as the slope of the nominal shear stress vs. the nominal shear curve t can be calculated
as:
eq. 6-3
t tan = G
6.2.2.5 Simplified method to measure strength for different adhesive systems
The following approach was suggested by Sika. This applies to flexible adhesives bonded to
painted surfaces. The basic requirement is that the adhesive system shows cohesive failure in the
adhesive in the lap-shear strength test and that one obtains very good results with the bead peel
test. If this adhesive is then evaluated for a new painted surface then only the bead peel test is
repeated. The adhesive is acceptable if it passes the bead peel test. The rational is that experience
has shown that for flexible adhesives the paint is always stronger than the adhesive also long-
term. Hence it is quite acceptable to use the standard lap-shear strength values for different
painted surfaces.
6.2.3 True stress / true strain behaviour, EModulus, Poissons ratio
6.2.3.1 Introduction
This section describes how the nonlinear behaviour of an adhesive in a one-dimensional tensile
stress state can be established by a simple test. This is normally required as input to numeric
simulation of bonded joints that take account of the non-linear behaviour of the adhesive.
The ultimate tensile strain measured as described below is normally not considered
representative of the ultimate fracture of a bonded joint because both the stress state and the
strain energy in the test setup is quite different from that in bonded joints.
The mechanical behaviour of adhesives depends strongly on temperature. Adhesives become
more flexible and more ductile as temperature rises. On the other hand adhesives could become
brittle at low temperatures. The service temperature range varies for different adhesives. Hence,
it is necessary to establish the temperature-dependent properties of an adhesive by testing.
Reference is made to the standard DIN 53 504: May 1994, testing of rubber and DIN EN ISO
527 2, plastics determination of tensile properties.
6.2.3.2 Samples
Samples are made of pure adhesive with the same sample size for each adhesive. In addition to
total curing in PTFE- templates (at 23C/50% rH) the samples are refinished to ensure perfect
geometry. Samples made out of rigid adhesives (e.g. Vantico Araldite 420 or Plexus MA 550)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 68
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
could be punched out of an adhesive plate with a special stamp. Figure 6-4 shows a test sample
and its dimension:
Figure 6-4: Pure adhesive sample for tensile testing.
Typical specimen dimensions:
l
3
, overall length: 150 mm
l
2
, min. distance between clamped sections : 109 mm
l
1
, narrow section length : 80 2 mm
r (+), fillet radius : 20 25 mm
b
1
, width of measure distance : 10 0.2 mm
b
2
; width of clamped section : 20 0.2 mm
h, sample thickness : 4 0.2 mm
L
0
, measure length : 50.0 0.5 mm
L, max. distance between clamped sections: 115 1 mm
6.2.3.3 Test procedure
Samples are fixed in a tensile testing machine. Samples are then loaded at a strain rate of 0.5
min
-1
for flexible adhesives and 0.05 min
-1
for rigid adhesives. Each adhesive is tested three times
at three different temperatures (-30C, RT, +70C). The samples are tested to rupture with
measuring of load, axial elongation and lateral contraction.
6.2.3.4 Definitions
The true strain c
true
is calculated from the nominal strain c
n
with:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 69
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
eq. 6-4
)
+ = =
l
l
n true
l
dl
0
) 1 ln(
'
'
c c
With the assumption of constant volume the true stress o
true
can be calculated from the nominal
stress o
n
and the nominal strain c
n
as:
eq. 6-5
) 1 ( + =
n n true
c o o
The Poisson `s ratio is defined as:
eq. 6-6
0
0
B m
L m
d
d
d
d
L
T
L
T

= =
o
c
o
c

m
T
and m
L
are the slopes of the transversal and the longitudinal deflection vs. force curve in the
linear range. L
0
and B
0
are the initial length and width of the sample.
The Youngs Modulus E is defined as:
eq. 6-7
n
n
E
c
o
=
The E-Modulus is evaluated from the slope of the longitudinal stress-strain curve.
6.2.4 Compression Tests Rhagava equivalent stress
6.2.4.1 Introduction
Adhesives and other polymers usually show different behaviour in compression and in tension.
Besides the temperature dependency of both kinds of stress, the absolute value of transition to
non linearity is not necessarily the same. This section describes a compression test that allows to
determine the compressive properties of an adhesive to complement the tensile properties as
established by the test in Section 6.2.2.5.
Compression tests are made with reference to DIN EN ISO 604: 1997-02, Druckversuche an
Kunststoffen.
6.2.4.2 Samples and Testing
The geometry of compression samples has to be different from the true stress/ true strain
samples. Samples should not crack because of their slenderness ratio. The specimen geometry
shown in Figure 6-5 is recommended.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 70
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
h = 10 mm
d = 20 mm
Figure 6-5: Compression test sample
Three samples of each adhesive are tested at 3 different temperatures. To receive comparable
results with true stress / true strain tests these temperatures are the same as for the tensile tests:
- -30C
- RT
- +70C
The samples have to be as homogeneous as possible. They need to be totally cured out with no
bubbles or any other irregularities. The top and bottom surfaces have to be parallel and as
smooth as possible.
Samples are compressed in the testing machine until failure occurs or if no failure occurs to a
compression of 60 %. Compression strain rate should typically be 0.05 min
-1
rigid adhesives
(e.g. Plexus MA 550 and Vantico Araldite 420) and 0.5 min
-1
for flexible adhesives (e.g.
Terostat 8590 UHV/M). To receive comparable results with tensile tests these compressive strain
speeds must be identical to the tensile strain speeds.
6.2.4.3 Definitions
For polymers an equivalent stress is defined by Rhagava. The equivalent stress which considers
the special physics of polymers to some extent is the Rhagava equivalent stress which is a
modified von Mises equivalent stress:
eq. 6-8
( ) ( ) ( )
R
R R R
V

+ + +
=
2
3 4 1 1
2 2 2 2
t o o o
o and
eq. 6-9
yield
tension
yield
n compressio
R
o
o
=
with
o
v
: equivalent stress
o : sum of the maximum applied normal stresses
t : sum of the maximum applied shear stresses
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 71
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The Rhagava equivalent stress R is the ratio between the compressive yield stress and the tension
yield stress. The stress value where the nominal stress vs. nominal strain curve becomes
nonlinear is defined as the yield stress. The compressive stress itself is defined as:
eq. 6-10
0
A
F
cn
= o
with F as the compressive load and A
0
as the starting compression area of the sample.
The compressive strain is defined as
eq. 6-11
0
h
h
cn
A
= c
This formula assumes that the cylindrical shape of the specimen is maintained during the test.
This is realistic because the plates of the testing machine between which the specimen is placed
are lubricated to allow the specimen to expand during testing. Furthermore, measurements of the
true stresses and strains are valid in the elastic range where deviations from the cylindrical shape
are negligible.
As for the true tension stress and strain, the true compressive stress o
ct
and true compressive
strain c
ct
are defined as:
eq. 6-12 and eq. 6-13
) 1 (
cn cn ct
c o o =
) 1 ln(
cn c
c c =
6.2.5 Creep strength
6.2.5.1 Introduction
This section describes how to determine the long term shear properties of an adhesive at room
temperature. No ageing effects, like humidity and raised temperature are considered or
evaluated. Static shear properties are found first, thereafter stress rupture properties.
The test is primarily intended for flexible adhesives. If used for rigid adhesives, it is
recommended to replace the specimen geometry with one with less peel stresses. The thick
adherend specimen of section 6.2.1 or double-lap or double-lap specimens may be more
representative of well designed joints.
Basis of the static shear tests are EN ISO 527 1 and ISO 4587. In addition the DNV Offshore
Standard DNV-OS-C501 is also referred.
6.2.5.2 Requirements
The following points are required for the test procedure:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 72
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Consideration of each application case with qualified surfaces according to the results of
the screening test programme.
Documentation and evaluation of the adhesive and the bond surface - adhesive and
cohesive behaviour of the specified adhesive.
Six samples have to be tested for the static test and at least 15 samples should be used for
the long term properties with fairly even distribution over the plot of log time.
6.2.5.3 Surfaces and samples
The surfaces should be suitably prepared prior to bonding. The test specimen geometry is shown
in Figure 6-6.
Figure 6-6: Form and dimensions of test specimen.
6.2.5.4 Testing
The specimens are tested in room temperature at 23 C
0
. The pH- value should be in-between 6
and 8.
Static shear strength test procedure:
- The geometric properties such as adhesive layer thickness and overlap area,
0
A , is
determined.
- The samples are loaded in shear. The load and the cross head deflection are measured. The
samples are tested until failure.
- The test speed has to be found experimentally. Fracture should occur approximately within
180 sec. (For Sikaflex 292 the test speed was 3 mm/min.). Test speed is a very important
parameter in this test: in BONDSHIP velocities between 3 (DNV) 20 (Sika) mm/min have
been used.
- Load F
u
to rupture has to be measured. The lap shear strength o is determined:
eq. 6-14
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 73
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
0
A
F
u
= o
- Characteristic shear strength values,
C
o , is established with 95 % confidence. References are
made to DNV Offshore Standard DNV-OS-C501. The characteristic shear strength value is :
eq. 6-15
o o o

m C
k =
where
m
k is constant depending on number of test samples. o is the standard deviation
estimated from the measurements.
6.2.5.5 Stress rupture strength
- If the linear relationship cannot be documented, an equivalent approach shall be used, taking
the non-linearity into account.
- Values shall be based on the data that are fairly evenly distributed over plot of log time to
failure vs. log load. At least 15 data points should be used.
- To obtain the characteristic curve the mean stress rupture curve of the form:
eq. 6-16
t log log log
0
| o o =
shall be converted to the form:
eq. 6-17
o log log log
0
k X X =
where X represents the time to failure under a sustained stress o , and k =1/ | .
- The mean curve can be transformed back into the standard formulation of a stress rupture
curve using the equations as given above.
6.3 Testing of full scale joint samples representative of real Application Cases
6.3.1 Introduction
Real application cases may be qualified on the basis of testing of full scale samples of bonded
joints representative of the application case. Such tests are briefly discussed in this section.
Alternatively, full scale tests could also be used to provide the basis for validation of theoretical
models. Preferably, such tests should also follow the recommendations below. Two examples
are discussed to shed light on the practicalities of complying with the recommendations.
6.3.2 General recommendations
The joint tests must be specially designed on a case by case basis to ensure that the test is
representative of the application case. The geometry of the joint should as far as possible be
identical to that of the real full-scale joint. The loading of the joint and the critical failure
mechanisms must be fully understood. That normally requires a response analysis of the
structure where load cases are applied that reflect the critical events that the structure may
experience during its lifetime. The test samples usually represent a rather small part of the
structure, sometimes with dedicated load introduction and support devices. The extent and
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 74
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
geometry of the test sample as well as boundary conditions and load introduction points must be
carefully designed such as to be representative of the real joint, how it is loaded in the real
structure and the resulting failure mechanism. The loading mode that governs the critical failure
modes must be applied. Such loads may be static, dynamic, cyclic or long term creep loads.
Environmental conditions that may influence the failure mechanism should be included. This
includes temperature and humidity.
6.3.3 Example 1: Bonded pillar
A relevant application of adhesive bonding in shipbuilding is to bond vertical pillars (and similar
items) to the deck as support for equipment [1]. The primary loading on such pillars would be
compression. However, this would load the bonded joint in its most favourable mode, and joint
failure is unlikely. More problematic would be a transverse loading on the pillar which would
produce bending and shear loading of the bonded joint. The joint must be qualified to survive
such loads.
An analysis of such a loading case would provide information on the relative magnitude of
bending and shear loading at the joint. A simple test setup is illustrated in Figure 6-7. The test
should be designed in such a way that the loading arm (i.e. the vertical distance from the bonded
joint to the clamping device, see Figure 6-7) produces the appropriate combination of bending
moment and shear force at the joint. If it is justified that the bending behaviour dominates, one
would only need to ensure that the loading arm is sufficiently long to ensure that bending
dominates also in the test. Taking due account of these issues, the test setup of Figure 6-7 can be
used to qualify this bonded joint.

Figure 6-7: Bonded pillar sample prepared for testing (left) and sketch of test setup (right).
6.3.4 Example 2: Bonded joint in steel sandwich panels
Another relevant application of adhesive bonding in shipbuilding is joining of prefabricated
laser-welded steel panels by adhesive bonding [1]. Figure 6-8 shows one such joint. A
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 75
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
structural load-response analysis of the structure with the joint would reveal the primary mode of
loading that the joint is exposed to.
Figure 6-8: Bonded joint between steel sandwich panels (left) and four-point bending test
setup (right).
A common test method for sandwich structures is the so-called four-point bending test. Figure
6-8 shows the bonded joint tested in four-point bending. The picture illustrates that the test
sample is so flexible in the mode in which it is loaded that it fails to expose the bonded joint to a
critical load. This is also the case in practice: if the sandwich panel is loaded transversally, then
this particular bonded joint will remain virtually unloaded. However, when exposed to in-plane
loading, the forces in the steel face sheets will have to be transmitted through the bonded joint.
A structural response analysis would show that this would be the critical loading mode. A
tensile test as shown in Figure 6-9 represents this loading case and would be suitable for
qualification of this bonded joint.
Figure 6-9: Schematic of steel sandwich joint tested in tension.
6.4 Environmental degradation ageing and fatigue
6.4.1 Introduction
Qualification must be based on a joint strength estimate that is representative for the whole
intended lifetime of the joint. To the extent the service environment may affect the joint
strength, this must be accounted for.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 76
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
This section first identifies some reasonable assumptions to be made and then identifies the most
important environmental factors. It concludes with some recommendations for how to obtain
representative strength estimates for qualification.
6.4.2 Major assumptions
The major assumption that can be made is that the material combination specified for the joint
has passed a screening test programme that aims at eliminating materials and material
combinations that are not suitable for the intended application. Recommendations for such
screening tests are provided separately in chapter 3. On this basis, one may assume that the
material combinations specified for a particular joint have the following properties:
o The materials (adhesives, primers, adherends etc) are chemically compatible such that no
chemical attack occurs when they are combined as specified.
o The individual materials are chemically stable when used as specified in the intended
operating environment (chemicals, UV light, moisture, temperature etc.)
6.4.3 Environmental factors
Polymeric materials, such as adhesives, primers, paints etc, absorb moisture when used in a
humid environment. Such moisture absorption causes swelling and plasticization. Swelling
causes internal stresses that may affect joint strength. Plasticization is a process whereby the
polymer material changes mechanical properties. Typically, the shear stress at fracture decreases
whereas the shear strain at fracture increases. Furthermore, one should expect that the creep rate
increases. This would affect the strength of the joint.
Although materials that are prone to attack by chemicals would be eliminated via a screening test
programme, the polymeric materials may also absorb chemicals and hence change properties in a
similar way as described for moisture absorption.
The properties of polymeric materials also depend on temperature. With increasing temperature,
one should expect that the shear stress at fracture decreases whereas the shear strain at fracture
increases and creep rate increases.
Cyclic loading may initiate and propagate cracks that could eventually cause fatigue fracture.
The fatigue behaviour may depend on the environmental factors discussed above. A constant
sustained loading may cause creep deformation and eventually creep rupture. Again, the
behaviour may depend on the environmental factors discussed above.
As explained above, materials prone to chemical and physical ageing would be eliminated
through the screening test programme. However, such screening tests must be cheap and quick
to be suitable for screening. Therefore some uncertainty would remain regarding slowly
propagating degradation mechanisms. When metal adherends are bonded, attacks on the metal
surface would be of particular concern. Accelerated ageing tests may shed light on such
degradation mechanisms, but within reasonable testing times, some uncertainty will remain
regarding degradation during the intended lifetime of the joint. An extra safety margin based on
previous service experience and professional judgement needs to be included to account for this.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 77
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
6.4.4 Representative testing
For the governing failure modes, care should be exercised so as to ensure that the effects of
environmental factors are properly documented and accounted for.
For joints bonded with elastic adhesives where creep normally governs, the creep behaviour in
different environments needs to be established. In particular, effects of temperature would be
important and, depending on the location of the joint, perhaps also the effect of moisture
absorption.
For joints bonded with rigid adhesives, extreme load events normally govern. Then, strength
degradation due to prolonged exposure to the operating environment before the extreme load
event occurs must be assessed. Furthermore, if intended to operate at low or high temperatures,
tests may need to be carried out at representative temperatures in addition to room temperature.
The tests to establish representative long term properties may be carried out using the standard
methods described in Section 6.2 in a controlled environment and, when relevant, on suitably
conditioned (aged) samples. Special tests may also be designed if that would help obtaining
representative results.
6.5 References
[1] Markus Brede, Test Results from Testing the Prototype, BONDSHIP Report Number:
5-52-D-2002-03-0; 2003
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 78
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
7 FIRE SAFETY
This chapter is based on references [1] and [2].
7.1 Introduction
Many structural divisions in a marine vessel have requirement to fire resistance, i.e. fire
insulation, smoke and flame tightness, and structural integrity. Unlike the joint in welded or
riveted structures, the adhesive bond will usually be the weak link in a load carrying structure
during a fire. The temperature development and distribution in a bonded joint is very complex
and dynamic, and although thermal FEM-analysis can be used to predict the temperatures in a
structure, the only reliable way of verifying the capacity of the joint is to measure temperatures
in the joint during a fire test while applying load during a realistic fire condition. Alternatively, if
the thermo-mechanical properties of the adhesive are known, it might be sufficient to measure
temperature in the (unloaded) joint and then predict structural lifetime numerically.
7.2 Governing documents and general requirements
The IMO Code of Safety for High Speed Craft (HSC-code) of Jan. 1996 introduced new fire
safety requirements for high-speed crafts; they utilise a functional approach i.e. bonding is
an option (FRP is also allowed), if performance can be proven.
The SOLAS (Safety Of Life At Sea) convention is simple and prescriptive: Steel or equivalent
material in all structures (also deckhouses), and all structural and insulation materials shall be
non-combustible. Approval of bonding might be given on a case-by-case basis. Recently
SOLAS has opened up for possibilities of using alternative designs on SOLAS ships in their
amendments published in 2000 (Chapter II-2: Construction - fire protection, detection,
extinction; Part F: Alternative design and arrangements).
SOLAS have defined one area where aluminium or bonding will not be allowed: The enclosure
around the engine rooms and the casing (up to the fire dampers). This should be made of welded
steel, because a fire can (often) last for much longer than the 60 minutes required for the fire
resisting divisions. Boundary cooling will usually be used actively, but high temperatures will
develop in areas with difficult access, and bonded or aluminium structure cannot withstand this
scenario.
For cruise vessels, the most critical approval authority is the US Coast Guard (USCG). USCG
often requires classification society approval first, and this implies a certain conservatism of the
classification societies: It can have large economic consequences for the yard and ship owner if
USCG refuses a ship that has class approval, because this can lead to added work, also for the
classification tasks, and maybe also delays in final approval (risk for customer).
Formal Safety Assessment (FSA) gives an opening to meet the requirements in other ways than
prescribed. FSA involves considerable work. Procedures are described by IMO. DNVs fire
department advise for the BONDSHIP programme: SOLAS is not written with bonding in
mind, i.e. do not look too much to SOLAS, but focus on function and equivalence in fire safety,
define application areas very carefully, propose solutions and tests, and verify in accordance with
agreements when it comes to real applications (often full scale = not scope of BONDSHIP).
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 79
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The FTP code (1998): International Code for Application of Fire Test Procedures (Res.
MSC.61(67) including fire test procedures referred to in and relevant to the FTP Code) is a
document where fire test procedures, fire tests and requirements for SOLAS-ships and HSCs is
gathered in one volume.
7.3 Fire engineering terms
There are two distinct areas of fire performance and fire behaviour, and hence also fire
engineering:
7.3.1 Fire reaction
Fire reaction describes a materials reaction to fire. Fire reaction properties are of major
importance for the fire and smoke development in the early stages of a fire. Some typical fire
reaction terms and properties are:
- time to ignition
- spread of flame
- heat release
- smoke, toxicity, etc .
Tests that determine these types of properties are all material tests. The surface of a product is
often the governing parameter for the fire reaction properties. However, when the materials
behind the surface are combustible, as for composite structures, they can give significant
contribution to the fire. Small-scale tests are often developed to evaluate the surface on a non-
combustible backing. However, the IMO HSC-code specifies a full-scale test to qualify a so-
called Fire Restricting Material, i.e. where more than just the surface is classified as
combustible.
7.3.2 Fire resistance
Fire resistance is a description of the ability for a construction to prevent a fully developed fire
spreading from one compartment to a neighbour compartment within a specified time. The most
important parameters are:
- fire insulation (T
ave
<140C on unexposed side of structure)
- smoke & flame tightness
- structural integrity during fire (load carrying)
Combustibility, heat release, smoke is not fire resistance requirements, and such parameters is
not measured during a fire resistance test. Fire resistance is a requirement to a fire resisting
division (e.g. B30, A60, ) and this can be a deck or a bulkhead, or the shipside. Tests that
determine these types of properties are tests on constructions. These tests are designed to
represent a realistic full-scale fire division (outside the scope of the BONDSHIP project).
7.4 SOLAS requirements
7.4.1 General
In section 7.2 of this report it is stated that SOLAS is prescriptive. However, the FTP code also
specifies some alternative routes, in Part I, section 7.1:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 80
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
To allow modern technology and development of products, the Administration may approve
products to be installed on board ships based on tests and verifications not specifically
mentioned in this Code but considered by the Administration to be equivalent with the applicable
fire safety requirements of the Convention
Procedures for this is given in Part I, section 7.2
The Administration shall inform the Organization of approvals referenced to in paragraph 7.1
in accordance with regulation I/5 of the Convention and follow the documentation procedures as
outlined below:
.1 in the case of new and unconventional products, a written analysis as to why the existing
test method(s) cannot be used to test this specific products;
.2 a written analysis showing how the proposed alternative test procedure will prove
performance as required by the Convention; and
.3 a written analysis comparing the proposed alternative test procedure to the required
procedure in the Code.
These two paragraphs makes it possible to propose relevant test procedures for bonded joints,
even if tests are not described by the FTP Code or other IMO documents. The same procedure
has been formally introduced into the SOLAS rules by the amendments published in 2000
(Chapter II-2: Construction - fire protection, detection, extinction; Part F: Alternative design and
arrangements).
7.4.2 Fire reaction requirements
The conservative approach is to try to follow all statements in the code regarding adhesives and
bonding, some examples from the FTP code are:
Part I, Annex 1, Part 3, section 3.1: The integrity of class B constructions shall be achieved
with non-combustible materials. Adhesives used in the construction of the specimens are not
required to be non-combustible; however, for the purpose of this Code, they shall have low
flame spread characteristics.
This is obviously relevant for adhesives that might be exposed to fire, e.g. adhesives in
honeycomb structure or adhesives for insulation, i.e. large areas. The requirement is not relevant
for bonded structural joints, due to the fact the bondline is always covered by a steel or
aluminium component only the thin edges can be exposed, and this can only happen if no fire
insulation is applied. Flame spread over free surface is therefore a phenomenon that cannot
occur for such use of adhesive, and the IMO test described for surface flame spread (IMO Res.
A653(16)), is not suitable for testing bonded joints. If there should be any fire reaction
requirement to the adhesive, it must be based on other tests (Oxygen index/self extinguishing or
similar). However, due to the non-exposed nature and also the small amounts it makes little
sense to put fire reaction requirements on the adhesive in the BONDSHIP type of joints. The
amount of adhesives used for bonding is also very small compared to the rest of the structure and
the combustible matters that are allowed in e.g. bulkhead liners and ceilings. It is also important
to have in mind that all additives used to make a so-called fire retarded adhesive system have a
negative impact on the mechanical properties.
Based on the above arguments, fire reaction tests are not identified as relevant for the bonded
applications. This analysis should fulfil the requirements of the paragraphs in the FTP-Code that
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 81
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
is referred to in section 7.4.1 (page 79). However, US Coast Guard, or others, might disagree.
This is mentioned to highlight the fact that this is an area where no good regulations exist that
can be used to avoid subjective evaluations.
7.4.3 Fire resistance requirements
7.4.3.1 General
According to SOLAS, both A and B class divisions shall be tested with the standard
(cellulose) fire curve as described in ISO 834, Part 1. In the SOLAS regulations it is stated that
fire divisions of A type shall be constructed of steel or other equivalent material and be
insulated with approved non-combustible materials, while the B type shall be constructed of
non-combustible materials. Common for the A and B divisions are full scale test
requirement, i.e. minimum 2.44 m x 2.5 m for the bulkhead test samples, and minimum 2.44 x
3.040 m for the decks. Details on test specimens and test procedures can be found in the FTP
Code, Part II IMO Res. 754(18) Recommendation on fire resistance tests for A, B and F
class divisions
Following requirement is found in the FTP Code, Part I, Annex 1, Part 3, paragraph 3.2:
Materials placed at B class panel joints for avoiding vibration or noise transmission shall be
of low flame spread characteristics and fire tested with B class divisions along which they are
used. However, such materials shall be non-combustible if they are necessary to support the
non-combustible B class structure or to achieve the required integrity.
This paragraph together with paragraph 3.1 referred to in section 7.4.2 (on page 80) seem to be
quite effective in stopping the use of adhesives as long as they are load bearing during the fire!
However, the requirement of non-combustibility has no direct link to structural performance, and
it should be possible to convince the Administration that evaluations in line with those done
when aluminium were allowed should be possible also for bonded joints, i.e. equivalency can be
achieved by specification of proper fire insulation.
However, if the joints are designed in a way that ensures integrity without strength in the
adhesive joint, e.g. with some mechanical locks, this will make the approval process easier.
7.4.3.2 A class divisions
A constructions shall be tested for smoke and flame tightness and structural integrity for 60
minutes. The time given together with the character designation gives the time requirement to
reach temperature limits on the backside of the fire divisions. A15 has temperature
requirements for 15 min, but shall still be tested for tightness and integrity for 60 minutes. A60
has temperature requirements for 60 minutes. Aluminium can be accepted in A divisions,
provided that the temperature rise in the structural core (includes e.g. stiffener flanges) never
exceeds 200C (to ensure structural capacity of more than 50 % compared to RT properties).
The consequence of this is that an A0 division in aluminium needs almost equal amount of fire
insulation as an A60 division in aluminium, if the division is load bearing.
The temperature requirements for A class divisions are:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 82
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
- Average temperature rise of less than 140C on the unexposed side for the given time
period (0, 15 , 30 or 60 min)
- Maximum temperature rise of less than 180C on the unexposed side, including joints,
stiffener connections, for the given time period (0, 15 , 30 or 60 min)
7.4.3.3 B class divisions
B class divisions shall be tested for flame tightness for 30 minutes. The time given together
with the character designation gives the time requirement to reach temperature limits on the
backside of the fire divisions. B15 has temperature requirements for 15 min, but shall still be
tested for tightness and integrity for 30 minutes. Aluminium can be accepted also in B
divisions, provided that the temperature rise in the structural core (includes e.g. stiffener flanges)
never exceeds 200C, provided that the structure is load bearing. The consequence of this is that
an B0 division in aluminium needs almost equal amount of fire insulation as an B30 division
in aluminium.
The temperature requirements for B class divisions are:
- Average temperature rise of less than 140C on the unexposed side for the given time
period (0, 15 , or 30 min)
- Maximum temperature rise of less than 225C on the unexposed side, including joints,
stiffener connections, for the given time period (0, 15 , or 30 min)
7.4.4 Areas of fire resisting divisions
SOLAS describes in Chapter II-2, regulation 26 and 27 the fire integrity requirements of
bulkheads and decks for ships with more and less than 36 passengers, respectively. From the
definitions and tables presented there it is clear that close to all divisions that separate different
areas need to be fire resisting. Even a deck toward open space needs to be A0, i.e. load
bearing, smoke and flame tight for 60 min. Stairways and lifts enclosures are also minimum A0.
7.5 HSC-code requirements
As stated in Section 2, the HSC-code is far less prescriptive than the SOLAS convention
regarding the use of materials. Phrases like the requirements below apply to all crafts
irrespective of construction material, and the use of other fire-restricting materials may be
permitted provided the requirements of this chapter is complied with should prove this
statement.
7.5.1 Fire reaction requirement, and discussion
The HSC-code requires that most of the structure and the fire resisting divisions shall be made of
non-combustible materials or fire-restricting materials. If there is a combustible liner on a non-
combustible structure, it must pass the same surface spread of flame test and smoke and toxicity
test as linings on a SOLAS vessel. A fire-restricting material is a term also used for a
construction where more than the liner is combustible, e.g. the case for all FRP structures. This
must then be tested, in end use condition/as built, in the full scale Room/Corner fire reaction
tests according to ISO9705, and meet the requirements given in IMO resolution MSC.40(64)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 83
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Standard for qualifying marine materials for high-speed craft as fire-restricting materials.
Insulation materials can also be fire-restricting, i.e. they need not be non-combustible.
The HSC-code has no special fire reaction requirement to the adhesive in BONDSHIP type of
bonded joints. In section 7.4.3.2 it is stated, The exposed surface of vapour barriers and
adhesives used in conjunction with insulation materials should have low flame spread. This
shows clearly that the requirement is linked to exposure in a fire. Argumentation used on page
80 (section 7.4.2) can be repeated here.
7.5.2 Fire resistance requirement
7.5.2.1 General
There are no A and B class divisions in the HSC-code. Instead, the terms 30 and 60 min
fire resisting divisions is used. They might be load bearing or not, dependent on the function
during a fire. The shall be tested with the standard (cellulose) fire curve as described in ISO 834,
Part 1 for 30 or 60 min. Fire divisions shall be constructed of approved non-combustible
materials or fire-restricting materials (refer to requirements described in section 5.1). Both 60
and 30 min divisions shall be qualified in full scale test, i.e. minimum 2.44 m x 2.5 m for the
bulkhead test samples, and minimum 2.44 m x 3.04 m for the decks. Details on test specimens
and test procedures can be found in the FTP Code, Part II IMO Res. A.754(18)
Recommendation on fire resistance tests for A, B and F class divisions, with the
addition of IMO Res.MSC.45(65) Test procedures for fire resisting divisions for high-speed
craft
7.5.2.2 60 and 30 min fire resisting divisions
From IMO Res.MSC.45(65) Test procedures for fire resisting divisions for high-speed craft
following extracts have been made:
If the divisions are non-load bearing, they shall be tested as B class divisions in
Res.A.754(18).
If the divisions are load bearing and made of steel and aluminium, they shall be tested as A
class divisions in Res.A.754(18). The temperature rise in the aluminium structural core shall
never exceed 200C.
If the divisions are load bearing and made of fire restriction materials (e.g. a FRP structure +
adequate protection), they shall be tested as B class divisions in Res.A.754(18), but in
addition they should be tested with prescribed load, and for minimum 60 min if the target
classification is a 60 min load bearing fire resisting division. There are criteria for maximum
deflection and deflection rates.
30 min fire resisting divisions shall be tested for minimum 30 min, 60 min divisions for
minimum 60 min. Temperature requirements are:
- Average temperature rise of less than 140C on the unexposed side for the given time
period (30 or 60 min)
- Maximum temperature rise of less than 180C on the unexposed side, including joints,
stiffener connections, for the given time period (30 or 60 min)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 84
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
7.5.2.3 Areas of fire resisting divisions
The Tables 7.4-1 and 7.4-2 in the section 7 of the HSC-Code gives the fire integrity requirements
of bulkheads and decks for passenger crafts and cargo crafts, respectively. Section 7.3 defines
the classification of spaces.
From the definitions and tables presented it is clear that there are no fire resisting requirement for
superstructure side facing open space, unless there is a storeroom with flammable liquid or
dangerous goods or similar then a 60 min fire divisions is needed (if aluminium or fire
restricting material is used in the division).
The bridge on a HSC is a so called control station, i.e. those spaces in which the crafts radio
or navigation equipment or the emergency source of power and emergency switchboard are
located, or where the fire recording or fire control equipment is centralised, or where other
functions essential to the safe operation of the craft such as propulsion control, public address,
stabilization systems etc, are located. Between control stations and areas of minor fire
hazard, such as public spaces, there is a requirement of 30 min fire resisting divisions.
7.6 Fire testing
Full scale testing, as required by both SOLAS and HSC Code for qualification and certification,
were not part of the BONDSHIP project. A large amount of information can be obtained from
relatively small-scale tests, using the same fire load as described for the full-scale tests. In
addition, the joint geometry and bondline thickness should be full scale as is the thickness of
the fire insulation. Using small-scale tests for product development is very cost efficient if a
prototype fails in this test, it will for sure fail in the full-scale test. If details such as panel and
insulation-joints behave well, then experience has shown that they will also perform well in full-
scale fire tests.
7.6.1 Description of furnace
A gas fired horizontal small-scale furnace is used for indicative fire resistance testing. See
Figure 7-1 for illustration.
- Sample size: Approx 800 x 800 mm, dependent on the boundary condition wanted.
- Fire exposed area of specimen: 600 x 600 mm.
- Loading of 1-8 kN is applied to the centre of the specimen, dependent on the test
specimen.
- Maximum 12 channels with thermocouples are available.
- Temperature - time histories are programmable. Standard cellulosic fire curve was used
for all tests in the BONDSHIP programme. This is the curve specified by IMO. See
Figure 7-2 for fire curve.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 85
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 7-1: Small-scale furnace at FiReCo. The picture shows a FRP/sandwich panel
during test.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 86
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
0
200
400
600
800
1000
1200
0 20 40 60 80 100 120
Time [mins]
T
e
m
p
e
r
a
t
u
r
e

[
d
e
g
.
C
]
Hydrocarbon
Cellulosic (BS 476)
Figure 7-2: Standard fire curves for furnace fire testing (BS 476 curve is identical to
ISO 834/IMO Res 754(18)
7.6.2 Description of loading arrangement
The loading can be applied vertically, using a pneumatic cylinder with a maximum capacity of 8
kN (at 10 bar air pressure). This acts as a follower load on the specimen. A sketch of the load
frame is seen in Figure 7-1. The load can be applied anywhere on the sample, and can be
distributed to an area, line or two load points if needed.
7.6.3 Test specimens and fire protection
The fire protection systems used was the same as is commonly used for insulation of steel or
aluminium ship structures (ROCKWOOL density 100 Kg/m
3
and FIREMASTER X607, 96
kg/m
3
). The only difference was that the thickness was increased to achieve a lower temperature
in the joint. Otherwise there were no adhesive bonding specific differences. The insulation
material was fasted using standard wire hooks or weld pins.
For many of the BONDSHIP ACs two tests were performed: The first test was using fire
protection that would have been used if the structure was welded, in order to compare the
structural lifetime of bonded and welded structures, everything else being identical. The
second test used an upgraded fire protection in an attempt to reach the required structural
lifetime.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 87
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
7.7 Application example: aluminium casing bonded to steel deck
7.7.1 General
Figure 7-2 shows the bonded aluminium structure. A relatively thick bondline of flexible
adhesive is used in order to avoid hard spots due to global deformation of the steel hull, but also
to avoid high stresses due to thermal expansion of the aluminium structure in case of solar
heating. It is assumed that the minimum class for the aluminium structure will be A0 which
means 60 min test, max 200C in the aluminium and still integrity in the joint (at least smoke and
flame tight).
Figure 7-3: Aluminium casing bonded to steel deck
7.7.2 Description of fire test specimens
Two test specimens were tested (panel-I and panel-II). The only difference is the fire protection.
The test assembly can be seen of Figure 7-4 and Figure 7-5. The test specimens is designed in
order to introduce shear load in the bonded joint during test, see Figure 7-4. On the right the
shear lag in the adhesive can be seen after load is applied. The adhesive thickness is approx 16
mm. The thermocouple positions are also marked on Figure 7-5.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 88
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 7-4: Test assembly on furnace, with load shear load applied (to the right)
TC8
TC2
TC3
TC6&7
TC9
TC1
TC4&5
TC1
TC2
TC9
TC3
TC8
TC4
TC6
TC7
TC5
Steel side!
Figure 7-5: Fire test specimen with thermocouple positions indicated (TC#)
7.7.2.1 Materials used for test specimens
Adhesive
Type: 1 component polyurethane
Trade name: SIKAFLEX 292
Steel panels
Grade(s) Fe 510
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 89
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Surface Shot-blasting (SA 2.5)
Primer CERABOND by Boat S.p.A.
Two-component Zinc-Silicate shop primer
Qualitative composition: zinc powder & Extenders
Catalyst: ethyl-silicate
Solvent: alcohol
Application: airless spay - brush for retouching
Solvent: CERABOND Thinner
Primer thickness: 15/25 m
Aluminium panels
Plates Alloy 5083
Primer NO
7.7.2.2 Loading and boundary conditions
A line load of 6 kN in total is placed on the aluminium part as close to the joint as possible. The
steel structure is supported close to the joint see Figure 7-4. The load in the joint is close to
pure shear, and the average stress level is 6 kN/(50 x500mm) = 0.24 MPa, which is about 10 %
of the short-term shear strength of the adhesive.
7.7.2.3 Fire protection, panel -I
Fire protection is ROCKWOOL density 100 Kg/m
3
, Thickness. 30mm+ 30mm, tot. 60mm.
Fire insulation is fixed to panels by the means of a sufficient number of wire wall hooks
(harpoons), 23 mm in diameter, welded to the structure, see Figure 7-6 and Figure 7-7.
Figure 7-6: Weld pins for fire insulation mounting
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 90
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 7-7: Insulation of specimen (right), and edge protection of specimen prior to fire test
(left).
7.7.2.4 Fire protection, panel-II
The fire protection of panel-II is identical to panel-I, apart from an added layer of 30 mm
ROCKWOOL density 100 Kg/m
3
locally over the joint, increasing the total thickness from 60 to
90 mm.
7.7.3 Test results panel-I
7.7.3.1 Observation during test
Table 7-1: Observations during first test run
Time (min) Observation Deflection (mm)
0 Test start 0
35 Deflection starts 56 to 82C in the joint 0.1
38 Rapid deflection rate 0.9
39 Rapid deflection rate 3.4
39:10 Joint fails
7.7.3.2 Observations after test
As shown on Figure 7-8, much of the adhesive is still left on the steel side, and Figure 7-9 show
what is left on the aluminium part it is most adhesive left towards the unexposed side of the
specimen.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 91
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 7-8: Sample just after joint failure. The adhesive failed near to the aluminium
panel, but there were still plenty of adhesive left on both the aluminium and steel part.
Figure 7-9: Adhesive left on the aluminium part
7.7.3.3 Temperature measurement
Figure 7-10 shows the temperature readings during the 39 min of test.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 92
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
TEMPERATURE REGISTRATIONS, Specimen FIN AC 6-I
0
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
800
850
900
950
0 5 10 15 20 25 30 35
Time (min)
T
e
m
p
e
r
a
t
u
r
e

(

C
)
T1
T2
T3
T4
T5
T6
T7
T8
T9
T12
Figure 7-10: Temperature readings during test of panel-I.
7.7.3.4 Comments and discussion to the temperature registrations and results:
- Temperatures on the back of the panel (T1 and T2) is far from reaching the limit of a
temperature rise of 140C the temperature rise at 39 min is 49C on average.
- Temperature rise in the core (SOLAS/IMO term) aluminium structure outside the joint
(T9) is far from reaching its limit of 200C the rise at 39 min was 99C.
- The temperature on the steel (T3) and aluminium (T8) flat bar in the joint is high on the
outside, i.e. just under the insulation (178-384C). This might be a local temperature rise
due to many joints in the insulation here.
- The temperature at 39 min in the interface of the adhesive was 64 to 80C (T7 and T6) on
the steel side, and 72 to 161C (T5 and T4) at the aluminium side. The highest
temperatures are on the exposed side. However, T4 showed a rapid increase after 38.5
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 93
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
min, when it was only 93C. This indicates failure, and corresponds well with deflection
measurements. This also explains why there is less adhesive left on the aluminium part,
and almost nothing left on the hottest part of the aluminium (see Figure 7-9)
7.7.4 Test results panel-II
7.7.4.1 Observation during test
Table 7-2: Observations during second test run
Time (min) Observation Deflection (mm)
0 Test start 0
35 From inspection through window in furnace:
It seems to be a gap in a joint in the added
layer of insulation, directly over the joint
0
39 Deflection starts 0.1
43 Rapid deflection rate 1.6
46 Joint fails test stopped 5
7.7.4.2 Observations after test
Identical to panel-I: Much of the adhesive is still left on the steel side, and less on the aluminium
side, where it is most adhesive left towards the unexposed side of the specimen, and nearly zero
on the hottest side.
7.7.4.3 Temperature measurement
Figure 7-11 shows the temperature readings during the 46 min of test.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 94
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
TEMPERATURE REGISTRATIONS, Specimen FIN AC 6-II
0
50
100
150
200
250
300
350
400
450
500
550
600
650
700
750
800
850
900
950
1000
0 5 10 15 20 25 30 35 40 45
Time (min)
T
e
m
p
e
r
a
t
u
r
e

(

C
)
T1
T2
T3
T4
T5
T9
T6
T7
T8
T12
Figure 7-11: Temperature readings during test of panel-II
7.7.4.4 Comments and discussion to the temperature registrations and results:
- Temperatures on the backside of the panel (T1 and T2) is far from reaching the limit of a
temperature rise of 140C the temperature rise at 46 min is 53C in average.
- Temperature rise in the core (SOLAS/IMO term) of the aluminium structure outside
the joint (T9) is far from reaching its limit of 200C the rise at 46 min was 82C (this
point of measurement is behind the added insulation)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 95
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
- The temperature on the aluminium (T8) flat bar in the joint is high on the outside, i.e. just
under the insulation (330C). This might be a local temperature rise due to many joints in
the insulation here.
- The temperature at 46 min in the interface of the adhesive was 61 to 82C (T7 and T6) on
the steel side, and 69 to 83C (T5 and T4) at the aluminium side. The highest
temperatures are on the exposed side. This is somewhat lower than the failure
temperature for panel-I on the aluminium side. However, inspection after test showed
that T4 was not in totally contact with the aluminium, and this has influence on
temperature recording.
7.7.5 Conclusions application example
- The limiting factor for the structural lifetime is the adhesive joint.
- The maximum allowable temperature in the adhesive is approx. 90C if requirement is
structural capacity.
- Possible classification of the structure is A30. The fire insulation on top of the joint needs
to be increased considerably to reach a 60 min load carrying capacity.
- However, the nature of the loads during a fire in this structure (compressive load from own
weight) is such that some mechanical looks can be used instead of the adhesive bond these
bonds can be designed in such way that they do not interact with the steel structure during
operational conditions.
7.8 Overall conclusions from BONDSHIP fire tests
7.8.1 Classification and testing of adhesive joints
For joints that have to carry load during fire, adhesive systems with heat resistance less than 80-
100C is not recommended, as this will require excessive amount of fire insulation: There are no
technical problems connected to fire protection of the bonded joints to any target value, but the
thickness of insulation needed can in some cases be in conflict with volume or area
requirements, and of course weight and cost requirements.
In order to assess the requirements of a bonded joint in a fire, a detailed load analysis has to be
carried out in each case. The key questions are:
- What does happen if the joint fails during fire?
- Does this have any impact on the fire safety of the vessel?
- How can this be avoided by other means than structural strength of the adhesive during fire?
As the current experience with adhesive bonding for ship applications is marginal, it is of vital
importance that the designer works in close contact with the classification society from the very
start of the project.
Flaming or smoke from the adhesive joint was not a problem for any of the tested application
cases. Even if the joints had been unprotected, the fire reaction (heat release, toxicity) properties
of the adhesive did not represent any decrease in fire safety level. This is due to the small areas
of adhesive exposed to fire, but also the small amount used compared to other combustible
materials used on a ship (paint, decorative surfaces, deck covering).
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 96
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The load carrying bonded joints failed at relatively low temperatures, from approx. 40 for the
rigid epoxy (Vantico Araldite 420), to approx. 80 C for the flexible PU-adhesive (Sikaflex 292),
to approx 120C for the semi-rigid methacrylate adhesive (Plexus MA550). None of the
adhesives were selected for optimum temperature resistance, and were not post-cured at elevated
temperatures. Difference from 40 to 120C as temperature limit for the adhesive has major
impact on structural lifetime for a joint during a fire, or major impact on the amount of needed
insulation in order to fulfil a given requirement. In terms of fire resistance, the Plexus MA550
represents a better candidate than the other two adhesives tested. Epoxy adhesive can, on the
other hand, be post-cured and many systems can have service temperature above 120C. The
type of joints, the available heating equipment and the size of the bonded structure will
determine if this is technically and economically feasible.
7.8.2 Joint design
The research in BONDSHIP confirmed that there are some joint design details that can make a
bonded joint more fire resistant:
- Load bearing frame structure: aluminium sandwich panels were bonded to a welded load
bearing frame made out of aluminium box-section members. Conclusion from fire tests was
that the joints that behave satisfactorily in a fire, also with respect to load carrying capacity.
- Door frame or hatch frame bonded to bulkhead: this was an example of a well designed
adhesive joint that decouples the adherends using a flexible adhesive, as well as taking the
local loads in normal operation, but then using a mechanical arrangement (= bolts that are
loose in normal operation) to keep the element in place during a fire and still ensure smoke
and gas tightness.
- aluminium superstructure bonded to steel deck: this was an example of a joint that decouples
steel and aluminium structure in terms of structural interaction. This joint could most likely
benefit from mechanical locks to prevent failure during fire, as the shear forces here are
relatively small and the major load acting is compressive. (see application example on page
87).
- Critical to the joint performance is also were in relation to the temperature distribution the
bondline is positioned and how much ventilation is available. There was one joint in the
BONDSHIP test programme (bonded steel sandwich) where the joint failed long before the
required time. This was because according to measurements taken during the experiment
high temperatures were measured in the most critical part of the bond with little cooling from
the backside structure. For this joint, if it shall be load carrying, a very high temperature
resistant adhesive must be used in order to avoid excessive amount of fire insulation. It is
however not obvious that this joint needs to be load carrying during fire this is dependent
on the supporting structure, the nature of the load, the global interaction etc.
7.9 References
[1] Bjrn Hyning, Final report on fire testing, BONDSHIP Report No.: 1-12-D-2002-02-
0; 2003
[2] Bjrn Hyning, Final design of fire test specimens and fire protection systems ,
BONDSHIP Report No.: 1-14-D-2002-02-0, 2002
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 97
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8 PRODUCTION OF BONDED JOINTS
8.1 Production procedures
Joint fabrication and the required quality control features are governed by the production rate
(joints per hour) and level of mechanisation (according to [1]):
- High level of mechanisation: pre-qualified procedures, on-line measurement and process
control
- Low level of mechanisation: training, inspection, pre-qualified procedures
A quality plan for the bonding operation(s) shall be prepared by the yard/fabricator specifying
working procedures, check points, tests, inspections etc. The following sections specify in detail
the requirements. Inspection and repair are dealt with in chapters 8.5 and 10 respectively.
8.1.1 Joint specification
This information was specified in the design process, see also chapters 2 and 5:
- Base materials
- Geometry with tolerances, e.g. bond width and length, bond line thickness with tolerances
- Documentation/Reporting
8.1.2 Receipt and storage of materials
8.1.2.1 Receipt and handling of materials and consumables
- Adhesives: resins, hardeners, etc.
- Cleaning agents
- Equipment for mechanical surface preparation: grinding, milling etc.
- Agents for chemical surface preparation: etching, deoxidising, etc.
- Verification of product specifications
- Verification of manufacturers or 3
rd
party certificates
- In-house acceptance testing/analyses
- Verification, registration and marking of shelf life
- Marking of cleared materials
- Documentation/Reporting
8.1.2.2 Storage condition of materials
- Range of acceptable storage temperatures
- Range of acceptable storage humidity
- Method of recording of temperature and humidity
- Shelf life
- Requirements to cleanliness
- Requirements to handling of opened packages
- Documentation/Reporting
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 98
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8.1.3 Bonding operation
8.1.3.1 Working environment and handling
- Range of acceptable working temperatures during fabrication
- Range of acceptable working humidity during fabrication
- Method of recording of temperature and humidity
- Requirements to cleanliness or workshop
- Requirements to acclimatisation of materials taken from storage
- Handling of substrates and materials during fabrication
- Documentation/Reporting
8.1.3.2 Surface preparation
It is important to document the whole surface preparation process. Show sequence of the
individual steps and minimum and maximum time intervals and how it should be applied e.g.
certain kind of movement. Health and safety aspects and waste disposal need to be considered.
- Methods and tools for mechanical surface preparation, e.g. grinding, milling (edge
preparation should also be included)
- Methods, tools and agents for chemical preparation, e.g. etching or priming
- Methods and agents for cleaning (note that compressed air may contain oil which can
contaminate the surface again!)
- Documentation/Reporting
8.1.3.3 Application of adhesive
- Verification of shelf-life
- Control of pot-life
- Methods, machines and tools for mixing (needs to be specified for each type of adhesive;
specify conditions e.g. temperature and also cleaning procedures)
- Methods, machines and tools for application; can be manual application or fully mechanised
spatulas of various shapes, pneumatic or electrical handguns or fully robotised systems
- Quantity: weight per surface area, application thickness etc. as relevant.
- Shape and pattern: specify shape of adhesive bead(s) and pattern of application of bead
- Open time?
- Documentation/Reporting
8.1.3.4 Fit-up
- Check of fit-up prior to adhesive application needed?
- Tolerances to fit between substrates, e.g. positioning
- How do you control bondline thickness? (e.g. add spacers, glass beads in adhesive, special
features in adherends such as grooves or ridges)
- Removal of excessive adhesive; specify shape and size of squeeze out and tool to achieve it
if required
- Jig or clamping needed? (specify tools required and sequence of application; important to
maintain target bondline thickness, avoid bondline starvation!)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 99
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
- Documentation/Reporting
8.1.3.5 Curing
- Method of heating (is heating necessary?)
- Cure-cycles: temperatures, heating times, holding times, cooling times
- Clamping: tools used, surface pressure, clamping force, duration etc.
- Method of temperature control
- Humidity curing adhesives: specify minimum storage time before structure can be moved
- Documentation/Reporting
8.1.4 Storage of finished bonds
Storage conditions of finished bonds:
- Structural support
- Range of acceptable working temperatures during storage
- Range of acceptable working humidity during storage
- Method of recording of temperature and humidity
- Requirements to cleanliness
- Documentation/Reporting
8.1.5 Health and safety
Adhesives, primers and solvents are chemical substances that require careful handling. However,
provided precautions are taken they are safe to use. The following list shows the main safety
precautions to be taken (this list has to be modified to adapt to the local requirements):
- Follow safety data sheets provided by the adhesive or paint suppliers
- Follow regulatory requirement regarding maximum allowable concentration of volatile
gasses
- Provide protection to workers for skin, eyes and inhalation
8.2 Manufacturing technology
8.2.1 Introduction
Once the materials for the adherends and the adhesive(s) have been selected (see chapter 3) the
manufacturing process is mostly defined. Again, the table of requirements presented in chapter 2
will answer many of the following questions (and more):
- Hybrid joining with rivets or bolts for example?
- Degree of automation, handheld gun or robot?
- How portable has the process to be (in dry dock or workshop)
- One-off or series; is it economical to make jigs?
- Is it possible to use a dedicated workshop or shipyard conditions?
- Handling of structure how soon can you move the joint?
- Open time, gel time and viscosity of adhesive
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 100
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8.2.2 Dispensing of adhesives
One of the key factors for a successful bonding operation is the application or dispensing of
adhesive on the adherend surface. How, where and how fast the adhesive is applied depends on
the requirements from the application case. The dispensing gun can be handheld (as shown in
Figure 8-3) or it can be a fully automated system as in Figure 8-5. Variable-orifice guns are used
for dispensing adhesives with robots. This ensures that the amount of adhesive being applied
remains constant independent of the angle and velocity of the robot. There are a variety of
application techniques and material deposition methods that can be used in manual or automated
application as shown in Figure 8-1. There are many different nozzle shapes that determine the
shape or size of the adhesive bead.
Figure 8-1 Dispensing of adhesive on surface: (a) extruded beads, (b) droplets, (c) spray;
(from [3])
- Extrude. The adhesive is extruded in one of many bead shapes the most common form of
application for hand tools (Figure 8-1 (a)).
- Flow brush. The adhesive is extruded into the centre of a flow brush. This allows the
adhesive to be spread out or brushed out into seams.
- Droplets. A droplet of adhesive is dispensed simultaneously out of multiple automatic guns
that are mounted in a manifold assembly (Figure 8-1 (b)).
- Spray techniques can be airless, air-assisted or swirl based (Figure 8-1 (c)):
Airless. The adhesive is dispensed at high pressure through a nozzle that is designed to
disperse the adhesive into a specific spray pattern.
(A) (B) (C) (D) (E) (F) (G)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 101
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Air-assisted is similar to airless, except that air is directed at the adhesive stream as it
exits the nozzle to contain the width of the spray pattern.
Swirl. Air impinging on the adhesive stream as it exits the nozzle creates cohesive
adhesive fibres that are rotated into a swirl pattern.
According to De Vries [3] the degree of automation of dispensing systems can be divided into
the following categories:
(i) Manually operated dispensing systems (see section 8.2.3)
(ii) Dedicated dispensing machines (see section 8.2.4)
(iii) Programmable multi axial robots (see section 8.2.4)
8.2.3 Manual adhesive processing
8.2.3.1 Introduction
There are many possible bonding applications in shipbuilding where it is necessary to use a
manual bonding process. This could be because of the high costs to mechanise or robotize a
bonding operation or because it is more convenient to carry the (lightweight) handheld gun
especially during outfitting. A very useful reference for the preparation of this section was
reference [2].
8.2.3.2 Prepare workbench
- Prepare work surface, e.g. disposable cover such as a paper or use PTFE
- Get containers for weighing and mixing adhesive and rod for mixing adhesive
- Get solvents to clean tools
- Provide sufficient paper towels
8.2.3.3 Weighing and mixing of the adhesive
- Select correct adhesive (check date and temperature!)
- Dont mix more than 50g, but at least 5g!
- Resolution of scales (has to be more accurate than the amount being measured)
- Determine mixing ratio either by volume or weight (more common)
- Dispose properly of un-used adhesive
8.2.3.4 Apply adhesive
- Cover area which shall not receive adhesive with tape
- Use spatula to apply adhesive (or gun)
- Tooth comb to spread adhesive if needed
- Bondline thickness control (spacers, e.g. glass beads)
- Clamping? (use weights or C-clamps)
- Remove excess adhesive (squeeze out)
- Check open time of adhesive
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 102
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8.2.3.5 Automatic mixing and dispensing systems
Automatic mixing and dispensing systems consist of a metering machine and a hand held
dispensing gun. Two examples are shown in Figure 8-2. They require low capital investment but
are only suitable for small scale production.
Figure 8-2 Commercial metering system (DOPAG)
It is very common to use hand held guns with singe or dual cartridges. This ensures that two
component adhesives are always mixed at the same ratio. The handheld devices have usually
static mixers as shown in Figure 8-3. The mixing head is connected to a dualbarrel cartridge.
Static mixers are good for adhesives which have a similar viscosity ratios (e.g. 1:1, but not
suitable for larger than 1:10) like e.g. epoxy resins. The mixing tube is discarded after use.
Figure 8-3 Handheld dispensing gun with disposable static mixer (DOPAG)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 103
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8.2.4 Mechanised adhesive processing
Reference [2] has been used extensively for the preparation of this section.
8.2.4.1 Basic setup of a mixing and dispensing unit
Dedicated mixing and dispensing machines are purpose built for a specific application which
they perform fully automatic. However it is not possible to adapt them to new task as they need
to be modified or redesigned. The most flexible dispensing system is the programmable multi
axial robot. It allows to perform complex dispensing applications. In case the geometry of the
component changes, the robot can be reprogrammed. A typical application is shown in Figure
8-5. A system is usually made up of a number of standard components:
Storage container, e.g. barrel. For simple hand held devices the resin components come in small
cartridges. For small quantities the two components are usually supplied in a dualbarrel
cartridge which ensures that the mixing ratio is kept constant. For larger systems resin
components are stored in barrels which are much more economical than cartridges.
Pumps are used to deliver the adhesive to the mixing and dispensing unit. Valves are used to
regulate the adhesive volume or pressure. The hoses and pipes used in the unit can be heated.
The items mentioned in this paragraph will be discussed in the following chapters.
Figure 8-4 Automated application of adhesive to wind screens, [5]
8.2.4.2 Pumps suitable for adhesive delivery
There are many different types of pumps that are used for adhesive delivery. They are usually
distinguished according to the following criteria:
- What is the useful viscosity range?
- How much pressure does it generate?
- How high an additive content can it deliver (also permissible particle size)?
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 104
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
- Constant or fluctuating supply of adhesive?
- Is it possible to heat the pump?
Using the above criteria it is possible to select suitable pumps for the job at hand.
An example of a common type of pump is the variable ratio or slave arm metering where the two
pumps are locked together by a lever. This ensures that the ratio remains constant for each pump
stroke. It is possible to change the ratio of the components by shifting the position of the pivot
arm. A typical machine is shown in Figure 8-5. For low-viscosity adhesive (typically less than
200 mPa s, [3]) it is also possible to use gear pumps which deliver the adhesive continuously.
Here the mixing ratio is controlled by selecting different gear ratios or by electronically
controlling the motors.
Figure 8-5 Variable ratio metering and mixing machine, [4]
8.2.4.3 Valves
Again, there are many different types of valve to choose from, the needle valve is one example.
The following list provides suitable questions for the selections of valves:
- What is the viscosity range?
- Operating pressure?
- Is the valve heatable?
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 105
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
- Is adhesive displaced when the valve is closed? (important for accurate dispensing of
adhesives)
- Does it work with adhesives that contain particles or fillers?
8.2.4.4 Mixer
The mixing usually occurs at the (handheld) gun which is used to apply the adhesive. The
adhesive components are mixed just before the nozzle (see Figure 8-3) so that only very few
parts of the gun need to be cleaned after use. To mix the resin components either static or
dynamic mixers are used [4]. The handheld devices have usually static mixers as shown in
Figure 8-3. The mixing tube is thrown away after use. It is important that the flow of adhesive is
adjusted such that at least 3 times the volume of the mixer passes through the mixer tube within
the duration of the potlife of he adhesive [2].
There are also dynamic mixers which portable if they are driven by compressed air. Dynamic
mixers have a cavity or chamber with a rotating paddle. To achieve thorough mixing it is
important that the adhesive stays at least one second in the mixing chamber [2]. It needs to be
cleaned with solvents after use. An alternative way to clean the mixer has been developed in the
aerospace industry [4]. The mixing unit is stored in a deep freezer after use (at about 40
o
C). The
low temperature inhibits cure of the adhesive. The next time the mixing unit is used, it is thawed
and the mixing chamber is flushed out with freshly mixed adhesive. This method works both for
static and dynamic mixers [4].
8.2.4.5 Ancillary equipment
The adhesive needs to be delivered from the barrel to the dispensing gun where it is mixed prior
to application. This is done by the header system [3]. The header system is usually composed of
rigid pipes and flexible hoses with a length of up to 90 m. For humidity curing adhesives it is
critical to use hoses with PTFE liner as only PTFE prevents diffusion of water. If not the
adhesive could start to cure prematurely and block the header system equipment.
The header systems will also contain a considerable number of valves. These valves regulate the
flow of adhesive in the system. Of all the process parameters, viscosity of the adhesive is the
most important one as it determines the required pumping pressure and hence the dimensions of
the components pipes, hoses e.t.c. have to be designed accordingly. For adhesives with high
viscosity it is necessary to heat the header system to reduce the flow resistance.
The main benefit from heating is that it can keep viscosity independent of fluctuations in room
temperature (e.g. heat to 35
o
C). Moreover, the adhesive becomes easier to pump and dispense
and the curing reaction is accelerated.
8.2.5 Cure of adhesive
Most adhesive relevant for shipbuilding cure at room temperature. To achieve consistent quality
it is important to control the work environment as much as possible. Key parameters are:
temperature, time and humidity. Furthermore, it is essential that the structure is not moved or
loaded while the adhesive cures. These parameters vary from adhesive to adhesive and they must
therefore be specified clearly in the production procedure.
It was already discussed in section 7.8 that using heat to cure the adhesives could increase the
fire performance of a bonded joint. A typical temperature range for epoxies is between 80C to
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 106
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
120C. If it is technically feasible (for larger structure probably not) one could obtain additional
benefits by heat curing. It was shown by [6] that the cooling rate influences the joint strength.
The slower the cooling down rate the higher was the ultimate shear strength of the samples. [6]
also observed an influence of ageing on the shear strength. Samples which where cooled very
fast after cure and stored in a dry environment for 30 days had an ultimate strength which was
approaching the results for slow cool down.
8.3 Application example: Production procedures for aluminium
superstructure
The following case has been adapted from references [7] and [8].
8.3.1 Casing design
The casing under consideration has the following main geometrical dimensions:
Transverse extension (width) 21600 mm
Longitudinal extension 7700 mm
External base perimeter 57752 mm
Total base perimeter 94736 mm
Frontal area 64.810
6
mm
2
Lateral area 23.110
6
mm
2
The casing has a symmetry plane coincident to the ship longitudinal symmetry plane.
Considering half casing, there are two internal longitudinal bulkheads, placed respectively at
1000 mm and 6600 mm from the ship centreline, and one transversal bulkhead, which extends
from the centreline to the external wale. The ceiling and the longitudinal bulkheads have two
frames, and the ceiling is longitudinally stiffened. The scantlings of the aluminium casing has
been calculated according to RINA rules, which prescribes to use a design pressure of 3 kPa.
Using this pressure the following scantlings have been calculated:
Thickness of aluminium plating 6 mm.
Ceiling longitudinal stiffeners bulbs 100x6, 700mm spacing
Ceiling transversal frames 318x100x4x8
Longitudinal bulkhead stiffeners bulbs 120x6, 917mm spacing
Longitudinal bulkhead frames 156x80x4x6
Transversal bulkhead stiffeners bulbs 120x6, 700mm spacing
The aluminium alloy scantlings obtained for the casing ceiling are shown in Figure 8-6.
8.3.1.1 Bonded joint design
After considering different possible joint configurations, the one shown in Figure 8-8 seems to
be the most simple and effective. In order to completely define the joint, proper values for the
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 107
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
overlap length and the adhesive thickness need to be defined. In Figure 8-8 is shown a 2D view
of the joint, while in Figure 8-7 a 3D view.
The main dimensions of the joint has been defined by simple calculations, on the basis of the
thermal load and the global ship deformation, assuring that the maximum allowable stresses and
strains previously set for the adhesive are not exceeded.
Figure 8-6: Drawing of half aluminium casing ceiling
Figure 8-7: Bonded joint lay-out: 3D view
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 108
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 8-8: Bonded joint scantlings
8.3.2 Material information
The following materials were used for the application example:
STEEL
Grade(s) Fe430 / Fe 510
Surface Shot-blasting (SA 2.5)
Primer CERABOND by Boat S.p.A. (see Annex 2)
Two-component Zinc-Silicate shop primer
Qualitative composition: zinc powder & Extenders
Catalyst: ethyl-silicate
Solvent: alcohols
Application: airless spay - brush for retouching
Solvent: CERABOND Thinner
Primer thickness: 15/25 m
ALUMINIUM
Plates Alloy 5083 or 5383 - H321 or H111 Skin passed
Channel/profiles Alloy 6082 - F28 (T5 - 6) Extruded or alloy 6005
Primer Epoxy primer - only for external side surfaces.
Primer is applied only after completing structures.
Primer thickness: 50 m
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 109
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
8.3.3 Surface preparation
8.3.3.1 Steel surfaces
As previously reported, steel surface is pre-treated by CERABOND primer. Depending on the
state of the surfaces, the following points/actions are recommended.
1. The surfaces temperature must not be less than 10 C. It is also recommended to the
operator: to use protective gloves and glasses, to ventilate the workshop, if the priming is
done in enclosed space, and at last to have no heat source close to the priming surfaces.
2. If surfaces are very dust and/or with rust deposits they must be cleaned by sand paper grade
40/80 and the use of a vacuum cleaner. The sharp edges of the surface to be bonded must be
rounded off to guarantee minimal thickness and good coverage of anti-corrosive primer, thus
avoiding the rusting of the steel surfaces and in consequence the efficiency loss of the joint.
3. Cleaners ( i.e. SIKA Cleaner 205) must be used on the surfaces. Application must be done
by a white linen cloth or, better still, by a paper towel. Apply the cleaner sparingly to the
cloth or towel, then wipe off the surface with a series of straight, light strokes: rotary
polishing actions are not recommended. Turn the cloth or towel often. After the applications,
the cleaner should leave neither film nor residuals, any remaining excess must be wiped up.
Treat the bond faces only. If cleaner is accidentally deposited in the surrounding area, it
must be wiped of immediately.
4. Ten minutes after the application of the SIKA Cleaner 205, surfaces must be treated by the
epoxy-anticorrosive primer SIKA - ICOSIT - EG1. Application could be done by mean of
brush or paint roller, which must be clean and dry.
5. Twelve hours later the surfaces are suitable to be bonded. Before the bonding procedure the
surfaces must be re-cleaned fast by SIKA Cleaner 205.
6. A good approach is to apply the primer in the evening and to bond the day after in the
morning.
7. The bonding should be done not later than four weeks after the priming of the surfaces.
8.3.3.2 Aluminium surfaces
As previously reported, aluminium surface is treated by Epoxy primer.
1. The surfaces temperature must not be less than 10 C. It is also recommended to the
operator: to use protective gloves and glasses, to ventilate the workshop, if the priming is
done in enclosed space, and at last to have no heat source close to the priming surfaces.
2. If surfaces are very dust and/or with rust deposits they must be cleaned by sand paper grade
40/80 and the use of a vacuum cleaner. The sharp edges of the surface to be bonded must be
rounded off to guarantee minimal thickness and good coverage of anti-corrosive primer, thus
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 110
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
avoiding the rusting of the steel surfaces and in consequence the efficiency loss of the joint.
3. Cleaners (i.e. SIKA Cleaner 205) must be used on the surfaces. Application must be done
following the same procedures and recommendations made for the steel case.
4. Ten minutes after the application of the SIKA Cleaner 205, surfaces must be treated by the
primer SIKA 204 EP. Application could be done by mean of brush or paint roller, which
must be clean and dry.
5. Twenty-four hours later the surfaces are suitable for bonding. Before the bonding procedure
the surfaces must be re-cleaned fast by SIKA Cleaner 205.
6. The bonding should be done not later than four weeks after the priming of the surfaces.
8.3.4 Bonding procedures
8.3.4.1 Selection of bonding procedure
The "curing" time of SIKA flexible adhesives depends on the environmental humidity and
temperature. Generally, in standard environmental conditions, the curing process begins after 4
or 5 minutes after the applications, starting from the external exposed surface toward the
internal regions, giving rise after 20-30 minutes (depending on the adhesive type) to a skin
formation which will impair the wetting properties of the adhesive and consequently will render
impossible to obtain adhesion on the substrate.
Since the bonded joint of casing has a very large extension, it is not possible to apply the
adhesive in the entire area of joint before placing the casing in its position. In fact, the time
estimated to apply adhesive and to place the casing in the right position is greater than the open
time of the adhesive, i.e. 20 minutes. According to the experts of SIKA, it was decided to adopt
an injection bonding procedure. The casing will be placed on its place, (proper spacers will be
placed to ensure a suitable adhesive layer), and than the adhesive will be injected inside the joint.
In this way the open time of the adhesive will not be a critical issue.
8.3.4.2 Selection of adhesive
After selecting the injection procedure, the experts also indicate that the joint has a quite large
width and this means a long time for curing, if "normal" adhesive is used.
In fact as said before, using normal adhesive, the curing process starts on the external surfaces
and proceeds towards the internal parts, Figure 8-9. After the external sides are cured, they
become sealants for the internal uncured adhesive. So, the time for a complete curing may be too
long, and not compatible with the other needs in the shipyard.
For this reason a two-component adhesive, not cured by humidity, has been selected: Sikaflex -
254 Booster. In this case the curing reaction does not depend on the environmental conditions
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 111
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
but it occurs within the whole adhesive volume. The final strength will be reached in approx. 1-2
weeks. The adhesive will have after 4 h more than 50 % and after 24 h more than 70 % of the
final strength.
Figure 8-9: Adhesive curing process without "booster"
8.3.4.3 Procedure for "spacers" selection
We will use approx. 40 spacers distributed all around the bond. The spacers are made of
Sikaflex-254, with a thickness of 17 mm and a surface of 20 x 50 mm2. So the compression
stress in the spacers will be approx. 0.2 MPa. [= 8500 N / (40 x 20 x 50 mm2)]. This will press
the spacer down to approx. 16 mm [= 17 mm * (1 - 0.2 / 5)]. 5 MPa is the compression stiffness
of the Sikaflex-254. For assembly the spacers can be fixed with SIKAFLEX.
8.3.4.4 Description of bonding equipment
The Sika Booster pump is based on the idea of using an existing one-component adhesive pump
extended with an add-on Kit (V-Kit) for supplying, dosing and mixing Sika Booster-Paste to
Sikaflex -254 (see Figure 8-10). The base-pump may come from Reinhardt Technik, Graco or
other qualified pump manufacturers. However, not all pump-types from these manufacturers are
suitable. Certain components, e.g. the pump motor, must meet specific requirements to suit the
processing of Sikaflex -254.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 112
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 8-10: one component adhesive pump with an add-on Kit for supplying, dosing and
mixing Sika Booster-Paste to Sikaflex -254
8.3.4.5 "Due point"
The recommended application temperature is in the range of 10C to 30C. The storage of the
pump equipment at room temperature before use is helpful.
8.3.4.6 Bonding operation
Proceed to the following operations:
1. Secure a stopper-band around all perimeter of the joint, at one end. This band must have a
section higher than the thickness of the joint , must be compressible and the adhesive must
not have adhesion on it. Optimal choice is closed cells PE foam normally used as back-
filling joints in construction (i.e. Sika Ethafoam) Figure 8-11A
2. Mask the perimeter of the bonding area on the opposite end with adequate masking tape
Figure 8-11A
3. Inject Sikaflex-254 Booster into the prepared joint slot, subsequently compressing the
adhesive with a spatula to eliminate air entrapment, to ensure full contact with the sides of
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 113
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
the joint and to leave a cavity which will be then filled with the refining sealant Figure
8-11B and C. Immediately remove masking tape from surface.
4. Once the adhesive is cured, remove the stopper band.
Stopper band
Masking tape
A
Adhesive
B
Spatula
C
Figure 8-11: Filling of the joint
8.3.4.7 Final surface refinement
After bonding operations, the following procedure could be adopted in order to obtain a surface
refinement:
1. Adhesive tape is placed over and below the adhesive layer Figure 8-12A.
2. Fill the empty space with Sikaflex 292 (due to poor smoothing properties of Sikaflex-254
Booster, a different material has to be selected to obtain a smooth finish) Figure 8-12B.
3. To facilitate smoothing operations, spray SIKA-Tooling-N on the surface of the adhesive.
Remove the adhesive surplus smoothing by mean of a broad knife - Figure 8-12C.
4. After smoothing, adhesive stripes must be removed in maximum 15 minutes, before
Sikaflex 292 is cured.
Masking tape
A
Sealant Sealant
B
Spatula Spatula
C
Figure 8-12: Final surface refinement
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 114
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Please note that if only on one side of the joint a smoothed surface is a must, it would be possible
to follow a less time-consuming procedure for bonding and refining:
1. Secure the stopper band on the side of joint which lately will have to be smoothed as in
Figure 8-11A.
2. Gun Sikaflex-254 Booster into the joint, compressing and removing excess of adhesive with
a broad knife without leaving any cavity in the joint.
3. After Sikaflex-254 Booster is cured, remove stopper band and proceed as per above reported
refining procedure only on this side of the assembly.
8.4 Application example: Production procedures for bonding of lashing
devices
The following case was extracted from reference [9].
8.4.1 Lashing device in mock-up
Figure 8-13: Set-up bonding AC5 mock-up components
The following adhesive parameters were used for AC5:
Table 8-1: Main characteristic features to work out a list of requirements
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 115
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 8-2: Steps of production procedure AC5 Vantico Araldite 420 mock-up
8.4.2 Lashing device samples for laboratory testing
Additional lashing devices were tested at IFAM in Bremen. Therefore the same lashing devices
as in mock-up are bonded to a pick-up which can be clamped into a test machine at IFAM.
Figure 8-14: Set-up bonding AC5 IFAM components
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 116
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Table 8-3: Bonding parameter AC5 IFAM test specimen
All eight pieces of adherend 1 and eight pieces of adherend 2 are coated with shop-primer and
bonded at Meyer-Werft by IFAM. For testing at IFAM four out of eight specimens were bonded
on a shop primed surface, while the remaining four specimens were glued on a grinded surface.
Table 8-4: Steps of production procedure AC5 Vantico Araldite 420 IFAM test specimen
8.5 List of references
[1] Espie et al. Quality Assurance in Adhesive Technology, Woodhead, 1998
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 117
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
[2] Course notes European Adhesive Engineer (week 5), IFAM, Bremen, 2002 (in
German)
[3] J. E. De Vries, Dispensing and Application Equipment for Adhesives and Sealants,
in Adhesives and Sealants, vol. 3, Engineered Materials Handbook, H. F. Brinson, Ed.,
1 ed: ASM International, 1990, pp. 693-702.
[4] W. Devlin, Metering and Mixing Equipment, in Adhesives and Sealants, vol. 3,
Engineered Materials Handbook, H. F. Brinson, Ed., 1 ed: ASM International, 1990,
pp. 687-692.
[5] H. Turner, Automation and Robotics for Adhesives and Sealant Use, in Adhesives
and Sealants, vol. 3, Engineered Materials Handbook, H. F. Brinson, Ed., 1 ed: ASM
International, 1990, pp. 716-725.
[6] W. P. De Wilde, G. Van Vinckenroy, L. Tirry, and A. H. Cardon, Effects of the
environment and curing on the strength of adhesive joints, Journal of Adhesion
Science and Technology, vol. 9, pp. 149-158, 1995.
[7] Luca Demattei, Final production procedures for Fincantieri AC6 and AC1,
BONDSHIP Report Number: 4-42-D-2002-01-0; 2002
[8] Leopoldo Ungarelli, Preliminary design of prototypes, BONDSHIP Report Number: 4-
41-D-2001-01-0; 2001
[9] Gnter Preuth, Final production procedures - JLM, BONDSHIP Report Number: 5-52-
D-2002-01-0; 2002
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 118
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
9 REPAIR OF BONDED JOINTS
9.1 Repair procedures
The need for repair can occur during new building to correct mistakes or in-service as a result of
wear and tear or accident damage. Furthermore modifications or upgrades may require the use of
a repair method. Hence a repair method(s) should cover both the production phase and the in-
service phase:
- The same items as described in section 8.1.3 apply here as well
- Additional required information includes the method of removal of the old adhesive (e.g.
cutting wire or vibration cutter for Polyurethanes) and perhaps how to remove of part of the
structure if it is damaged.
- It is important to specify an appropriate adhesives: in some cases one can bond onto the
existing adhesive (e.g. PU) while other adhesives require complete removal of the old
bondline and proper surface preparation (e.g. epoxy)
- In addition it may be necessary to design some kind of patch to effect the repair
- Documentation/Reporting
9.2 Application example: repair of aluminium superstructure
This section is an edited version of reference [1].
9.2.1 Background
The mock-up is constituted of: An aluminium deckhouse (Alloy 5083) of dimension 5504mm x
2100mm bonded on a steel deck portion (Fe 510 / primed with Cerabond) (Figure 9-1 and Figure
9-2).
Figure 9-1: Bonded joint configuration Figure 9-2: Casing mock up - Outside View
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 119
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
9.2.2 Damages
The joint was damaged in two different ways:
1) The adhesive was removed from the bonded line; the bond line was drilled and a steel wire
was get inside, the cutting was made by one man inside the deckhouse and one man outside
(Figure 9-3). It was decided to remove completely two portion of adhesive of length about 30
centimetres in order to test two different restoring methodologies (Figure 9-4). This kind of
damage (removing the total part of the adhesive) is well representative of the case in which is
requested to remove and replace a panel portion of the casing.
Figure 9-3: Removing of the adhesive
Figure 9-4: Bonded line to be replaced
2) A portion of the bonded line, about 1 meter, has been subjected to the heat of an
oxyacetylene torch (Figure 9-5). This kind of damage is well representative of an
accidentally damage due to passage of a welder or an oxyacetylene torch on the bonded line.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 120
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 9-5: Carbonisation of a portion of bonded line
9.2.3 Repairs
The bonded line was damaged in three different points, the mock up before the repair was in the
below condition (Figure 9-6).
Figure 9-6: Bonded line before the repair
1. A portion of joint which has been mechanically damaged can be completely restored in two
different way:
- To use Sikaflex-292 Booster and follow the same operations such as done during the
manufacturing of the joint:
I. Clean the surfaces to be re-bonded with Sika Activator with a paper towel
(assumption: in the case that an aluminium portion have to be replaced, it must be
already primed with Sika 204 EP) activate also adjacent parts of cured polyurethane
adhesives..
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 121
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
II. Secure a stopper-band at one side of the joint to be repaired. This band must have a
section higher than the thickness of the joint, must be compressible and the adhesive
must not have adhesion on it. Optimal choice is closed cells PE foam normally used as
back-filling joints in construction (i.e. Sika Ethafoam).
III. Mask the perimeter of the bonding area on the opposite side with adequate masking
tape(see document Production Procedures for Fincantieri AC6 and AC1).
Fill Sikaflex-292 Booster into the prepared joint slot, if the joint to repair have not a large
extension a normal gun can be used and a Booster System pump is not needed (
IV. Figure 9-7). Subsequently compress the adhesive with a spatula to eliminate air
entrapment, to ensure full contact with the sides of the joint and to leave a cavity
which will be then filled with the refining sealant. Immediately remove masking tape
from surface.
V. Once the adhesive is cured, remove the stopper band and proceed with final sealing
with Sikaflex 292 on both sides of the joint.
Figure 9-7: Filling of Sikaflex 292 Booster
- To use Sikaflex-292 (without booster) and follow the following operations:
I. Clean the surfaces to be re-bonded with Sika Activator (assumption: aluminium
portion to be replaced has already been primed with Sika Primer 204 EP) activate
also adjacent parts of cured polyurethane adhesives.
II. Position a stopper-band (the same described above) in the middle of the joint to be
repaired (Figure 9-8 - Figure 9-10.A). This results in two bonded joints, one on the
inside and one on the outside. These joints are sufficiently narrow to allow the use of
standard moisture curing polyurethanes. The stopper-band may provoke leaks and
corrosion, so the best way is to remove it before filling from the other side, may be
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 122
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
after one day.
Figure 9-8: stopper-band at the middle of the joint
III. Mask the perimeter of the bonding area to be repaired on the external side with an
adequate masking tape, its high should be greater than the spatula one (Figure 9-10.A).
IV. Gun Sikaflex-292 into the prepared joint slots (internal and external side) (Figure 9-9 -
Figure 9-10.B), subsequently compressing the adhesive with a spatula to eliminate air
entrapment and to ensure full contact with the sides of the joint. Immediately remove
masking tape from surface.
Figure 9-9: Filling of Sikaflex 292
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 123
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Stopper band
Masking tape
A
Adhesive
B
Spatula
Spatula
C
Figure 9-10: Repairing of the joint with Sikaflex 292
For each of the two reparations made on the mock up the final surface refinement using the
masking tape was not made (due to imperfection of steel and aluminium surfaces also during the
manufacturing of the mock up no surface refinement was made) the complete restoring of the
joint (about 30 centimetres), in each case, needed about 1 hour, with one man working.
2. The portion of bonded line (about 1 meter), subjected to the heat of a oxyacetylene torch,
needed only a superficial repair, even though the adhesive caught fire, see Figure 9-11.
Figure 9-11: Adhesive burning
The bonded joint can be repaired with the following operations:
I. The carbonized adhesive must be removed by a cutter (Figure 9-12).
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 124
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 9-12: Removing of the carbonised
adhesive
Figure 9-13: Filling of Sikaflex 292
II. Mask the perimeter of the bonding area to be repaired with an adequate masking tape
(Figure 9-14.A).
III. SIKA Activator (bond promoters, surface activator) must be used on the old
adhesive. Application must be done by a paper towel.
IV. Inject Sikaflex-292 into the prepared joint slots (Figure 9-13 - Figure 9-14.B),
subsequently compressing the adhesive with a spatula to eliminate air entrapment and
to ensure full contact with old adhesive. Immediately remove masking tape from
surface.
Adhesive
Masking tape
A
new Sikaflex 292
old Adhesive
B
Figure 9-14: Superficially Repair of the joint with Sikaflex 292
During the repair made on the mock up the masking tape was not used, the complete repair of the
joint (about 1 meter) needed about 1 hour and half, one man working.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 125
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
9.3 List of references
[1] Luca Demattei, Final repair guidelines (completed repair of prototype), BONDSHIP
Report Number: 4-43-D-2002-01-0; 2003
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 126
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
10 NON-DESTRUCTIVE INSPECTION
This chapter has been adapted from [1].
10.1 Overview
The aim of this chapter is to document the application of non destructive testing (NDT) to
support the design, manufacture and analysis of adhesive joint assemblies in the BONDSHIP
programme. As the research was carried out consideration was given to standards, certification
and guidelines and their impact on the testing work for the BONDSHIP application cases.
This chapter serves as an introduction to bonding and the role of NDT, and the reader is referred
to the literature [8] and online resources (www.ndt.net, www.bindt.org) for more detailed
insights into NDT technology.
A review of currently available relevant NDT techniques and an analysis of the capability of
each technique for adhesive bond inspection is provided in section 10.2.
Section 10.3 provides a summary of the requirements to be considered when employing NDT
for adhesive bond inspection.
As ultrasonics is the main test technique for adhesive bonding this project naturally
concentrates on this technology for adhesive bond inspection. To aid the reader an overview
of ultrasonic theory and an introduction to the inspection techniques is provided in section
10.4.
Section 10.5 covers the key results and research activity for the use of NDT techniques
within the BONDSHIP programme. This was not limited to inspection of the manufacturers
application cases and hence NDT was involved in the materials selection process, joint
design and mechanical testing programme.
Section 11.4 of the appendix contains a glossary of the most common NDT terms, lists
current standards and certification requirements and shows a sample NDT procedure.
10.2 Adhesive bonding and NDT
10.2.1 Definition of bonding parameters
A typical adhesive joint topology is given in the following figure:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 127
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-1: Basic adhesive joint topology.
This topology can be extended to include laminates (layers of bonded material to form
composites) and multilayer joints. Assuming good adhesion the important parameters for the
performance of the joint are the bonding area (bead width) and the bondline thickness.
10.2.2 Definition of defect parameters
The following figure shows typical defects and their location in an adhesively bonded joint.
Figure 10-2: Typical defect parameters and location in joint.
In general, defects can be classed into two groups:
- Interfacial
Delamination or disbonding caused by incorrect application of adhesive or loading of the
joint during service.
Kissing bond a term describing an adhesive joint with little or no strength (shear
stiffness). Although there is no standard definition it is widely accepted to be an
interfacial phenomenon.
- Bulk
Adherent
Adhesive
Adherent
Coating (paints etc)
Bead width Bondline thickness
Adherent
Cohesive failure Adhesive failure/delamination
Adhesive
Adherent
Porosity, voiding
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 128
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Adhesive failure cracking in the bulk of the adhesive, for example, caused by loading
of adhesive during service.
Voids large pores or bubbles trapped in the bulk of the adhesive, often caused by
trapping of air during joint production.
Porosity small pores or bubbles of air trapped in the bulk of adhesive. The air is often
present in the adhesive prior to application and is released during joint production and
adhesive cure (out gassing).
Further details of adhesive joining chemistry and engineering can be found in the literature ([7],
[3]). The following sections cover the available test technologies that could be employed to non
destructively inspect adhesive joints.
10.2.3 Description of available test technologies
There are a wide range of NDT technologies available which are often application specific
depending on the materials being bonded. Even within the main test method of ultrasonics there
are several different techniques in routine use. The aim of this section is to give the reader an
appreciation of the technologies that could be employed. A summary of the test capabilities are
given below (taken from [2]) and a brief summary of the applicability to bond testing is
provided. A list of equipment manufacturers for the listed techniques is given on page 210
(section 11.4.4).
10.2.3.1 Acoustic methods
Acoustic methods (also known as Sonic testing, vibration testing, mechanical impedance testing)
depend on exciting vibrations in a specimen by a local impact and then measuring some
properties of the vibrations, e.g. resonant frequency, decay time, etc. The commonest method is
the modern equivalent of the railway wheel tapper, using a controlled-energy impact instead of a
hammer, and a transducer detector instead of ears. Electronic or digital data analysis and
recording may also be used. Hard objects will have natural resonances and these are dependent
upon the microstructure, hardness and presence of defects. Changes in any of these
dependencies will cause changes in resonant frequencies, and the components or areas of
components having differences can be detected. Resonant inspection can be used to detect
cracks, voids, delaminations, lack of bonding and changes in hardness in a wide range of
metallic and composite bonded joints. The technology cannot detect or identify specific defects,
but can detect changes in components relative to standard references.
10.2.3.2 Radiography
Radiography uses X-rays or gamma-rays to produce an image of an object on film. The image is
usually natural size. X-rays and gamma-rays are very short wavelength electromagnetic
radiation which can pass through solid material, being partly absorbed during transmission.
Thus, if an X-ray source is placed on one side of a specimen and a photographic film on the
other side, an image is obtained on the film of the thickness variations or other inhomogeneities
in the specimen, whether these are surface of internal. This is a well established technique which
gives a permanent record and is widely used to detect internal flaws in weldments and castings
and to check for mis-constructions in assemblies. Due to the relatively small change in
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 129
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
absorption contrast achievable in bonded joints the application of Radiography to bond
inspection is limited to specific applications.
10.2.3.3 Shearography
Laser shearography uses an interferometer to detect the out-of-plane derivative of deformation of
the test piece due to the non-homogeneous strain field caused by subsurface flaws. Creating an
out-of-plane stress by applying vacuum, vibration of thermal stress, plays upon the weakest bond
of a laminar or bonded structure. The combination of correct stressing and shearography
provides a rapid, non-destructive evaluation of composites, honeycombs and sandwich
constructions both in production and in service, to qualify repairs and impact damage.
Production shearography equipment is based around the deployment of vacuum test chambers.
High definition shearography systems are able to discern single honeycomb cell failures in a
metre square panel in seconds. The systems are configured for numerous applications and the
incorporation of shearography into production is proving to reduce inspection time drastically.
This technology is generally applied to the inspection of thin sandwich structures and composite
honeycomb panels.
10.2.3.4 Thermography
Thermography is a technique of obtaining an image of the heat distribution over the surface of an
object. The usual method is to use a special television camera with an infra-red sensitive
detector and a lens which transmits infra-red radiation. Such cameras can operate at normal
video rates. Temperature variations in the test sample are then displayed as shades of grey or
can be converted into pseudo-colour images and temperature variations as small as 0.2
o
C can be
detected.
The method of application for bonded joints is to provide a pulsed source of heat to the test
sample and examine the surface for non-uniformities in infra-red emission which could
correspond to internal inhomogeneities or large flaws. This technique is generally limited to the
inspection of plastic composites for the detection of near surface (within 10 mm) defects.
10.2.3.5 Ultrasonics
Ultrasonic methods of NDT use beams of mechanical waves (vibrations) of short wavelength
and high frequency, transmitted from a small probe and detected by the same or other probes.
The frequency is in the range 0.1 to 20 MHz and the wavelength is in the range of 1 to 10 mm.
The velocity depends on the material and is in the range 1000 to 6000 m/s.
Usually pulsed beams of ultrasound are used and in the simplest instruments a single probe,
hand-held, is placed on the specimen surface. An oscilloscope display with a time based shows
the time that it takes for an ultrasonic pulse to travel to a reflector (a flaw, the back surface, or
other free surface) in terms of distance across the oscilloscope screen the so-called A-scan
display. The height of the reflected pulse is related to the flaw size and the position to the flaw
depth. The relationships of flaw size, flaw distance and flaw reflectivity are complex, and
considerable skill is required to interpret the display. Ultrasound and sonic techniques have wide
application for the inspection of numerous types of bonded systems and is the most common test
method employed.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 130
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
10.2.3.6 Visual
Visual inspection, with or without optical aids is the original method of NDT. Many defects
which are surface breaking or hidden in inaccessible areas (bonded structures) can be detected by
careful direct visual inspection. Optical aids include low-power magnifiers, microscopes and
specialised devices such as boroscopes, endoscopes, and other fibre-optic devices for the
inspection of restricted access areas. These devices can also be used with television camera
systems. Much of the success of visual inspection depends on access to the test subject and the
lighting arrangements. A typical application of visual NDT in bonding is for the detection of
spewing or squeezing out of the adhesive during bond assembly indicating that good coverage
and a continuous bond has been obtained.
The following table summarises the capability of the above techniques to aid selection for a
particular bonded joint type.
Table 10-1: Guidance list of NDT techniques and capabilities.
Joint type Defect type Operation consideration
Compo-
site
Metallic Porosity Voids Adhesion Safety
issues
Skill Cost Coverage
Acoustic
9 9 9 9
Low Low Low Single
point
Radiography
9 9 9
High High High Large
area
Shearography
9 9 9
Med. High High Large
area
Thermography
9 9 9 9
Low Med. Med. Large
area
Ultrasonics
9 9 9 9 9
Low Low to
med.
Low to
med.
Single
point
and
large
area
Visual
9 9 9
Low Low Low N/A
10.3 Procedures, documentation, requirements and application
10.3.1 Procedures and documentation
It is recommended that where possible procedures for NDT inspection are developed by the
manufacturer or operator of the structure in accordance with relevant standards. The following
issues need to be considered when developing an inspection procedure:
- Suitability of technique for proposed inspection
- Is the minimum defect size or parameter to be measured within the capability of the test
method
- Method for verification of measurement and calibration of equipment
- Operator training required and documented
- Assessment of defect detection capability probability of detection analysis
- Establishment of written procedure
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 131
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
An example procedure and documentation, developed for one of the BONDSHIP application
case inspections, is described in section 0.
10.3.2 Production and in-service requirements
The NDT requirements for adhesive joint inspection regarding bonding parameters and possible
defect types can be divided into two groups: production NDT (quality assurance) and in-service
NDT (structural integrity).
- Common questions concerning adhesive joint production
Is the adhesive located correctly?
What is the extent (bead width etc) of the adhesive bonding?
How thick (bondline thickness) is the adhesive?
Is it in contact with the adherends?
Has it cured or reached full strength?
Are there any voids or porosity in the adhesive and what size are they?
Are there any defects at the interface and what size are they?
- Common questions concerning adhesive joints during in-service
Is the adhesive still in contact with the adherend?
Are there any defects at the interfaces and what size are they?
If I can see visual evidence of damage what is the extent of the damage in the joint?
10.3.3 NDT application areas for adhesive bond inspection
The following section looks at the various issues associated with adhesive bonding technology
which may require NDT. Many of these application areas where considered and investigated in
the BONDSHIP programme.
10.3.3.1 Design impacts
Joint/structure design access for NDT inspection methods at different stages of production.
Materials selection cladding materials etc can hamper some NDT methods. Certain materials
may limit the capability of a particular NDT technique: for example, porous adherends can
seriously attenuate ultrasound signals
10.3.3.2 Mechanical testing, materials selection
- Coupon tests use of NDT to screen samples prior to mechanical testing
- Condition monitoring NDT techniques to measure joint parameters during testing
10.3.3.3 Production
- Adhesive placement NDT for bead-width and bond-line measurement, detection of
adhesive movement
- Void detection detection of bubbles created when joint is closed, detection of delamination
due to opening of joint during cure
- Cure monitoring tracking of adhesive cure and measurement of final properties
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 132
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
10.3.3.4 In-service
- Condition check test of key structural areas that are prone to high load forces or
environmental attack
- Impact damage testing of joints subjected to high impacts or mechanical damage
There is much debate in industries which currently utilise adhesive bonding technology
concerning the role of NDT for production monitoring. In some cases it is has been argued that
good process control can negate the need for an NDT inspection programme. It can also be
argued that there is a role for NDT to provide tuning of a production technique and to provide a
safety net to detect breakdowns in production equipment. In general the requirement for NDT
will depend on the criticality of the joint in the overall structure and the safety and knock down
factors employed in the design phase.
10.4 Ultrasonics equipment and inspections
10.4.1 Ultrasound basics
The term ultrasound generally covers sound waves of frequency from the upper audible range
(20 kHz) up to radiofrequency (100 MHz). Ultrasound, up to frequencies of 1 MHz, will readily
propagate through gas, liquids and solids. Above this frequency ultrasound propagation is
limited to liquids and solids.
A theoretical description of ultrasound propagation is beyond the scope of this report and only
the basic properties of ultrasonic waves are covered. In general, ultrasonic waves travel by
elastic disturbance of molecules in gases, liquids and solids as compression waves with the wave
speed dependent on the bulk modulus of the material. As solids also have a shear modulus
ultrasound can also propagate as a shear wave. The following figures show the two types of
wave motion and the equations relating wave speed to elastic modulus and density.
Figure 10-3: Principles of ultrasound wave propagation in liquids and solids.
Compression waves (liquids)

2
c
v K =
Compression wave
Shear wave motion Shear waves (solids only)

2
s
v G =
Compression waves (solids)
) (
3
4
) (
2 2

s c
v v K =
Direction of wave propagation
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 133
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Here K is the bulk modulus, and G the shear modulus. These are related to the material
properties of density () and ultrasound wave speed for compression (v
c
) and shear (v
s
) waves.
In this way ultrasound velocity measurements can be used to measure the mechanical properties
of materials such as the shear modulus in adhesives which can be related to curing of the
polymer [6].
Compression and shear wave propagation is also characterised by an attenuation coefficient
which describes losses in the wave energy as the sound travels through a material. These losses
can be caused by frictional losses in visco-elastic materials (e.g. plastics and resins) or by
scattering (e.g. porosity) or a mixture of both (e.g. carbon fibre reinforced resins).
The reason ultrasound is useful as inspection technique for adhesive bonding is that it is very
sensitive to interfaces and discontinuities in the propagation medium (e.g. presence of bubbles).
Using a concept called acoustic impedance the change in amplitude of ultrasound waves at
boundaries can be described as shown in the following figure.
Figure 10-4: Ultrasound propagation across interfaces.
Here the amplitude of reflected (R) and transmitted (T) waves are directly related to the wave
velocity (c) and the density () of each medium. If Z
1
and Z
2
are the same (matched interfaces)
then all the energy will be transmitted across the interface and none reflected. If Z
2
is very much
less than Z
1
, then most of the energy is reflected with very little transmitted. It is these physical
relationships that allow ultrasound to detect changes in acoustic impedance within adhesive
joints and to detect defects such as air bubbles (low acoustic impedance) which are strong
reflectors compared to the surrounding adhesive (medium acoustic impedance) and metallic
adherends (high acoustic impedance).
Ultrasonic waves are usually excited using a machined disc of piezo-electric material which is
made to resonate by the application of a short pulse of electrical energy. The frequency of
resonance is determined by the thickness of the disc and the amplitude by the voltage of the
exciting pulse.
R
T I
Transmitted and reflected amplitudes
1 2
1 2
Z Z
Z Z
I R
+

=
1 2
2
2
Z Z
Z
I T
+
=
Medium 1 Medium 2
Medium 1:
1 1
c Z =
Medium 2:
2 2
c Z =
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 134
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-5: Conversion of electrical pulses into ultrasound pulses.
The disc, when arranged within an ultrasonic probe, can be used in two basic configurations,
where either the same probe is used to excite and receive the ultrasound (pulse-echo) or where a
separate transmitter and receiver probe are used (through transmission). For the inspection of
solid materials a liquid gel is required to couple the ultrasound from the probe into the test
material.
10.4.2 Inspection types
The following describes some of the basic inspections that can be performed on adhesive joints.
10.4.2.1 Pulse echo
Figure 10-6: Pulse echo ultrasound inspection.
In pulse echo an incident pulse of ultrasound is transmitted into the test sample and the reflected
echoes are analysed. The amplitude and position of the echoes are used to detect the presence of
defects and measure physical properties such as the joint thickness. This inspection is preferred
since it only requires access to one side of the structure.
Electrical
pulse
Ultrasound pulse
Piezo-electric
energy conversion
Adherent
Adhesive
Adherent
Probe
Top surface signal
Interface signals
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 135
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
10.4.2.2 Through transmission
Figure 10-7: Through transmission ultrasound inspection.
In through transmission the incident pulse of ultrasound is transmitted through the test sample
and received on the other side. The method makes detection of defects easier since voids will
cause a dramatic loss of signal but requires access to both sides of the structure.
The standard piece of equipment used to make these types of ultrasonic inspection is called a
flaw detector. This unit contains the electronic circuitry to drive the ultrasonic probes and
receive the small amplitude echoes from the test sample. Most units have an oscilloscope
display to allow the operator to view the ultrasound waveforms and a number of measurement
tools to analyse the amplitude and position of the ultrasound echoes for defect detection and
thickness measurement. Examples of using this type of equipment in BONDSHIP are given in
later sections.
10.5 Application examples
10.5.1 WP 2: Acceptance and qualification of joints task 2.2: Selection of
materials
During this part of the project all the work horse materials selected by the project partners were
characterised using ultrasound to enable preparation of NDT techniques for the inspection work.
This enabled methods to be developed to inspect the adhesive materials based on modelled data
from simulated joints.
10.5.1.1 Adhesive and adherend characterisation
The adhesive types screened by the BONDSHIP partners were drawn from flexible adhesives
(low modulus) and rigid adhesives (high modulus). When the selection of these materials was
finalised they were characterised by NDT Solutions. This was done using a technique developed
by the author to excite both compression and shear wave ultrasound in the test samples as shown
below [4].
Probe
Probe
Adherent
Adhesive
Adherent
First signal through joint
Echoes in layers
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 136
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-8: Test arrangement used for adhesive characterisation.
The adhesive is formed into a thin sheet of material and placed in a water bath between an
ultrasonic transmitter and receiver. An incident compression wave, with the sample set at an
angle, will excite both compression and shear waves in the adhesive through a phenomenon
known as mode conversion. The characterisation of the two wave types provides important
physical data (ultrasound velocity and attenuation) to enable the ultrasonic response of a joint
formed with the adhesive to be predicted.
For each of the application cases the adhesive materials where characterised using these
techniques. Typical results from the rigid and flexible adhesives used in the BONDSHIP
programme are given in the following figures.
Araldite 420
-10
0
10
20
30
40
0 5 10
Frequency (MHz)
L
o
s
s

(
d
B

m
m
-
1
)
A comp.
B comp.
A shear
B shear
Figure 10-9: Repeat compression and shear wave ultrasonic attenuation measurements for
Araldite 420.
Test sample
S wave
C wave
Receiver Transmitter
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 137
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-10: Comparison of compression wave attenuation data for rigid and flexible
adhesive employed in the BONDSHIP materials selection process.
An important result from this research was the measured attenuation in the Sikaflex material
which was found to be very low compared to other similar performance materials. This meant
that testing of joints form with thick bondlines would be easier to test.
Most of the substrates employed in the BONDSHIP programme were metallic being either steel
or aluminium alloys. The basic ultrasonic properties of these materials is well understood and in
most cases published values for these materials were used, unless a calibrated measurements was
required (e.g. for thickness measurement).
10.5.1.2 Adhesive joint simulations
Using data from the project partners concerning projected joint designs, and the expected
adherend and adhesive thicknesses it was possible to simulate the ultrasonic response for a wide
variety of joints with different types of defects. This allowed NDT Solutions to begin work on
developing inspection methods before the joints were designed and constructed. A multilayer
wave propagation model was used to simulate the adhesive joints [5]. This model was extended
as part of the BONDSHIP work to enable simulation of thick joints with low frequency
ultrasound pulses and required considerable reworking of the software.
Typical joint simulations produced by the model are given below. In this case a 19 mm thick
joint was simulated consisting of a 10 mm thick composite layer, a 1 mm adhesive layer and a 8
mm thick aluminium layer.
Rigid
0
10
20
0 5 10
Frequency (MHz)
L
o
s
s

(
d
B

m
m
-
1
)
Loctite 410
Araldite 420
Flexible
0
10
20
0 5 10
Frequency (MHz)
L
o
s
s

(
d
B

m
m
-
1
)
Sikaflex 265
Sikaflex 292
Sikaflex 552
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 138
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-11: Joint simulations for a composite to aluminium bonded and disbonded joint.
The top graph shows a simulation of an intact composite to aluminium joint and the
characteristic ultrasound response of the structure using a 5 MHz ultrasound probe. The bottom
graph shows the joint response with the rear adherend disbonded and the change in ultrasound
response can clearly be seen.
Using these techniques joint simulations were prepared for each of the major application cases
and the resultant data used to investigate optimal ultrasound techniques for the inspection work.
10.5.1.3 Coupon testing
As part of the adhesive materials selection by the project partners a range of coupon test samples
were produced. These were passed to NDT Solutions for testing and were compared to
simulated data. The following data shows measured data from a joint coupon consisting of 2.0
mm aluminium substrates bonded with 2.5 mm of Sikaflex 265 flexible adhesive. The
measurements were made at 10 MHz in pulse echo.
Composite
Adhesive
Aluminium
Probe
Composite
Adhesive
Aluminium
Probe
C-A bond
-0.03
0.02
0 10 20 30 40
Time ( s)
A
m
p
l
i
t
u
d
e

(
a
r
b
.
)
C r-dbond
-0.03
-0.01
0.01
0.03
0.05
0 10 20 30 40
Time ( s)
A
m
p
l
i
t
u
d
e

(
a
r
b
.
)
0 10 20 3
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 139
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-12: Measured data from an intact coupon and a delaminated bond which had
suffered cohesive failure in the adhesive.
In all cases good results were obtained when comparing simulated data with the measured
coupon data.
10.5.2 WP3: Application case 1 - task 3.4: Performance of joints and critical
defects
The Vosper Thornycroft application case presented a range of joint types and made use of
composite honeycomb panels bonded to an extruded aluminium frame. The welded frame was
constructed in modular form with modules bonded together as shown in the following figures.
Figure 10-13: Pictures of Vosper Thornycroft application case.
Intact bond
-1.0
-0.5
0.0
0.5
1.0
0 5 10 15 20
Time (s)
A
m
p
l
i
t
u
d
e

(
a
r
b
.
)
Delaminated bond
-1.0
-0.5
0.0
0.5
1.0
0 5 10 15 20
Time (s)
A
m
p
l
i
t
u
d
e

(
a
r
b
.
)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 140
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
A test sample was provided which had examples of the main joints to be inspected as shown in
the following figure.
Figure 10-14: VT test structure showing principal joints.
- Key to joint types:
1. Honeycomb panel to frame joint
2. Inter-module butt-strap joint
3. Frame to frame joint
4. Honeycomb panel skin to core bonding
The use of box section extrusions introduced considerable problems with access to the module to
module joints (joint number 3 in figure). For this reason techniques for joints number 1 and 2
were investigated. NDT was used to support strength testing of the primary joints and for
inspection of the main structure and a trial was undertaken to investigate the types of defect that
could be detected. The testing programme for the VT application case is described in the
following sections.
10.5.2.1 University of Southampton monitoring of joints
Ultrasound imaging has the capability of screening test coupons prior to mechanical testing. To
demonstrate the sensitivity of ultrasound to cohesive and adhesive failure modes a DBC test
coupon was scanned directly after mechanical testing as shown below:
1
2
3
4
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 141
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-15: Ultrasonic C-scan of DBC sample after DBC test.
The results demonstrate the sensitivity of ultrasound to both adhesive and cohesive failure modes
with strong correlation between the ultrasound images and photographs of the bondline
interfaces.
10.5.2.2 University of Southampton monitoring of joint displacement
A box beam section from the VT application case was monitored using 10 MHz through
transmission ultrasound. The figure below shows the joint topology and probe arrangement.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 142
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-16: Instrumentation and test assembly using through transmission ultrasound to
monitor joint.
The aim of this test was to monitor changes in bondline thickness as the joint was loaded to
provide additional information on the stresses in the adhesive joint. The following figure shows
a plot of the adhesive bondline displacement as a function of applied load to the joint up to the
point of failure.
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
0.2
0 100 200 300 400 500
Time (s)
D
i
s
p
l
a
c
e
m
e
n
t

(
m
m
)
0
5
10
15
20
25
L
o
a
d

(
k
N
)
Adhesive
Load
Figure 10-17: Through transmission compression wave data versus time during bend test.
This is a novel result and further research in this area is planed between the University of
Southampton and NDT Solutions.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 143
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
10.5.2.3 VT test panel inspection
The test panel was investigated using ultrasound to inspect the accessible joints and good results
were obtained. However one of the primary requirements indicated by VT was the inspection of
the honeycomb panel to frame bonding. Ultrasound could be used to inspect the honeycomb
panel itself (skin to core bonding) but not the panel to frame joints (see section 10.5.2.4). It was
agreed therefore that the panel would be sent to Laser Testing Instruments (LTI) for inspection
using a laser shearography technique.
Figure 10-18: Test sample sent for laser shearography testing showing primary test areas.
Figure 10-19: LTI 5100 Laser shearography system.
LTI tested the panel and the following is a summary of their findings: The LTI laser
shearography technique is a non contact NDT method that measures a material response to an
applied force or stress. The process uses a CCD camera to take a set of phase stepped reference
sheared images of the laser speckle pattern generated on the surface at one stress state and after a
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 144
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
stress change, another set of phase stepped images is taken. The stress change can be introduced
by thermal excitation or by positive or negative ambient pressure changes. The correlation
between the image in the newly stressed state is then compared with the base line image.
Interference fringes, which are due to the optical interferometer, are lines of iso-strain which
indicate the mechanical integrity of laminate or sandwich structures. Unwrapping the images
removes the fringes yet keeps all the phase data.
The VT application case panel was tested in an LTI 9000 vacuum chamber, viewed by a LTI
5100 phase stepping camera. Using vacuum stressing with a pressure change of 1.08 psi the
upper and lower sections of the butt-strap indicated on figure 18 were imaged. The results are
shown below.
Figure 10-20: Shearography image of upper section of centre inter panel joint.
The inspection of the upper section showed variations in the bondline, but as no phase change
was detected it was assumed that this is showing variations in the bondline position rather than
changes in bondline strength.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 145
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-21: Shearography image of lower section of centre inter-panel joint.
The inspection of the lower section showed an indication on the right hand side and as the phase
change is 180 degrees out of phase with the response from the bondline it assumed to be a local
weakness such as an air bubble trapped in the bondline.
Figure 10-22: Shearography image of lower left-hand corner panel.
The dark indication is associated with a change in strain compared with the rest of the bondlines
running vertical and to the right of the corner. No build information was supplied to LTI as this
was a blind trial but wooden bondline spacers were placed in the bondline at this position and
the shearography successfully detected this pseudo defect.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 146
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The shearography results have clearly shown the potential of the technique for the large scale
imaging of bonded composite panels and has demonstrated sensitivity to large voids (spacers)
and changes in the bondline parameters (position and air bubbles)
10.5.2.4 VT mock-up inspection
A photograph of the VT mock-up is shown in the following figure.
Figure 10-23: Photograph of VT mock-up during inspection work.
Visual inspection of the mock-up during production had identified several issues including voids
in the adhesive layers caused either by movement of the structure during bonding or by
inadequate filling of large gaps. The honeycomb panels had been used both as floor and wall
panels and there were several areas where the dropping of tools had caused impact damage.
Discussions with personnel at VT had identified an interest in being able to measure the extent of
bonding on the butt-strap joints. Two areas of the VT mock up were therefore identified for
inspection; the inter module butt-strap joints (bonding quality) and the honeycomb panel skin
bonding (impact damage assessment).
The large flat areas of bonded structure made it suitable for inspection using a new large area
rapid scanning technique developed by NDT Solutions. The RapidScan system employs a water
filled rubber wheel sensor and an ultrasonic array to generate images (C-scans) of the test
structure in real time.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 147
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-24: RapidScan large area ultrasound inspection system.
Butt-strap inspection
The following figure shows a photograph of the RapidScan being used to inspect an inter-
module butt-strap joint.
Figure 10-25: RapidScan inspection of inter-module butt-strap joints
The RapidScan system performed well and was able to produce images of the extent of bonding
as shown in the following figures.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 148
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-26: Repeat ultrasonic scans of butt-strap joint with diagram of scanned area.
The Colourmap on the ultrasonic imaging system was set to show areas of good bonding
(green/yellow) and areas of poor bonding (red/white). Black areas indicate areas where
measurements could not be made (poor signal quality). The black line indicates a join between
two butt straps. The areas of poor bonding shown by red patchiness are circled and it is believed
that this is caused by squeezing and relaxing of the joint during manufacture causing rippling of
the adhesive as shown in the following figure.
Figure 10-27: Rippling of adhesive which could give rise to observed patchiness.
During the inspection dust contamination of the equipment became a concern and the keyboard
required a dust cover. The inspection work carried out in a shipyard environment demonstrated
the need that equipment should have dust covers or utilise dust protection to IP65.
Honeycomb skin to core inspection
Scans of the honeycomb floor panels were made to assess the damage caused by tool impacts.
Typical ultrasonic scans of the delaminations are shown in the following figures. In each case
the measured impact diameter and the visible impact diameter are given.
Box section extrusions
Bonded regions Butt-strap plate
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 149
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-28: Large impact area, ultrasonic size (16 mm), visible size (10 mm)
Figure 10-29: Barely visible impact damage (BVID), ultrasonic size (8 mm), visible size (4
mm)
These results demonstrate that although the use of advanced materials may be attractive in terms
of structure performance (e.g. strength to weight ratio) more attention to working practices is
required to minimise tooling damage during manufacture. The inspection work on the floor
panels demonstrates the capability of ultrasonic NDT to monitor and assess impact damage.
Conclusions
The VT inspection demonstrated the capability of ultrasonic NDT for the inspection of a range of
joint types and potential defects. The main limitation for the inspection work was the limited
access afforded to large parts of the structure. This demonstrates the need for QA and NDT to be
considered early in the design stage to ensure that suitable access to structure is provided. The
inspection results showed that although ultrasound offered an inspection solution, a scanning
device is required in order to reduce the inspection time for large bonded areas.
10.5.2.5 VT defect test sample inspection
Vosper Thornycroft prepared an aluminium-to-steel butt-strap test sample containing a range of
defects to simulate possible production and in-service problems. The following defects were
placed in the adhesive bondline:
1. PTFE tape insert (to simulate lack of interfacial adhesion)
2. Air voids (to simulate lack of adhesive material)
3. Bubble wrap insert (to simulate large pores)
4. Water bubble (to simulate moisture ingress)
A photograph of the test piece and a diagram of the test piece showing the location of the defects
are given in the following figures.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 150
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-30: Defect test sample containing a range of simulated defects.
Figure 10-31: Location of defects in test sample.
The test sample was scanned with the RapidScan with the system set to interrogate the interface
between the but-strap plate and the adhesive bondline.
Figure 10-32: Resultant ultrasonic C-scan of defect test sample.
Water Air void PTFE insert Bubble wrap
Water Air void PTFE insert Bubble wrap
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 151
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
As can be seen the system successfully all the defects although sensitivity to the bubble wrap and
water filled sample was lower than that for the PTFE insert and air void. This is due to acoustic
impedance matching between the compressed bubble wrap and water with the adhesive material.
10.5.3 WP4: Application case 6 casing
Only one joint type was inspected in this application case which was the scantling joint on the
casing structure. This utilised a 16 mm thick flexible adhesive joint. A test structure was
supplied for non destructive testing prior to fatigue testing by Fincantieri.
This joint was tested at NDT Solutions laboratory with an ultrasound research instrument using
both high (10 MHz) and low (1 MHz) frequency probes. A schematic of the application case
and a photograph of the test joint are given below.
Figure 10-33: Fincantieri AC6 - engineering drawings and photograph of test joint.
The requirement for this inspection was to inspect the bulk of the adhesive bondline and the steel
and aluminium interfaces. A 10 MHz pulse echo measurement was made at each on each side of
the joint to test the interfaces as show in the following figures.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 152
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-34: Photograph of measurement of aluminium interface and resultant test data.
Figure 10-35: Photograph of measurement of steel interface and resultant test data.
No interfacial defects were found in the test sample. However, the inspection only interrogates
the interfaces and not the bulk of the adhesive. As this type of flexible adhesive was widely used
in the BONDSHIP project it was decided to investigate the use of low frequency ultrasound to
the test the bulk of the adhesive in two planes. Measurements made earlier in the project (see
section 10.5.1.1) suggested that a low frequency inspection was possible. The results of the
inspection are shown in the following figures.
10 MHz pulse echo, steel interface
-0.10
-0.05
0.00
0.05
0.10
2.0 4.0 6.0 8.0 10.0
Time (s)
A
m
p
l
i
t
u
d
e

(
V
)
10 MHz pulse echo, aluminium interface
-0.10
-0.05
0.00
0.05
0.10
2.0 4.0 6.0 8.0 10.0
Time (s)
A
m
p
l
i
t
u
d
e

(
V
)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 153
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-36: Photograph of measurement of bulk joint and typical test data.
Figure 10-37: Photograph of measurement of bulk adhesive and typical test data.
No defects were found although signal anomalies were observed due to presence of pre-cured
spacer blocks in the joint. Good sensitivity to the joint topology was observed and transverse
flexing of the joint resulted in changes in alignment of the substrate faces which could be seen in
the ultrasound signals.
The test results indicated that provided low frequency probes (< 1 MHz) can be used with
standard flaw detectors, interrogation of thick (> 10 mm) flexible adhesives is possible.
10.5.4 WP5: Application cases at Meyer shipyard
A range of joint types were inspected on the Meyer Werft application case including a load
bearing wall structure (AC1), deck foundation fixings (AC2), gutter ways (AC7) and balcony
structures (AC9). These applications employed several adhesive types (flexible and rigid) with a
range of bondline thicknesses. It was agreed that these joints would be inspected using a
standard digital ultrasonic flaw detector to investigate the inspection capability of current NDT
instrumentation.
1 MHz test across bead
-0.006
-0.004
-0.002
0.000
0.002
0.004
0.006
0.0 20.0 40.0 60.0 80.0 100.0
Time (s)
A
m
p
l
i
t
u
d
e

(
V
)
Reflection from
bottom of bead
1 MHz test across joint
-0.010
-0.005
0.000
0.005
0.010
0.0 20.0 40.0 60.0 80.0 100.0
Time (s)
A
m
p
l
i
t
u
d
e

(
V
)
Reflection from
bottom of joint
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 154
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The instrument used was a Sonatest Masterscan 340 digital ultrasonic flaw detector operating
with 0.5 MHz, 1.0 MHz, 5.0 MHz and 10 MHz pulse echo probes.
The following sections review each application case structure showing the bondline location,
measurements taken and the inspection results.
10.5.4.1 AC1 secondary load bearing wall structure
This structure was bonded with Terostat 8590 UHV/M with an approximate bondline thickness
of 20 mm. A drawing and photograph of the structure is given below.
Figure 10-38: Meyer Werft application 1, engineering drawings and photographs
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 155
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The adhesive bondline on the external wall lip (circled in white) was inspected using a 10 MHz
pulse echo probe to determine the extent bonding and to detect voiding at the aluminium
adhesive interface. Measurements were attempted with a lower frequency probe (1 MHz) in
order to measure the bondline thickness but the signal attenuation proved excessive and no
measurement was possible.
It was decided to focus on the outer wall lip bondline since this would be most sensitive to
environmental attack during service and thus it was deemed important to establish that there
were no large voids or areas with no adhesive where corrosion could be initiated.
The lip area was divided into 3 strips and inspected at approximately 100 mm intervals as shown
below.
Figure 10-39: Test grid for external wall lip
At each measurement point the echo amplitude of the echo from the aluminium-adhesive
interface was monitored using a measurement gate and the structure was marked to record the
presence or absence of adhesive. Typical ultrasound signals from a bonded and disbonded
region are shown in the following figure.
Figure 10-40: Typical bonded and disbonded data reading from flaw detector.
The data was processed to measure the relative change in amplitude in the measurement gate and
a map of the bonded region produced as shown below.
Bonded Disbonded
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 156
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
0.7-1
0.4-0.7
0.1-0.4
-0.2-0.1
Figure 10-41: Colour map of bonded and disbonded regions
It can be seen that there are areas of no adhesive in the corner of the lip and these could indicate
hot spots where corrosion would have a greater chance of establishing itself on the structure.
These areas could be repaired by additional application of adhesive to completely seal the joint
against moisture ingress.
10.5.4.2 AC7 gutter ways
A number of gutter ways were bonded using flexible adhesives (thickness approximately 10 mm)
to seal the gutter way as shown below.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 157
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-42: Meyer Werft application 7, engineering drawings and photographs
During structure testing, visual inspection had shown that the gutter ways were subject to leaking
of water where there had been insufficient bonding during construction. It was decided to see if
the ultrasonic flaw detector could be used to assess the adhesive bead pattern to enable the
operator to detect areas where there may be insufficient bonding. An 80 mm by 30 mm
rectangular area was scanned to try and determine the extent of bonding. The bonded area was
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 158
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
marked with a pen to show the extent of bonding. The markings were later successfully
compared to visual analysis of the bead position during the repair trials.
Figure 10-43: Testing of gutter way and close-up of flaw detector display.
0.7-1
0.4-0.7
0.1-0.4
-0.2-0.1
Figure 10-44: Adhesive bead width measurement showing extent of bonding on gutter way.
In order for this type of inspection work to be routinely used a method for rapid inspection of the
structure would need to be developed, similar to that used for the butt-strap inspection on the VT
application case (see section 10.5.2.4).
10.5.4.3 AC9 balcony structure
The balcony structure was of particular interest since this was a load bearing structure and
therefore could require NDT to assess the bondline thickness during production and monitoring
of the bondline interfaces during service. Engineering drawings and photographs of the structure
are given below.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 159
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-45: Meyer Werft application 9, engineering drawings and photographs
The ultrasonic flaw detector was used to measure the bondline thickness in several locations on
the structure as shown below.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 160
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 10-46: Typical ultrasound data with automated thickness reading to assess adhesive
bondline.
The following figure shows a detail of the balcony structure and the locations where the bondline
thickness was measured.
Figure 10-47: Diagram showing measurement locations and recorded bondline thicknesses.
It can be seen that there are large variations (200 %) in the bondline thickness between the left
and right hand side of the balcony support. These measurements would be important for
0.46 mm
0.72 mm
0.54 mm
1.22 mm
1.74 mm
1.50 mm
1.68 mm
2.28 mm
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 161
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
production since they give quantitative information on the bondline thickness and hence an
indicator of joint strength and performance under load.
10.6 Conclusions
The research reported in this document has shown the applicability of NDT techniques to
support production and in-service monitoring of adhesively bonded marine structures. Research
in the early stages of the BONDSHIP project demonstrated the suitability of ultrasound as an
inspection technique for most of the proposed materials and joint designs considered for the
application cases. Novel use of experimental data and modelling techniques enabled the
simulation of many of the inspection prior to the construction of test assemblies and the
application case structures. Inspection strategies were developed for each of the application
studies based on the joint designs and measurements of the adhesive and adherends. In some
cases new inspection methods utilising lower ultrasound frequencies or large areas scanning
techniques were employed. Close working and consultation with the manufacturers of the
application cases enabled the key requirements for future NDT inspection of commercial
structures to be identified and the role of NDT for production monitoring and assessment of the
joining process to be assessed.
Valuable experience in deploying NDT equipment in a shipyard environment at Vosper
Thornycroft and Meyer Werft was gained when investigating the role of NDT for supporting
repair trials.
The issues of standards and development of guidelines has been reviewed and this report
contains a list of current standards and operator certification with guidelines for the selection and
deployment of NDT techniques for adhesive bond inspection. An inspection procedure for one
of the BONDSHIP application cases is contained in section 11.4.5 on page 212.
10.7 References
[1] Richard Freemantle, Non destructive adhesive bond inspection techniques and
guidelines, BONDSHIP Report No.: 2-24-D-2002-01-4; 2003
[2] British Institute of Non Destructive Testing year book (2003). Ed. David J. Gilbert,
ISSN 0952-2395
[3] Brewis, D. M. and Briggs, D. (1985). Industrial adhesion problems. John Wiley and
Sons, New York. ISBN 0-471-84005-X
[4] Freemantle, R. J. et al (1993). A model fitting approach to the broad band
measurement of ultrasonic wave velocities in thin samples of engineering material.
Meas. Sci. Technol. 4, 1129-1137.
[5] Freemantle, R. J. (1995). Ultrasonic compression wave evaluation of adhered metal
sheets and thin sheet materials. PhD thesis, Keele University, UK.
[6] Freemantle, R. J. et al (1998). Combined compression and shear wave ultrasonic
measurements on curing adhesive. Meas. Sci. Technol. 9.
[7] Kinlock, A. J. (1987). Adhesion and adhesives, Chapman and Hall, London, ISBN 0-
412-27440-X
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 162
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
[8] Krautkramer, J. and Krautkramer, H. (1990). Ultrasonic testing of materials. 4
th
Edition, Springer Verlag. ISBN 0-387-51231-4
[9] Lavender, S. J. (2002). Lavender International NDT Operator Certification Series
Article 4 Levels of Certification. (c) BINDT. http://www.lavender-
ndt.co.uk/pdf/oc_article4.pdf
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 163
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11 DEFINITIONS AND PROCEDURES
11.1 Test procedures
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 164
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.1 Measurement of pH value (IFAM test standard WP-AA-60)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 165
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.2 Lap shear test - strength
11.1.2.1 Sika flexible adhesives
Lap-Shear Test: for rigid and flexible adhesive
Standards: ASTM D 1002, DIN EN 1465, ISO 4587,
Sika SQP046-2
Adherend length: 100 mm (300 mm for composite)
Adherend width: 25 mm (40 mm for comp.)
Adherend thickness: 5 mm for steel, 6 mm for aluminium, 10 mm for composite
Overlap: 12.5 mm (50 mm for comp.)
Adhesive thickness: Depends from the process properties of the adhesive, as proposal:
Rigid: 0.2, 0.5, and 1 mm
Flexible: 3 and 5 mm. (The relationship of Overlap divided by thickness
has to be larger than 4, otherwise a defective shear deformation can occur)
Test speed: The test speed need also some different graduates, according to the EN
1465, so that the fracture occurs nearly in the same time (6520)s. This is
a good basis for comparable and useful tests.
11.1.2.2 CTA rigid adhesives
Specification of single lap joint specimens
Adhesive thickness: 1 mm
Sample width: 25mm (metal)
40 mm (composite)
Adherends thickness : 5 mm (steel)
6 mm (aluminium)
10 mm (composite)
Overlap :
12.5 mm
(metal)
50 mm
100 mm
(metal)
300 mm
(composit
50 mm (metal)
(Area in test grips The
tabs can be made with a
material different from
the adherent materials)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 166
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Preparation of the specimens:
5 specimens are prepared from two panels of dimensions :
Metals : Composites :
The two panels are bonded together then cut carefully into 5 specimens.
175 mm
100 mm
300 mm
300 mm
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 167
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.3 Lap-shear test - strain to failure (IFAM test standard WP-AA-11)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 168
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 169
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 170
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.4 Boeing wedge test (ASTM D 3762-98)
Written by Roland Joannic, CTA
11.1.4.1 Principle of the method
On aluminium alloy, one of the most sensitive durability experiments of the interface is the
Boeing wedge test, which is based on a constant displacement double-cantilever-beam (see
Figure). A wedge is inserted at one end of two bonded plates, producing a crack. The sample is
placed in the conditions of 50C and 95% relative humidity and the moisture tends to degrade
the adhesive and the interface. One then observes a propagation of the crack towards the other
end of the plates. For a durable surface treatment, the crack remains cohesive and propagates far
more slowly than for a bad surface treatment for which one observes an adhesive failure.
a
If we call d the wedge thickness, a the crack length at a given time (defined as drawn on Figure)
and h the adherend thickness (h=d=3.2 mm for our reference experiments on aluminium
2024T3), we can calculate the strain energy release rate G. This fracture mechanics parameter
represents the peeling forces exerted at the crack tip. From beam theory, G is calculated by :
4
2 3
16
3
a
d Eh
G =
where E is the adherend modulus.
11.1.4.2 Geometrical quantities recommended
Adhesive thickness : 0.5 mm
Sample width : 25 mm
Sample Length : 170 mm
Adherends thickness : 5 mm (steel)
6 mm (aluminium)
10 mm (composite)
170 mm
180 mm
45 mm
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 171
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.4.3 Preparation of the specimens
Two panels (150*180 mm) are bonded together then cut into 6 specimens. A zone of length
L=45 mm remains not bonded for initiating a crack.
11.1.4.4 Wedges:
Thickness : 2 mm for aluminium.
1.5 mm for steel.
11.1.4.5 Calculation of the wedge thickness :
We wish to adapt the standard test which is given mainly for one material (aluminium 2024T3 of
Youngs modulus E
ref
), and two thickness of the adherend and of the wedge : h
ref
=d
ref
=3.2mm.
In order to obtain a comparable evolution of the crack length a with the time, it is necessary to
verify :
2 3 2 3
ref ref ref new new new
d h E d h E =
where the subscript new refer to the new material under study.
This give the following table:
Elastic
modulus in
GPa
Adherend
thickness h in
mm
Wedge
thickness d in
mm.
Wedge
thickness d in
mm (for an
adhesive
thickness
0.5mm)
Recommende
d value of the
wedge
thickness d in
mm
ASTM D3762-79 for
aluminium 2024T3
72 3.2 3.2
Aluminium 72 6 1.24 1.74 2
Steel 200 5 0.98 1.48 1.5
Composite
glass/vinylester
30 10
25
40
60
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 172
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.4.6 Calculation of the length of the not bonded area :
We want to limit the forces that are necessary to insert the wedge. This force is given by :
3
8
3
L
dEbh
F =
where L is the length of the not-bonded area. L=19 mm in the standard.
Having the same value of the force implies :
3
3
3
3
ref
ref ref
new
new new
L
h E
L
h E
=
The values of L are listed in the next table :
Value of L in mm Recommended
value of L in
mm
ASTM D3762-79 for aluminium
2024T3
19
Aluminium 40 45
Steel 41 45
Composite glass/vinylester 44 45
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 173
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.5 Measurement of electrical resistance (Sika Test Procedure 316)
Determination of electrical specific volume resistance
Contents
1. Scope and objective
2. Apparatus
3. Procedure
3.1 Measurement with sheet-method
3.2 Measurement with bonded sample
3.3 Calculation
4. Test report
5. Appendices
1. Scope and objective
This test procedure delines the determination of the electrical volume resistance of
adhesives and sealants.
a) Standard-method on sheets made from adhesives
(according to DIN 53482/VDE 0303)
b) Bonding method which determines the electrical values on a bonded sample.
2. Apparatus
2.1 Method A
- Voltmeter / Ampere meter / power supply
- Electrodes type C
- Test samples made according to TS 308
- Conditioned room 23 / 50 %
- Distilled water
2.2 Method B
- Ohmmeter
- Copper sheets 1 x 100 x 200 mm
- Copper strips 1 x 15 x 20 mm
- Teflon bars 5 x 20 x 100 mm
- Masking tape
- Sharp knife
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 174
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
3. Procedures
3.1. Method A (Sheet-method)
A sheet of adhesive / sealant is produced according to TS 308 and measured after
the 2 x 7 days curing cycle.
Range of measurement
10 - 10
4
O : 0,1/mm
10
4
- 10
9
O : 1 V/mm
10
6
- 10
12
O : 100 V/mm
10
8
- 10
15
O : 1000 V/mm
A second measurement is done after an additional 14 days immersion in distilled
water at 20C.
3.2 Method B (bonding method)
Copper sheet and three copper strips are cleaned with scotchbrite very fine and than
cleaned with SC-205.
Masking tape is applied on the copper sheet giving a free space of 3 times 15 mm
(use gloves to prevent surface contamination).
Fix the Teflon bars on each end of the sheet with a masking tape. Apply the adhesive
or sealant in a triangular rope 15 x 15 mm on the non covered copper. Apply the
copper strip on the top, press it down with a flat profile. Watch the correct
positioning.
Let cure the samples for 24 hours at 23C/50%. Fill up cavities with the same lot of
test product and let it cure for total 7 days. Teflon's are removed. The overstanding
adhesive is cut with a knife and removed with the masking tape.
Testing is done after another 7 days curing at 23C/50%. A second testing is done
after additional immersion of the whole sample in distilled water at 20C for 14 days.
3.3 Calculation
Volume resistance =
U (Volt) x section sample (cm
I ( Ampere) x sample thickness ( cm )
Method A :
Volume resistance =
U x 24
V x t
= ( cm)
Method B :
Volume resistance =
U x l x b
V x t
= ( cm )
2
)
O
O
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 175
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
From each sample 3 independent measurements are made
4. Test report
The following shall be indicated in the test report, with reference to this standard.
- Adhesive / sealant
- Batch
- Method
- Measuring dates
- Results (mean)
- Variation of this standard
- Test date
- Tester
5. Appendices
DIN 53482 / VDE 0303
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 176
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 177
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.1.6 Bead test (Sika SQP033-0 and Sika SQP034-0)
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 178
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 179
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 180
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 181
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 182
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.2 Analytical analysis methods for joints
Ship-building industry wishes to have so called easy-to-use design rules. Easy-to-use design
rules, from an initial point of view, are used to be seen as based on simple analytical formulas
and equations, which can be used to design adhesive joints. The formulas not only have to be
simple, but they also must be reliable. In many cases certain assumptions are necessary to find
analytical solutions. These assumptions have to be carefully considered before analytical
solutions can be applied in a design process in order to avoid inaccurate results.
As pointed out in Section 5.1.1 adhesively bonded joints are especially suited for transferring
shear and compressive loads. Numbers of joints in the ship-building industry are of a lap shear
type, of which the single-lap joint is the simplest. Adhesively bonded single- and double-lap
joints have been extensively applied in, e.g., the aerospace and aircraft industry for a long period
of time. A huge amount of analytical analysis methods for such joints has been developed. In the
following some of the most well-known derivations for adhesively bonded single-lap joints will
be discussed. Some of the theories may be generalized to other kinds of joints. However, for
joints with complex geometries, boundary conditions and/or load transfer mechanisms it may be
difficult or impossible to derive suitable analytical analysis methods. In such cases numerical
methods should be applied.
Two main types of analytical approaches exist. The first and simplest approach to a joint design
uses nominal stresses and strains, while the second type of approach uses solutions of differential
equations. The latter solutions become more complicated the closer they describe reality.
Sometimes, much effort has to be put into the second type of design rules and they are not longer
easy-to-use. The nominal-stress-strain approach and some easy-to-use solutions of differential
equations will be discussed now. However, typical parameters required in the analysis of single-
lap joints will be introduced first.
11.2.1 Parameters for single-lap joints
The overlap region of an adhesively bonded single-lap joint is shown in Figure 11-1, which also
contains the most important geometrical parameters describing the joint. The complete list of
parameters required in most (analytical) analysis methods are as follows:
Elastic constants of the adherends: Youngs modulus, E, if the adherends are of the same
material, or E
1
and E
2
if they are made of different materials. For a complete three-
dimensional (3D) description of the elastic behaviour of a homogeneous and isotropic
material a second elastic constant is required. This might be the Poissons ratio or the shear
modulus. Many analytical solutions do not consider the general 3D case. Therefore, they
need only a single elastic constant (for each adherend).
The thicknesses, s
1
and s
2
, of the adherends, or s, if the thickness is the same for both
adherends
Elastic parameters for the adhesive. The shear stiffness, G, and for a 3D, elastic analysis an
additional elastic constant, Youngs modulus, E
a
, or the Poisson ratio, is required.
Experimental testing has shown that most adhesives behave non-linearly. For highly loaded
joints such effects (hyperelastic models for flexible adhesives, and elastic-plastic models for
rigid adhesives) should be included
The thickness, d, or t
a
of the adhesive layer,
The length, l, or l
,
of the overlap region
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 183
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The width, b, of the joint
The (nominal) tensile stresses denoted by o
10
= F/(bs
1
), in adherend 1 and o
20
= F/(bs
2
), in
adherend 2
The (adhesive) shear displacement, A, due to the tensile load, F, of the adherends.
Figure 11-1 Parameters to describe a single lap joint
11.2.2 Analysis using nominal stresses and strains
It is well-known that joints with flexible adhesives do not show considerable stress peaks at the
ends of the joints like joints with rigid adhesives. Therefore, the assumption of constant stress
and strain distributions can be used for a first design approach to flexible adhesive joints. Thus,
nominal stresses and strains can be adopted. Since flexible joints are often used to compensate
relatively large deformations, both stresses and strains should be considered during design.
The nominal adhesive shear stress, t, for the joint in Figure 11-1 is given by
eq. 11-1
0
A
F
b l
F
=

= t
Assuming linear elastic adhesive behaviour the relation between the shear stress and shear strain
reads
eq. 11-2
= t G ,
where is the shear strain in the adhesive. For small values of the shear strain, i.e. <<1, the
following relation may be introduced
eq. 11-3
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 184
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
( ) ~ =
A
tan
d
To illustrate the accuracy provided by the approximation above, it can be stated that with = 10
the difference between and tan() is 1%, with = 20 the difference is about 4% and with
=30 the difference is 10%.
For certain applications flexible adhesive joints may also be subjected to considerable through-
thickness loads. The nominal (through-thickness) tensile (or compressive) stress, o, may then be
expressed by
eq. 11-4
b l
F

= o ,
where F is the force acting in the through-thickness direction. Assuming linear elastic adhesive
and applying Hookes law yields
eq. 11-5
b l E
F
E
a a

= =
o
c ,
where c is the (through-thickness) normal strain of the adhesive. The normal strain may also be
expressed by
eq. 11-6
d
d A
= c ,
where Ad is the increase (or decrease) in the adhesive thickness due to the tensile (or
compressive) load F. By combining eq. 11-5 and eq. 11-6, we obtain
eq. 11-7
b l E
d F
d
a

= A
However, the adherends restrain the deformation of the adhesive transverse to the tensile axis.
Therefore, the tensile strain within the adhesive and the change in adhesive thickness are
overestimated by the predictions in eq. 11-5 - eq. 11-7. Thus, E
a
is the lower limit on the joint
stiffness in the through-thickness direction. On the contrary, by assuming the adhesive axial
deformations transverse to the (through-thickness) loading direction to be zero, one obtain the
upper limit, E
u
, for the joint stiffness. From linear elastic analysis this upper limit may be
expressed by
eq. 11-8
a u
E E
2
2 1
1
v v
v

= .
For a conservative design approach of a joint loaded in tension (i.e. in the through-thickness
direction) two situations have to be distinguished when using nominal stresses and strains:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 185
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Firstly, if the applied load is given, the adhesive strain should be calculated with the modulus of
the adhesive, E
a
, and the joint should be designed in a way not to exceed the maximum allowable
strain. Secondly, if the strain is prescribed, the highest possible nominal modulus, E
u
, should be
used to calculate the stress. Now, the joint should be designed in a way that this stress does not
exceed the strength due to tensile stresses.
11.2.3 Analyses of Volkersen and Goland Reissner
Based on the following assumptions regarding the deformation pattern and the material
behaviour of the joint;
- Axial deformations in the adherends
- Shear deformations in the adhesive
- Linear elastic behaviour of adherends and adhesive
Volkersen derived the first analytical solution of the shear stress distribution within the adhesive
layer of a single-lap joint in his classical work [Volkersen, O., Luftfahrtforschung 15, 41 (1938)].
This analysis predicts a non-uniform stress distribution which is also observed from
experimental studies on lap shear joints. Following ideas from the books of Adams and
Wiedemann [R.D. Adams et al, Structural Adhesive Joints in Engineering, 2
nd
ed., London
(1997), J. Wiedemann; Leichtbau 2: Konstruktion; Berlin, (1996)] the shear stress distribution,
t(x), is divided by the mean (or nominal) adhesive shear stress, t = F/(bl), yielding the non-
dimensional relation
eq. 11-9
( ) ( ) ( ) ( )
( )
( )
( )
( )
( )
(

o +
o


=
o
t
=
t
=
t
t
2 cosh
l x sinh
1
1
2 sinh
l x cosh
2 s
l x
F
l b x x
1 10
,
o
10
is the tensile stress in adherend 1 according to the applied load F in shear direction. o and
are defined by
eq. 11-10
2 2
1 1
s E
s E

= o
and
eq. 11-11
( )
d s E
l G
1
1 1
2
2

o + =
and the coordinate x is restricted to the overlap region
eq. 11-12
2
l
x
2
l
s s .
The adherends have to be numbered in a way to ensure that the stiffness ratio of the adherends,
o, is less than unity. The maximum value of the shear stress, which occurs at one of the overlap
ends (at x = l/2), reads
eq. 11-13
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 186
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
(


o +
o
+

=
o
t
=
t
t
2
tanh
1
1
2
coth
2 s
l
1 10
max max
A serious drawback connected to the theory of Volkersen is that it neglects the eccentricity in the
applied load, which is caused by the non-symmetric geometry of single-lap joints. The
eccentricity serves to introduce bending moments in the adherends, which in turn leads to
bending deformations of the adherends. This effect was taken into account by the analysis of
Goland and Reissner [Goland, M., Reissner, E., J. Appl. Mech., Trans ASME, 66, A17 (1944)]
which is summarized and compared with similar theories by Adams et.al. [R.D. Adams et.al.,
Structural Adhesive Joints in Engineering, 2
nd
ed., London (1997)]. The bending moment factor,
k, is defined by the ratio of the real adherend bending moment, M, observed at the overlap end
and the bending moment that would have been observed at the overlap end if no adherend
bending had occurred. Thus, the bending moment at the overlap ends might be expressed by
eq. 11-14
2
s
F k M =
,
where Fs/2 estimates the adherend bending moment at the overlap end of a single-lap joint in its
original geometry. The bending moment factor is equal to unity for (infinitely) stiff adherends,
where bending of the adherends is negligible (upper sketch of Figure 11-2), while k < 1 for
adherends subjected to bending as in the lower sketch of Figure 11-2. In the analysis of Goland
and Reissner the adherends are assumed to be identical, which means that they are of the same
material and have the same thickness, s. Thus, the adherend stiffness ratio, o, equals unity.
According to Goland and Reissner the ratio of the maximum stress at the ends of the joint and
the mean stress (nominal value) is given by:
eq. 11-15
( ) ( ) | | k 3 3 coth k 3 1
4
1
max
+ + =
t
t
,
where the bending moment factor may be expressed as
eq. 11-16
( ) | |
1
2 2 tanh 2 2 1 k

u + =
.
Here, u is defined by
eq. 11-17
( )
3
2
s E b
1 F 3
l

v
= u ,
where v is the Poissons ratio of the adherend. The main purpose of the analytical solutions is to
calculate the maximum values of the shear stress at the overlap ends and to use these values in
local failure criteria for design of adhesive joints. Finally, it should be mentioned that although
there is great agreement on the needs for introducing the bending moment factor when analyzing
single-lap (or other non-symmetric) joints, several expressions for k exist in the literature, see
e.g. the book of Adams et.al. cited above.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 187
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 11-2 Single-lap joints and bending moment factor
11.2.4 Approach of Bigwood Crocombe
During the BONDSHIP project an approach of Bigwood and Crocombe [Bigwood, D.A.,
Crocombe A.D., Int. J. Adhesion and Adhesives, 9, 4 (1989)] has been investigated. In this
approach the overlap region is reduced to a simple adherend-adhesive sandwich, see Figure 11-3.
Using such a sandwich model it is possible to analyse any joint that can be simplified to this
form, and for which the end loading values can be calculated.
X
V21 T22 M22 V22 M21 T21
T12 M12 V12 V11 M11 T11
t 1
t 2 Adherend 2
Adherend 1
Y
Figure 11-3 Adherend-adhesive sandwich with applied moments and forces
It has been shown that numbers of adhesive joints, including single-lap joints, L-joints, T-joints
as well as certain double-lap joints, may be analysed by this theory, see Figure 11-4. The
presumed restrictions and the effects taken into account are as follows:
- 2D plane strain or plane stress analysis in the plane introduced by Figure 11-3.
- Adherends subjected to general tensile, shear and moment loading.
- Axial and rotational deformations as well as axial stresses of the adherends.
- Isotropic adherends of constant thickness. The adherends may be of different materials and
thicknesses.
- Axial stresses (in the through-thickness direction) and shear stresses in the adhesive. The
stress components are assumed to be constant through the thickness of the adhesive layer.
- Isotropic adhesive of constant thickness, which has to be much smaller than the thickness of
the adherends.
k~1
k <1
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 188
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
- Linear-elastic material behaviour.
However, Bigwood and Crocombe claim that there is no major conceptual difficulty connected
to taking material non-linearity into account by their analysis, but this has not been investigated.
Figure 11-4 Possible configurations for joints described with the approach of Bigwood and
Crocombe
Based on these restrictions (which are less severe than those introduced by Volkersen) and
effects Bigwood and Crocombe derived various differential equations for the stress distributions
in the adherends and the adhesive. Simple, closed-form solutions exist for the adherend stresses.
However, unless additional restrictions are introduced, closed-form analytical expressions do not
exist for the adhesive stresses. Therefore, in the present project the complicated formulas for the
adhesive stresses have implemented into a Microsoft Excel spreadsheet [Outline of approach to
easy-to-use design rules, 1-11-D-2001-01-0; Modelling benchmark study on rigid adhesives, 1-
11-W-2001-04-0]. This approach, using the spreadsheet solution as an easy-to-use method, has
been applied for, e.g., the preliminary design of application case 1 of Fincantieri [Determining
the hyperelastic material parameters by fit_poly and fit_ogden, 1-11-W-2002-01-0]
11.2.5 Analysis of Wiedemann pre-design, overlap length of joints
In addition to selecting a reasonable geometry and suitable materials for the joint based on the
general guidance offered above, estimates on several geometrical parameters, such as the overlap
length and the thickness of the adherends and the adhesive layer, should be provided by the pre-
design. This may be obtained by the approach of Wiedemann [J. Wiedemann; Leichtbau 2:
Konstruktion; Berlin, (1996)], which relies on the well-known stress analysis of Volkersen.
Figure 11-5 shows the ratio of maximum shear stress at the ends of the overlap region to the
(nominal) tensile stress in adherend 1 as a function of the parameter (eq. 11-10 and eq. 11-11),
which is characteristic for the specific joint,
eq. 11-18
( )
d s E
G
1 l
1 1

o + =
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 189
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
For 5, the ratio of the maximum shear stress to the applied tension becomes almost constant.
For further illustration, the stress distribution within the adhesive layer in the joint, normalised
with respect to the applied adherend tension stress, is depicted in Figure 11-6 as a function of the
parameter . In practical use, we can estimate a minimum overlap length, l
*
, by choosing = 5.
From eq. 11-18 we obtain
eq. 11-19
( ) o +

=
1 G
d s E
5 l
1 1 *
If no bending moment appears, as, for example, in a double lap joint, the joint with an overlap
length according to eq. 11-19 has the lowest level of stress: In the middle of the joint the stress
will not be significantly reduced by further increase of the overlap length, and the maximum
stress at the ends of the joint will be almost constant with increasing overlap length. Looking at
eq. 11-13 and considering:
eq. 11-20
1
2
coth =
|
.
|

\
|
and 1
2
tanh =
|
.
|

\
|
,
as becomes infinite, we obtain for the ratio of the maximum shear stress at the ends of the
joint, t
max
, to the mean shear stress, t, which is called the shear stress factor:
eq. 11-21
( ) o +

=
t
t
1
max
.
With the same material and the same thickness of the two adherends we have o = 1, which
implies that t
max
= 2.5 t for = 5 as a rough rule of thumb. With eq. 11-21, eq. 11-13 and eq.
11-20 a minimum value for the peak stress ratio to the applied tension, o
10
, can be calculated by:
eq. 11-22
( ) ( ) o +

=
o +

=
o
t
1 d E
s G
1 l
s
1
1 1
10
max
The indices have to be selected in relationship to the adherends to ensure that os 1. In this
equation t
max
is the maximum shear stress at the ends of the overlap length and o
10
is the tensile
stress in adherend 1 according to the applied load F in shear direction. The ratio t
max
/o
10
increases with the thickness of the adherends and is independent of the overlap length, if no
bending occurs. A lightweight structure requires adherends as thin as possible which just can
carry the loading, including some safety factors. If a bending moment cannot be avoided the
maximum shear stress at the ends of the joint can be estimated with eq. 11-15 - eq. 11-17 and the
overlap length should be larger than given by eq. 14. Considering a lap-shear sample with either
aluminium (E = 73 GPa) or steel (E = 215 GPa), the rigid adhesive according to Table 5-1, with
an adhesive thickness of d = 0.3 mm, we can use eq. 11-19 with s
2
/s
1
>1 to calculate l
*
as a
function of s
1
(Figure 11-7). l
*
increases with the square root of the stiffness of the adherend, (E
1
s
1
). The minimum overlap length for steel is about 2.9 = 215/73 times larger than for aluminium,
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 190
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
if the thickness of the adherends is the same. The reason for this is that the principle of
corresponding deformation (Figure 5-7) is fulfilled closer in aluminium than in steel. The
minimum overlap length increases with increasing difference of thickness of the adherends.
Figure 11-5 t
max
/o
10
as a function of the parameter , o = 1

Figure 11-6 Ratio of shear stress distribution to the applied tensile stress according to
Volkersen as a function of the overlap length
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 191
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 11-7: Minimum overlap length as a function of the thickness of the adherends, d=0.3
mm
Figure 11-8: Tensile stress in the adherends vs. thickness of the adherends, d= 0.3 mm
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 192
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The load capacity, o
10
/t
F
, of the joint can be determined as a function of the thickness of the
adherends, using eq. 11-22. t
F
is the maximum shear stress at the applied load F. For the load
capacity of the joint the following relation can be developed from the last equation
eq. 11-23
( )
( )
1
1
1
*
max
10
F
10
s G
1 d E
s 5
l
1

o +
=

o + =
t
o
s
t
o
The load capacity o
10
/t
F
is proportional to the overlap length (eq. 11-23). The dependency
between the tensile stress in adherend 1 and the thickness of the adherends can be seen in Figure
11-8. Figure 11-7 and Figure 11-8 can be used to estimate the thickness of the adherends and the
overlap length for a joint with an applied load where the adherends are just loaded to the yield
point: Taking, for example, the steel QSte380 TM with a yield strength of about R
p0.2
= 380
MPa, which has been used as material for the sandwich plates of JLM AC4, the joint with
maximum load capacity can be realised with an adherend thickness of about 12 mm and an
overlap length of about 120 mm. If the thickness of the adherends is lower or the overlap length
is larger the adherends will fail and not the adhesive joint.
In order to take account for the effect of the bending moment, regarding also the reduction of the
peak stress at the ends of the joint due to plastic yield of the adhesive, the overlap length should
be taken to be
eq. 11-24
* * *
l 2 l > .
Then the full load capacity of the joint can be used. If a symmetrical double lap shear joint is
used, the recommended overlap length is half the overlap length of the single lap joint. For
further optimising the design, the spread sheet described above should be applied.
The equations above cannot be used for thick adhesive layers. According to the basic equations
of Volkersen, Goland, Reissner the stresses in the adhesive layer decrease with increasing
thickness of the adhesive layer. However, the experimentally measured lap shear strength
decreases with increasing thickness of the adhesive layer, which is apparently in contradiction to
what should be expected from the analytical equations. Therefore, at a thickness of the adhesive
layer above 0.5 mm, knock down factors have to be determined experimentally, which take into
account that the load capacity of an adhesive bonded joint decreases with increasing thickness of
the adhesive layer. No analytical solutions exist to describe the dependency of the stress
distribution on the thickness of adhesive layer.
As an example we consider a single lap shear sample with following characteristic parameters:
adherends steel QStE380 TM, shot blasted with corundum, overlap length 13.05 mm, width
24.95 mm, thickness of adherends 2.53 mm, thickness of adhesive layer 0.35 mm, measured lap
shear strength 38.41 MPa (three samples) which corresponds to F = 12480 N, test temperature
20C, rigid adhesive according to Table 5-1. Figure 11-9 shows the shear stress distribution
according to Volkersen for three different overlap lengths. The first overlap length is 13 mm as
in the experiment. The second overlap length is 55 mm as calculated from eq. 11-19 (with = 5)
and the third overlap length is 110 mm as required from eq. 11-24. The maximum stress
decreases from 42 MPa for the short overlap length to about 21 MPa in the two other cases. It
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 193
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
can be seen that the maximum shear stress is not reduced by increasing the overlap length from
55 to 110 mm as predicted for > 5. However, the mean stress decreases from 9.7 to 5.4 MPa in
the last case. Since failure occurs at the ends of the joint, it can be assumed that further increase
of the overlap length will not influence the failure behaviour of the joint.
Figure 11-9: Shear stress distribution at F = 12480 N according to Volkersen
Figure 11-10: Shear stress distribution at F = 12480 N according to Bigwood and
Crocombe
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 194
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 11-11: Shear stress and peel stress distribution at F = 12480 N according to Bigwood
and Crocombe
Figure 11-10 shows the shear stress distribution according to the solution of Bigwood and
Crocombe, which was calculated with the spreadsheet. The mean shear stress is the same in the
Volkersen and the Bigwood and Crocombe approach. Compared to the solution of Volkersen
(Figure 11-9), the stress gradient and stress level is higher at the ends of the joint, while the
stress level at x = 0 is lower in the Bigwood and Crocombe solution. The maximum stress
decreases from 51 MPa for the short overlap length to 40 MPa for the larger overlap lengths. The
maximum stress, again, is the same for the large two overlap lengths. The same can be observed
for the peel stress which can be calculated according to Bigwood and Crocombe. Figure 11-11
shows the peel stress compared to the shear stress. The maximum of the peel stress is higher than
the maximum shear stress and decreases from 60 MPa at the short overlap length to 56 MPa at
the larger overlap lengths. The maximum peel stress is about three times as high as the maximum
shear stress at the ends of the joint for the Volkersen solution. Therefore, we should change our
rule of thumb for the maximum stress to be o
max
= 7.5 t
mean
. As the result of the comparison
between Volkersen and Bigwood & Crocombe, the concept for calculating the overlap length,
which was described above, holds for both approaches. It is, therefore, recommended to estimate
first the overlap length for a joint loaded in shear with eq. 11-19 and eq. 11-24 before further
optimising the design with other methods.
With regard to a failure criterion the analytical formulas should be applied to an experiment with
a lap shear sample. The maximum values of the shear stress and the peel stress should be
calculated for an applied load, which corresponds to the strength of the joint. Using a relatively
short overlap length to determine the failure criterion yields a conservative approach.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 195
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
While Volkersens method yields a failure criterion 42 MPa maximum shear stress, Bigwood
and Crocombe yields a shear stress of 51 MPa and a peel stress of 60 MPa. Therefore, these
values should only be used within the own framework of the two theoretical approaches.
Furthermore, the thickness of the adhesive layer should be the same as in the experiment from
which the failure criterion was developed and should not be changed during any optimisation
loop and during any design of application cases for which the failure criterion has to be applied.
The reason for this is that any change of the thickness, d, of the adhesive layer influences the
calculation of the maximum stresses and the failure criterion as it has been defined here. As
already mentioned the experimental strength of a lap joint decreases with increasing thickness of
the adhesive layer, while the calculated stress peaks decrease with increasing thickness of the
adhesive layer. Since we can assume that experimental failure occurs at the same local stress
level the calculated behaviour is in contradiction to the experimental observation.
Increasing the applied load at an overlap length of 110 mm by 30% yields a maximum shear
stress of 51 MPa and a maximum peel stress of 73 MPa, which is higher than the estimated
failure stress (51 and 60 MPa); 20% yield 47 and 68 MPa, 10% yield 44 and 62 MPa,
respectively. Hence, increasing the overlap length by 30% increases the load capacity only by
20%.
Figure 11-12: Decreasing stress peaks at s
1
> 0.8 mm and F=12480 N
With an applied shear load of 12480 N, the tensile stress inside the adherends is 197 MPa. This
is significantly lower than the yield strength of the steel, which is 380 MPa. Reducing the
thickness of the adherends to 0.8 mm, the tensile stress in the adherends is approximately the
yield strength of the steel. Figure 11-12 shows the shear stress and peel stress peaks as a function
of the thickness of the adherends at s
1
> 0.8 mm, for various situations calculated with the
Bigwood & Crocombe spreadsheet. Both the peel stress and the shear stress are plotted for an
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 196
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
overlap length of 110 mm and 13 mm. While the peel stress is almost the same for both overlap
lengths, the shear stress decreases faster, and to a lower value, with increasing thickness of the
adherends for the overlap length of 110 mm. One curve shows the shear stress calculated with M
= 0. In this case the shear stress is lower than in the corresponding situation, where a moment,
M, has been included into the calculation. This curve is almost identical to the Volkersen
solution calculated with eq. 11-13. The maximum shear stress without considering the bending
moment is significantly lower than the stress, which appears with bending moment. Hence, the
Volkersen approach underestimates the real stress situation in the single lap shear samples,
significantly. Only some rules of thumbs for a first approach do an overlap length and to the
maximum stress at the ends of a rigid joint can be used from the Volkersen analysis. In a later
section, the solution of Bigwood and Crocombe will be verified with finite element calculations.
11.3 Numerical analysis methods the finite element method
During the past four decades, the finite element method has become a very popular technique for
the computer solution of complex problems in engineering [K.-J. Bathe, Finite Element
Procedures in Engineering Analysis, London (1982)]. In finite element modelling the structure is
divided into small elements of any shape. Therefore, the finite element method is considerably
more suitable for problems of complex geometry than, e.g., the finite difference method, which
relies on several restrictions with respect to the regularity of the problem domain. Moreover, the
implementation of boundary conditions is more general in the finite element method than the
finite difference method.
The finite element method is well suited for analysis of all kinds of structures. This is mainly due
to the flexible nature of the method. In addition to the possibility of performing analyses on any
geometry and with a number of applied boundary conditions, a wide variety of element types are
available. Thus, when studying the main or global behaviour of large structures, for example a
ship, one may apply a high, but limited, number of so-called structural elements, such as beam or
shell elements, instead of using an enormous number of general solid elements. Thereafter, local
analyses of small, critical parts of the structure might be performed using a fine mesh of solid
elements. In this way a large amount of valuable CPU-time is saved, and the applicability of the
method to analyse structures of real interest is increased considerably.
Generally, two main types of solid elements are available. So-called h-elements, which are
widely used, need a relatively fine mesh. The mathematical functions used to solve a problem are
usually polynomials of the order of one or two, while the ratio of one side of the element to
another side should typically not exceed three. In order to obtain convergence of the analysis
results, relatively fine meshes might be necessary (Figure 11-13). The second main element type
covers the so-called p-elements. These elements use polynomials up to a much higher order, and
the aspect ratio of the elements can be significantly larger. Therefore, convergence of analysis
results might be obtained even on coarse meshes (Figure 11-14). The programs that use h-
elements are widely spread, while the p-element types have become commercially available only
in the last decade.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 197
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 11-13: A lap shear sample for local analysis modelled with h-elements (2D)
11.3.1 Linear finite element analysis
The simplest way of using the finite element method is to apply a linear analysis. Then, the
behaviour of the material is typically assumed to be linear elastic. In addition, if the material is
isotropic, only two elasticity coefficients are required to describe the material behaviour
completely. Moreover, the analysis is performed with respect to the original or un-deformed
geometry of the structure. The maximum load or the entire design load may be applied directly.
Thus, for static (or quasi-static) problems the solution may be obtained in a single main step.
11.3.2 Non-linear finite element analysis
Very few materials offer pure linear-elastic behaviour until failure. In most cases considerable
material non-linearity will arise far below ultimate load limits. For example, the behaviour of
flexible adhesives is often well described by (strain-hardening) hyperelastic material models,
while most rigid adhesives possess some kinds of elastic-plastic behaviour. As a result,
additional material coefficients, some of which might be hard to determine, are required when
materially non-linear effects are taken into account. In addition to the increased number of
material parameters needed, the complexity of the main solution procedure increases
considerably. The entire maximum or design load can not be applied directly. On the other hand,
the load level must be increased in a step-wise manner. For each load step, incremental nodal
displacement values are calculated based on the actual load level. Also the material model
describing, e.g., the stiffness of the material accounts for the present state of stresses and strains.
Fortunately, major parts of these complicating factors are hidden for the users of most of the
commercial finite element software packages available. For users the increased complexity is
roughly limited to picking a suitable material model (and eventually a yield criterion) including
numbers of material parameters as well as defining a proper number of load steps and choosing
an iteration method, for example the Newton Raphson method, for the non-linear problem of
consideration. However, it might happen that the computer code does not contain proper models
for the elastic behaviour or yield criterion of the material. In such cases, one may have to
implement particular models into the software package used. This might be a very complicated,
or even impossible, task.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 198
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 11-14: A lap shear sample for local analysis modelled with p-elements (3D)
In addition to the materially non-linear effects discussed above, several geometrically non-linear
effects may be taken into account. These may be due to large displacements and/or large strains.
The former case may appear in combination with small or large strains. For example, although
the strains within rigid adhesive joints might be small, significant deflections due to the
eccentricity in the load path (or due to the non-symmetric geometry) of single-lap joints are
usually presented for loads far below ultimate limits. This effect plays a crucial role on the
overall joint stiffness, and it should therefore be included in the analysis. In other cases the
strains in critical locations of the structure may be large. Then the simple strains used in linear
analyses are not valid. Instead more generally applicable strain measures, for example Green-
Lagrange or logarithmic strains, should be included. In computer codes such effects are usually
taken into account by imposing the loads (or prescribed displacements) in a step-wise manner,
updating the geometry for each load step, and prescribing a suitable strain measure.
To conclude this sub-section containing general aspects it can be stated that the finite element
method is well suited for analysing a wide variety of engineering structural problems in general
and adhesively bonded joints in particular. However, the following items regarding the
modelling of adhesive joints should be remarked:
- The width and the length of adhesive joints are usually large compared to the thickness of the
joint. If local analysis of strains and stresses has to be accomplished, brick-elements are
necessary. If a joint is modelled with, for example, three layers of brick elements over the
thickness d, a high number of elements is necessary to model an extended joint. The finite
element models, at least with h-elements, become quite large.
- Adhesives are polymers. While the constitutive equations, which are necessary to describe
metals are implemented in the finite element codes, some special features of polymer
materials are sometimes not available. One example are hyperelastic material models, which
are necessary to describe the local behaviour of flexible adhesive joints. A second example is
the equivalent stress. In most programs the von Mises equivalent stress is implemented to
describe plasticity. In many cases the von Mises equivalent stress is used for the design of
structures. The von Mises criterion is not adapted to materials which depend on both,
hydrostatic and deviatoric stress components. Plastic deformation of metals depends, in most
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 199
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
cases, only on the deviatoric part, while the plastic flow of polymers depends on both
components.
- Visco-elastic material behaviour of polymers, which influences both the measurement of
characteristic values and the modelling of the structural behaviour, yields further
uncertainties in the design process.
11.3.3 Finite element analysis for flexible adhesive joints
The first approach of numerical analysis of flexible adhesive joints uses a linear-elastic material
model for the adhesive. With increasing complexity of the local structure and the applied
loading, the flexible adhesive layer is presented in the finite element analysis by 3 dimensional
solid elements using hyperelastic material models. The adhesive material is then described as a
rubber-like, nearly incompressible, and isotropic hyperelastic material characterised by a strain
energy potential, U(c), which defines the strain energy stored in the material per unit of
reference volume (volume in the initial configuration) as a function of the strain at that point in
the material. There are several forms of strain energy potentials available in many commercial
finite element codes to model incompressible, isotropic elastomers.
First we consider an exponential strain energy function as implemented in the StressCheck
program. In this FE-code the strain energy potential is written as
eq. 11-25
L
CW 2
L
e W W =
,
where
eq. 11-26
} ]{ E [ } {
2
1
W
T
L
c c =
and
eq. 11-27
( ) } ]{ E [ e CW 2 1
} {
W
} {
L
CW 2
L
c + =
c c
c
= o
The parameter C can be determined from a true stress vs. true strain relation of a tensile test with
a sample made out of adhesive. Figure 11-15 shows the experimental data and the fitted function
for a flexible adhesive (Terostat 8590 UHV/M). Up to about 30% true tensile strain the linear
elastic approach (green line) yields sufficiently good results. At higher strains the discrepancy
between linear and true behaviour becomes significant. Even if joints with flexible adhesives
should not be designed to reach strains above 10-30%, depending on the particular adhesive,
methods used in the design process should allow to give accurate results up to 80% strain or
more to identify hot spots in the joints.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 200
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 11-15: Nominal and true stress-strain of a flexible adhesive
Further examples are the Ogden form and various polynomial forms which are used in
ABAQUS. The form of the Ogden strain energy potential is
eq. 11-28
( )
_
=
o o o
+ +
o

=
N
1 i
3 2 1 2
i
i
3
2
U
i i i
,
where
i
are the principal stretches: the ratios of current length to length in the original
configuration in the principal directions. They are related to the principal nominal strains, c
i
,
by
eq. 11-29
i i
c + = 1 , i=1,2,3.

i
and o
i
are material parameters that may be determined from adhesive material test data. The
form of the reduced polynomial strain energy potential is
eq. 11-30
( )
_
=
=
N
1 i
i
1 0 i
3 I C U ,
where C
i0
are material parameters, and I
1
is the first deviatoric strain invariant defined as
eq. 11-31
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 201
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
2
3
2
2
2
1 1
I + + = .
The form of the neo-Hookean strain energy potential is
eq. 11-32
( ) 3 I C U
1 10
=
which can be obtained as a special case from the reduced polynomial function for N=1.
Figure 11-16: Ogden potential strain energy function to describe hyperelastic, flexible
adhesive Sikaflex 360 HC from first benchmark
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 202
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Displacement [mm]
0 2 4 6 8 10 12
F
o
r
c
e

[
N
]
0
200
400
600
800
1000
Experimental
CETENA: Mooney3
CETENA: 3. order
CETENA: Ogden
UoS, linear
UoS, nonlinear
UoS, nonlinear
UoS, hyperelastic
IFAM, Ogden, N=3
IFAM, Polyn., N=2
IFAM, Yeoh
IFAM, Neo-Hookean
Alusuisse
Numerical Results: Conclusion
Figure 11-17: Potential strain energy functions to describe hyperelastic, flexible adhesive
Sikaflex 360 HC from first benchmark
As already mentioned, various types of potential strain energy functions are available. It depends
on the type of adhesive, which function yields the best results. For the adhesive Sikaflex 360HC,
which was used for the first benchmark study [IFAM et al., Benchmark Test Problem for finite
element analysis1-11-W-2001-03-0], the Mooney and Ogden strain energy potential seem to be
yielding the best results (see Figure 11-16 and Figure 11-17). However, it is important that at
least two different types of experiments such as tension/compression and shear are used to
determine the free parameters (eq. 11-28 for the Ogden formulation). More details can be found
in [IFAM, Determining the hyperelastic material parameters by fit_poly and fit_ogden, 1-11-W-
2002-01-0, IFAM et al., Benchmark Test Problem for finite element analysis1-11-W-2001-03-0].
11.3.4 Finite element analysis for rigid adhesive joints
Since the design methods worked out in the BONDSHIP project will be used on ship-yards, the
focus of the discussion is put more on applicability than on scientific completeness. In most of
the examples, which are discussed here, the finite element program StressCheck was used. This
program was developed in the last years. It is working on PC-basis and uses p-elements up to the
order of eight. Due to the p-elements local modelling of adhesive joints can be done in relatively
short time. The program does not have capabilities to implement own material laws. Modelling
of hyperelastic materials with potential functions is only possible with one type of hyperelastic
potential function. In cases where more modelling features had been necessary the codes
ABAQUS and MARC were available during the project. In StressCheck the plastic behaviour of
materials is described with the von Mises approach. This situation can be found for the most
finite element codes, which are spread in the shipyard industry. Furthermore, in most cases only
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 203
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
linear-elastic modelling is common practice. From the point of view of fast modelling, of linear-
elastic approaches, and von Mises plasticity, the application of StressCheck has some practical
aspects with regard to finite element analysis in shipbuilding industry. The modelling of a lap-
shear sample will be used to show the limits of finite-element modelling, if von Mises plasticity
is implemented. The behaviour of the adhesive was measured with a tensile test [IFAM et al.,
Final report on coupon tests for characterisation of adhesives and surface coatings, 1-12-W-
2001-05-0] (Figure 11-18). The true-stress is evaluated as a function of true-strain and can be
described as input into the finite element program (Figure 11-19). The non linear material model
for the code uses five parameters and Poissons ratio. The model fits to the experimental curve.
The adherends can be modelled as elastic-plastic material with E-modulus, Poissons ratio and
yield strength. For QStE 380 TM the modulus was measured in a standard tensile test to be 255
GPa (the steel plates have a texture), the yield strength is 380 MPa and Poissons ratio was
assumed to be 0.33.
Figure 11-20 shows the numerical and experimental data for the lap-shear test. The displacement
was evaluated from the finite element model at the same positions outside the overlap where a
strain gauge was applied in the experiment. The linear-elastic solution is a reasonable
approximation only to the first part of the curve. The non-linear solution shows very good
agreement with the experimental results up to the calculated yield point of 23 MPa, where the
overall plastic deformation of the simulated behaviour starts. The experimental macroscopic
yield point is reached above 30 MPa, which is significantly higher than 23 MPa. The slope of the
plastic region is also higher in the numerical analysis than in the measurement. The failure
cannot be predicted a priori from this model. The same results have been obtained with
ABAQUS. Again, the calculated macroscopic yield point is reached before the experimental
yield point and the slope in the plastic region is too high.
StressCheck allows an estimation of the convergence of the analysis. For increasing number of
degrees of freedom, which corresponds to an increase in the polynomial order of the p-elements,
the error can be evaluated. The error bars in Figure 11-20 represent the error calculated by the
program at the end of the non-linear solution procedure, but before the solver failed due to large
deformations in certain elements.
Figure 11-18: True-stress true-strain behaviour of Vantico Araldite 420 at room-
temperature
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 204
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 11-19: Description of the material model as input to the finite element code
Similar calculations have been done in [CTA, Modelling benchmark study on rigid adhesives, 1-
11-W-2001-04-0]. These results predict the yield point approximately at the correct position, but
for only one overlap-length. The adhesive was simulated with an elastic-plastic material model
with the yield strength of 32.45 MPa which is about 13% higher than the yield strength
according to the experimental results (28.8 MPa) which are presented here (Figure 11-18).
Figure 11-20: Numerical simulation and experimental data of lap shear test
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 205
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
The von Mises equivalent stress is plotted in Figure 11-21. For the linear-elastic solution the von
Mises equivalent stress at the experimental shear strength of the joint shows peaks up to
90 MPa at the ends of the joint. In the non-linear elastic plastic simulation the von Mises stress
has a level of 28.8 MPa from the left to the right end of the joint. At an applied load of 15.6 MPa
large regions of the adhesive layer are already plastically deformed. The von Mises stress
reaches 28.8 MPa at the outer ends of the joints and remains constant at further loading. The
regions of this stress level move from the ends of the joint to the middle part of the joint until the
stress limit is reached through the entire joint.
In conclusion, the finite element method can be used for analysis of rigid adhesive joints. As we
have seen, linear-elastic analyses agree well with experiments for low levels of applied loads. In
addition, conservative, yet not very accurate, results may be obtained by linear methods for
higher load levels. Moreover, the predictions obtained by finite element methods taking
materially and geometrically non-linear effects into account are of higher quality than the results
offered by linear methods. However, more attention should be paid on developing models
describing the non-linear behaviour of polymeric materials.
Figure 11-21: Von Mises equivalent stress
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 206
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.4 Non-destructive inspection
11.4.1 Glossary
Acceptance criteria Size of allowable defects for a given object or component
Access Ease of access to enable to place sensor onto structure
Adherend Substrate of adhesive joint
Adhesion detection Inspection of interface between adherend and adhesive
A-scan Primary display mode for NDT equipment showing sensor
response as a function of time
Attenuation Loss of NDT radiation as it travels through test sample
Bond General description of an adhesive joint
Bondline General description for adhesive layer in an adhesive joint
B-scan Enhanced display mode for NDT equipment showing a
series of A-scans as a function of sensor position
BVID Barley visible impact damage (defect type)
Calibration Setting up of NDT equipment to ensure correct operation
Call Phrase referring to detection of suspected defect
Certification Formal training (EN, ISO) for NDT operators
Colourmap Colour guide for C-scans relating colour to NDT parameter
such as defect depth
Component Structure or body to be inspect (see object)
Compression wave Ultrasound wave with compressive (dilatational)
displacement
Coupling Method by which sensor transmits energy into component
C-scan Enhanced display mode for NDT equipment showing sensor
response at a particular time as a function of sensor x, y
position
Cure monitoring Use of NDT to track progress of cure (conversion from
liquid to solid state)
Defect Discontinuity or structural anomaly
Defect size Size parameter for defect (e.g. volume, area, depth)
Delamination Interfacial or planar defect often described for adhesive
bonds or layered composites (e.g. carbon fibre components)
Disbond Lack of adhesion between adhesive and adherend
False call Defect measurement recorded when no defect present
Flaw See defect
Flaw detector General name give to instrument for NDT work
Guided wave Ultrasonic wave that travels in plate like structures
Kissing bond Varying definitions but in general describes bonded joint
with zero shear stiffness
NDE Non destructive examination
NDT Non destructive testing
NDT procedure Formal written method for carrying out an NDT inspection
NDT technique Specific NDT method or measurement used for inspection
Object Refers to the test sample or structure
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 207
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Operator Refers to NDT inspector or operator
Pitch catch Ultrasonic inspection using separate transmitter and receiver
probes on same side of test sample.
Probability
of detection
Statistical analysis of ability of NDT technique to detect
defects of certain sizes
Probe Device for transmitting and/or receiving ultrasound or other
radiation
Pulse echo Ultrasonic inspection where the same probe is used as
transmitter and receiver
Reference block Sample used to check calibration and measurements made by
NDT equipment
Sensor See probe
Shear wave Ultrasound wave with shear (tangential) displacement
Through
transmission
Ultrasonic inspection using separate transmitter and receiver
probes on either side of test sample
Time of flight Measurement of arrival time of NDT signal, often used to
calculate thickness of sample with ultrasound.
UT Shorthand for ultrasound testing
Void Air filled cavity, larger than pore or bubble
Zero volume
disbond
See kissing bond
11.4.2 Standards
The standards listed here are based on existing British standards (BS) many of which have been
replaced by European standards (CEN) and are marked (BS-EN). CEN standards which have
reached the provisional stage (prEN), and have been numbered, are also listed for completeness.
Where possible it is recommended that inspection techniques selected for bond inspection are
based on existing or related standards. General and technique specific standards are given
below:
11.4.2.1 General standards
BS-EN-473:2000
Non-destructive testing Qualification and certification of NDT personnel General principles
BS-EN-ISO-9000-9004
Quality management and quality assurance standards
BS-EN-1330
Terminology
Part 1. General terms
Part 2. Common terms for NDT methods
11.4.2.2 Radiological methods
BS-EN-1330-3:1997
Terminology. Terms used in industrial radiographic testing
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 208
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
BS-EN-444:1994
NDT General principles for radiographic examination of metallic materials by X- and gamma-
rays
11.4.2.3 Ultrasonic methods
BS-EN-1330:4:1999
Terminology Part 4. Terms used in ultrasonic testing
BS-EN-583:1998/2001
Ultrasonic examination
Part 1. General principles
Part 2. Ultrasonic examination sensitivity and range setting
Part 3. Transmission technique
Part 4. Examination for perpendicular defects
Part 5. Characterisation and sizing
Part 6. Time of flight diffraction technique as a method of detection and sizing of discontinuities
prEN-14127
Ultrasonic examination, thickness measurement
11.4.2.4 Visual inspection methods
BS-EN-13018:2001
Visual testing general principles
BS-EN-1330:2000
Part 10. Visual inspection, terminology. Terms used in visual inspection
11.4.3 Inspections and operators
NDT inspections will often form part of a quality assurance system for production and in-service
applications and it may be required that the company carrying out the inspection have to
demonstrate that their NDT engineers and technicians have the necessary level of knowledge and
skill. This is particularly important since NDT and inspection activities are very operator
dependent. In the case of adhesive bond inspection there are few established standards in place
and thus reliance must be placed on the skill of the operator and advice sought from practitioners
and equipment manufacturers. Most European countries operate an NDT operator certification
scheme which covers basic certification to general standards with specialist applications for
particular inspections. In the UK a worldwide recognised certification scheme called PCN is
operated which complies with the standard BS-EN-473:2000. Generally, most certification
schemes offer three levels of certification (after Lavender, 2003):
NDT Level 1:
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 209
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
An NDT Level 1 individual should:
i. be qualified to properly perform specific calibrations;
ii. be qualified to carry out specific NDT and specific evaluations for acceptance or
rejection determinations according to written instructions and to record results;
iii. receive the necessary instruction or supervision from a certified NDT Level 2 or 3
individual.
NDT Level 2:
An NDT Level 2 individual should:
i. be qualified to set up and calibrate equipment;
ii. be qualified to interpret and evaluate results with respect to applicable codes,
standards, and specifications;
iii. be thoroughly familiar with the scope and limitations of the methods for which
qualified and should exercise assigned responsibility for on-the-job training and
guidance of trainees and NDT Level 1 personnel;
iv. be able to organise and report the results of NDT.
NDT Level 3
An NDT Level 3 individual should:
i. be qualified to establish techniques and procedures;
ii. be qualified to interpret codes, standards, specifications, and procedures;
iii. be qualified to designate the particular NDT methods, techniques and procedures to
be used;
iv. to be responsible for the NDT operations for which qualified and assigned and should
be capable of interpreting and evaluating results in terms of existing codes, standards,
and specifications;
v. have sufficient practical background in applicable materials, fabrication, and product
technology to establish techniques and to assist in establishing acceptance criteria
where non are otherwise available;
vi. have general familiarity with other appropriate NDT methods, as demonstrated by the
Level 3 Basic examination or other means;
vii. in the methods for which certified, be capable of training and examining BNDT Level
1 and 2 personnel for certification in those methods.
In the case of adhesive bonding where there are few established standards, it is recommended
that manufacturer ensure that the personnel or companies carrying out the inspection work have
suitable experience equivalent to Level 2 and, where required, Level 3 NDT.
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 210
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.4.4 List of equipment manufacturers
11.4.4.1 Acoustic methods
Advanced NDT Instruments www.advanced-ndt.co.uk
NDT Equipment Services www.ndtequipment.co.uk
Peak NDT www.peakndt.com
RTD www.rtd.nl
Sonatest www.sonatest.com
11.4.4.2 Radiography
Agfa NDT www.agfandt.com
Fidgeon www.fidgeon.co.uk
RTD www.rtd.nl
X-Tek Industrial www.xtekxray.com
11.4.4.3 Shearography
Computerised Information Technology www.cituk.com
www.laserndt.com
11.4.4.4 Thermography
FLIR Systems www.flir.com
L.O.T. Oriel www.lotoriel.co.uk
www.laserndt.com
11.4.4.5 Ultrasonics
Advanced NDT Instruments www.advanced-ndt.co.uk
Agfa NDT www.agfandt.com
GE Panametrics www.panametrics.com
NDT Solutions Ltd www.ndtsolutions.com
Phoenix Inspection Systems www.phoenixisl.co.uk
Physical Acoustics www.pacuk.co.uk
R/D Tech www.rd-tech.com
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 211
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
RTD www.rtd.nl
Sonatest www.sonatest.com
www.staveleyndt.com
Ultrasonic Sciences www.ultrasonic-sciences.co.uk
11.4.4.6 Visual
Olympus Industrial www.olmypusindustrial.com
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 212
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
11.4.5 NDT test procedure - rigid adhesive bondline measurement
Relevant standards
Equipment
Calibration and reference block
Component and location
Measurement technique
Acceptance criteria
Reporting
<< name of company >>
Non-destructive testing manual
Inspection procedure for bonded balcony fixture
1. Description
A. This general ultrasonic inspection procedure is for the inspection
of the bonded regions for the balcony support component.
B. The inspection is carried out on the accessible external bonding
plate which is not obstructed by weld lines or heavy corrosion.
C. The procedure is in one part:
General ultrasonic inspection for bondline thickness measurement.
1. Component or area to be inspected (Refer to figure 1)
A. A 40 mm wide area around the complete outside edge of the
bonding plate.
2. Description of possible indications (Refer to figure 1)
A. Lack of adhesive material on edge of bonding plate
B. Large areas of delamination
C. Thinning or thickening of adhesive
3. Related documents
A. EN-1330-444:1994
B. EN-583-1998/2001
Part 1. General principles
Part 2. Ultrasonic examination sensitivity and range
settings
4. Equipment and materials
(1) Instrument: Masterscan 340 digital flaw detector from
Sonatest Plc
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 213
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
(2) Search unit (refer to figure 2)
(a) Wedge: 0 degree replaceable delay line, part
no. TMP3-DL from Sonatest Plc
(b) Crystal: 10 MHz, 6.35 mm diameter transducer
crystal, part no. TMP3-10
(3) Couplant
(a) Crystal to wedge: DL-gel from Sonatest Plc
(b) Wedge to inspection surface: Thixotropic gel
Note: Any comparable ultrasonic equipment may be used
provided it satisfies the requirements of this
procedure and is capable of resolving the reference
indications in the reference standard at the required
level of resolution.
(4) Reference standard: MW-BI-3200 (refer to figure 3).
Figure 1: Inspection area.
Bondline
Inspection area
Bonding plate
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 214
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
Figure 2: Search unit.
Figure 3: Reference standard MW-BI-3200 and typical data from
position 2 and 3.
5. Preparation for inspection
Position 1 Position 2 Position 3
6.35 mm crystal
Wedge 0 degree
Search unit
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 215
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
A. Make sure that the following preparations are made.
(1) Paint (excluding primer), if applied, is removed from the
inspection surface for a distance of 40 mm from the outside
edge of the bonding plate.
(2) Heavy corrosion, if any, is removed from the inspection
surface for a distance of 40 mm from the outside edge of the
bonding plate (refer to figure 1).
B. Check the inspection area for any visible damage or
discontinuities.
6. Instrument calibration
Note: This procedure is based on isolation and gating of the
adhesive bondline echo. The inspector must have a good
understanding of the ultrasonic beam paths excited in the
structure.
A. Set the FREQUENCY (10 MHz)
B. set the GAIN to 90 dB. Set the RANGE to 6 mm for compression
waves in adhesive (2500 metres per second), and set the DELAY to
11 mm and locate the wedge echo. Confirm location by finger
damping.
C. Familiarisation (refer to figure 3)
(1) Couple the search unit to reference standard MW-BI-3200 at
position 1, as shown in figure 3, and identify the backwall
echo in the bonding plate. Confirm location of echo by finger
damping.
(2) Couple the search unit at position 2 in figure 3 and
identify the adhesive bondline echo. Confirm location by
finger damping.
(3) Couple the search unit at position 3 in figure 3 and
identify the adhesive bondline echo.
D. Calibration
(1) Couple the search unit to the reference standard at
position 3. Set the instrument GATE1 START to 14 mm, WIDTH 2.4
mm and THRESHOLD 30 %. Set the instrument GATE2 START to 15.5
mm, WIDTH 3 mm and THRESHOLD 20 %. Adjust the instrument GAIN
until the bonding plate backwall echo reads 100 % FSH. Set
the instrument measurement to G1-G2, enable HUD and check the
thickness reading from POSITION 3 is 2 mm.
(2) Couple the search unit to position 2 and check that the
thickness reading from POSITION 2 is 1 mm.
Note: The timebase positions obtained during calibration are
for guidance only. Signals from the balcony bonding plate
DET NORSKE VERITAS
Report No: 2004-0193, rev. 2
TECHNICAL REPORT
Page 216
Reference to part of this report which may lead to misinterpretation is not permissible.
Rev2_RPs_DNVRepNo_2004-0193.doc
structure may appear at other timebase positions, depending on
local geometry.
7. Inspection procedure
A. Calibrate the equipment in accordance with paragraph 6.
B. Inspect all areas of the bonding plate using lateral scanning
movements to ensure complete coverage
Note1: Loss of bonding material is indicated by a loss of
signal in GATE2
Note2: Excessive thinning or thickening of the adhesive may
cause GATE2 not to trigger
8. Acceptance criteria
A. All signal reading an adhesive thickness greater than 2 mm and
less than 1 mm must be recorded on the structure with a
description stating:
(1) The balcony number, measurement position referred to top
left corner and bondline thickness.
9. Final NDT requirements
A. Remove all couplant and any markings from the inspection area.