Vous êtes sur la page 1sur 8

/@@&

OTC 7799

Strength and Stiffness of Tubular Joints for Assessment/Design


Purposes
A.F. Dier and M. Lalani, MSL Engineering Ltd.

Copyright 1995, Offshore Technology Conference

This paper was presented at the 27th Annual OTC in Houston, Texas, U. S.A.,1-4 May 1995.

This paper was selected for presentation by the OTC Program Committee folfowing review of information contained in an abstsact submitted by the author(s). Contents of the paper,
as presented, have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s) The material, as presented, does not necessarily reflect
any position of the Offshore Technology Conference or its officars, Parmisslon to copy is restricted to an abstract of not more than 300 words. Illustrations may not be copied. The abstract
should contain conspicuous acknowledgment of where and by whom the paper is presented.

ABSTRACT INTRODUCTION

Ttiular joints are integral components of ofihore The continuing need to demonstrate the structural
structures and, as such, many codes aaliress their integrity and safety of offshore steel platforms
design. 7he strengths predicted by the codes ojien requires that they are subject to periodic structural
dl~er widely due to the adoption of dl~erent analysis and assessment. Initially, a linear elmtic
philosophies during codefimudation and to dl~erences analysis for extreme event loading is conducted for
in the underlying alztabases used during curve jitting the jacket structure, modelling the actual condition of
and calibration exercises. In the assessment of emkting the platform (as-built plus all subsequent
structures, a hierarchy of analysis methoak, j?om modifications). The joint components are subjected to
elastic analysis with improved code checti to non- code checking (to API RP2A(1J or HSE Guidance
linear 3-D pushover analysis, may be employed. To Notes@J, for example) or specific assessment
avoki potentially unnecessary and expensive o~hore depending on whether the joints are undamaged or
strengthening work, d ir impomnt that these darnaged. In a number of instances, joint utilisation
assessments are based on best estimates of joint ratios may exceed unity, as a result of increased
strength (mean or characteristic, depending on applied loads or code updates, for example. In these
application) and joint stiJlhess behavior. cases a hierarchy of analysis methods, of increasing
complexity, can be adopted to assess with greater
77ukpaper gives the background to and interim results accuracy the integrity and adequacy of the structure
of a major international joint industry project in this under consideration. This hierarchy can be defined as
area. Based on the most extensive carefilly screened follows:
database ever assembled, the robusmess of present
design formulations have been examined, and new (a) Elastic analysis with improved code checks
formulations have been developed where present codes for joints.
are found wanting. Other Lrsues, such as chord load,
can length and material yield to ultimate strength have Elastic analysis with modelling of joint
(b)
ako been dressed. For the first time, expressions rotation for the joints in question. Code
have been developed fir P-8 and M% joint stlfiess check to API RP2A.
responre, as may be used in advanced analysis
including the simulation of joint behaviour during (c) As (b), but with improved code checks for
pushover analysir. The paper goes on to describe the
joints.
expected jiture developments in this jieki of
technology.
References
andjtguresat endofpaper.

941
2 Strength and Stiffness of Tubular Joints for Assessment/Design Purposes OTC 7799

(d) As (b), but with full modelling of joint non- closely corresponds (ie. having similar joint
linear behaviour, uncoupled and coupled. parameters) to the joint being studied. Even
then, it is not clear how reliable or
(e) As (d), but with improved code checks for representative the reported curve is.
joints. Although some measure of a single curves
reliability may be gleaned from comparing its
(0 System reliability, utilising one of (a) to (e) ultimate strength with that obtained from
above for characterizing joint and system (mean) strength equation, there are no such
resistance parameters. parametric equations for, say, the deformation
at peak load. Thus, what is required is the
(g) 2-D pushover analysis (non-linear joint and formulation of closed-form expressions which
member behavior). can be used to generate non-linear load
deformation data for joints of practical
0-0 3-D pushover analysis. geometries. It may be noted that implicit in
generating reliable load deformation curves,
(i) System reliability, utilising (g) or (h) above it is necessary to have reliable estimates of
for characterisation. ultimate stren-@s, and this bears on issue (i)
above.
The above hierarchy is nominal, and may be (and
often is) adjusted depending on the structure under Recognizing the present day need for development of
consideration and the findings from preceding the above technologies, and following favorable
analyses. This is particularly valid for options (~ to findings from feasibility investigations, a joint
(i). Two specific technical issues become clear from industry initiative was instigated(3J with the following
the above analysis methods:- two main objectives:

(i) The improved joint code checks noted above to develop industry-accepted assessment
recognise that codes do not atways give best criteria for use in the reappraisal of existing
estimates of joint strength, for example API steel platforms
RP2A and the HSE Guidance Notes
essentially represent 1980 and 1985 to develop close&form expressions which will
technology, respectively, and no fundamental enable non-linear load deformation curves to
changes have been made since that time. It is be readily created for the range of joint types,
also recognised that a number of technological Ioadcases and geometries that commonly
advances have taken place regarding the occur in practice.
ultimate limit state of tubular joints from 1980
to date, including the generation of pertinent The results of the joint industry initiative will enable
new data and information. Refined criteria the above hierarchy of analyses to be conducted in a
can be established for use in the structural technical y-appropriate, safe and cost-effective
appraisal of existing installations, whereby manner. This, in turn, allows rational decisions to be
reliability of capacity algorithms can be made concerning the structural integrity of existing
increased without compromising safety joints and jacket systems without recourse to potential
(Assessment Criteria). expensive strengthening measures which might
inappropriately be deemed necessary through the
(ii) In order to apply options (b) to (i) above, adoption of outdated recommended practice& Whilst
information on joint stiffness has to be the first objective above concerns strength criteria for
available. For most options this implies that existing installations, it is recognised that the
knowledge of the fill non-linear load developed criteria will represent 1994-95 technology
deformation (I% and MOand their interaction) and, therefore, would be expected to be considered by
is required, although elastic L.JFs (local joint codedraflhg bodies. In particular, the current 1S0
flexibilities) suffice for applying options (b) efforts on producing a harmonised offshore code can
and (c). Non-linear load deformation data, in be mentioned in this respect.
tie form of P6 or M@ curves, is only
sometimes reported in the literature. It is not,
therefore, always possible to identify a load
deformation curve from the literature that

942
OTC 7799 A.F. Dier and M. L&mi 3

DATABASES affected by joint parameters (J!3, T etc.). These


differences may be attributed to variability in the
A considerable number of joints have been tested underlying databases, in the appraisals of those
worldwide over the last three decades, and testing still databases and in the various philosophies adopted (eg.
continues. These generally have been on steel models use of first crack loads or ultimate loads in tension
although, more recently, joint behaviour has been loaded joints). For each joint and load type, the
explored through the media of Finite Element predictive power of existing formulations were
analysis(45Jand lead-tin alloy models@). These latter compared with the present database as it was expected
studies are useftd in identifying specific aspects of that at least some of the formulations would capture
behaviour, but should always be calibrated against the the effect of joint parameters reasonably well. The
steel model data. mean and characteristic strengths of the present
database can be predicted by the existing formulations
Although steel model data is generally accepted as adjusted by numerical scaling factors. The predictive
being more reliable, a scrutiny of the literature shows performance of each formulation may be ascertained
that there is a wide range in the standard of reporting, by comparing the coefficients of variation (COVS)
and in the standard of quality of the testing procedures and the various sensitivity plots of the ratio of
adopted by various investigators. It is therefore measured to predicted strength against@, -f etc. This
necessary to formalise and impose appropriate process could be used to select the most appropriate
screening criteria such that derived strength formulation (from the scaled existing formulations) for
formulations using the screened data can be each joint and load type.
considered to represent real joints. There are obvious
criteria (eg. chord yield strength and thickness should The above approach is somewhat traditional and no
be measured and not just minimum specified or fundamentally new formulation capturing better the
nominal values quoted) but many are more subtle such real effect of joint parameters would result.
as ensuring that the test boundary conditions have not Therefore, for some appropriate cases, approaches
influenced the result or that chord/brace member were developed from first principles. In particular,
yielding has not affected the measured response. The the effect of chord loads in joint strength was
screening criteria established for the project were considered at the outset. Chord loads are often found
vetted by experts worldwide. to drive joint strength and are incorporated in present
codes through the following equation:
It is noteworthy that the size of the screened database
is significantly larger than those used in the
P = QfQu
development of the current API and HSE
FYT2
recommendations. Table 1 indicates the number of
screened results, by brace loading type, used to
develop API~, HSE(8) and the present formulations. In the above equation, P is the allowable load, FY and
In part, the large increase in the present database T are the yield strength and thickness of the chord
material, Qu is the basic strength factor (which varies
reflects the inclusion of test data generated in the last
for joint and load type) and Qf is the chord load
decade. However, it also reflects that the previous
limit on minimum chord diameter (140mm or 125mm factor. Traditionally, test data are examined and Qu
for API and HSE respectively) has been relaxed to formulations are derived, ignoring any effect of
loomm. equilibrating chord loads which arise in the tests, and
then Qf is subsequently applied. Although this may
bean appropriate approach for design, it is considered
STATIC STRENGTH to be unduly conservative for assessment criteria. In
the present work, studies were undertaken to identify
There are a number of codes(l 29-13)that address the the effect of chord loads ab initio, and then adjust the
static strength of joints and their provisions are database to account for chord loads before establishing
generally based on appraisals of test data. Inspection Qu. It should be noted that this is not that simple, as
of these codes and other guidance show that for any the best fit Qf term is dependent on the choice of Qu
given joint type (TN, K, DT/X) and load type (axial, and vice versa.
in-plane bending or out-of-plane bending), different
formulations exist. This extends beyond simple In some instances, radically new formulations were
scaling of numerical coefilcients to fundamental found to improve the otherwise rather poor COVS of
differences in which joint strength is supposedly existing formulations. As an example, consider K

943
4 Strength and Stiffness of Tubular Joints for Assessment/Design Purposes OTC 7799

joints, Figure 1, in which the API formulations (with Pa curve was rather poor with the predicted response
factors of safety set to unity) are compared to new oscillating about the experimental curve, see Figure 5.
MSL formulations. The API formulation is lower
bound whereas that of MSL is based on mean A parametric formula, incorporating just three
strengths. As can be seen, the strength of all K independent coefficients, was finally found and which
joints, and particularly those of overlapped K joints, models load deformation curves very well. It easily
are much better predicted with the new MSL adapts to load deformation curves which have a flat
formulation. post-peak plateau and has proven to be stable during
curve fitting. For a given joint and load type, fitted
Other issues, such as chord length, can length, effect curves were found for the available experimental
of end boundary conditions, and yield to tensile curves. Expressions were then found that related the
material strength ratio, have all been appraised during fitted coefficients back to the joint parameters. One
the development of the assessment criteria. The of the fitted coefficients corresponds to the ultimate
assessment criteria, taken as a whole, may be used to strength and therefore only two new expressions, one
ascertain more accurately the strength of tubular joints each for the other two coefllcients, had to be found.
in a safe and rational manner. Thus, complete P6 or Mi3 curves can be relatively
easily generated from surprisingly simple formulae.
The original fitted curve and the predicted curve are
LOAD DISPLACEMENT CURVES shown superimposed on the experimental response in
Figure 6. The simplicity and elegance of the load
Only about 30% of the tests in the screened database deformation prediction equations will ease
have associated load deformation data, in the form of implementation of joint non-linear response into
P6 or MO curves. These have been carefully algorithms.
appraised to establish the method of deformation
measurement, and any appropriate adjustments were Future work in this area will be directed at coupling
made (eg. to correct for the inclusion of brace the P6 and MO curves to simulate more realistically
deformations). Load deformation curves for axially actual joint behaviour within fiarnes. Also, studies
loaded joints are shown schematically in Figure 2. will be undertaken on how best to incorporate chord
The response is fairly linear up to point 1, after which load effects, although even now approximate account
the joint continuously softens until a peak (point 3) is can be made through the appropriate adjustment of PU
reached. For tension loaded joints, failure is sudden (or MU).
when cracks, initiated at point 2, have grown
sufficiently to cause gross separation of the brace
from the chord. For compression loaded joints a CONCLUDING REMARKS
reduced strength is observed which extends over a
considerable range of deformation. The load at the On the basis of the studies conducted to date,
peak (point 3) may be predicted by the (mean significant increase in the reliability of joint strength
strength) assessment criteria equations. It has been formulations can be achieved, leading to more rational
showni14J that it is possible to derive P and 6 appraisal of existing installations where joints
expressions to capture the other points along the constitute the weak points. In turn, this leads to the
curve. For instance, in the case of initial softening or availability of an important tool in the decision-
joint yield @oint 1) for compression loaded DT joints, making process for cost effective and safe IMR
Figures 3 and 4 illustrate that data is reasonably operations.
tightly banded.
The availability of formulations, for the first time,
From a computational standpoint for implementation that predict the full non linear Pa or MO response of
into frame analysis software, it is somewhat messy to popular joint types, loadcases and geometries which
integrate a patchwork of expressions describing the 3 occur in practice, allows the effects of both linear and
or 4 (P, 8) points along a curve and so a more elegant non linear joint flexibility to be accounted for in
approach was sought. Rather than having discrete advanced analysis, including pushover analysis.
points, by which the response curve would be Much interest is being shown in system failure rather
approximated to piecewise linear functions, a single than component failure at the present time.
expression was called for. In the first instance a Knowledge of joint loaddeformation response is a
simple polynomial expression was tried. However, pre-requisite to assess accurately system response.
even with a 6th order polynomial the fit to a typical

944
OTC 7799 A.F. Dier and M. Lalani 5

NOMENCLATURE 7. Yura, J. A., Zettlemoyer, N. and Edwards,


I.F. Ultimate Capacity Equations for
F De-sign strength of chord member Tubular Joints. Paper OTC 3690, Offshore
Nf Moment Technology Conference, Houston, May 1980.
P Axial force
Qf Chord stress factor 8. Department of Energy. Background to New
Qu Joint basic strength factor Static Strength Guidance for Tubular Joints in
T Chord wall thickness Steel Offshore Structures. Offshore
b Brace/chord diameter ratio Technology Report OTH 89 308, HMSO,
Y Chord radius/T London, 1990.
6 Axial displacement at joint
o Brace rotation 9. Det norske Veritas. Rules for Classification
of Fixed Offshore Installations. 0s10,
Norway, July 1993.
ACKNOWLEDGEMENTS
10. Norwegian Petroleum Directorate. Acts,
The authors wish to recognise the contributions of Mr Regulations and Provisions for the Petroleum
R Ossei and Ms V Trembath in compiling databases Activity. Volume 2, 1993.
and conducting spreadsheet calculations referred to in
this paper. 11. American Welding Society. Structural
Welding Code - Steel. ANSI/AWS D1. 1-92,
1992.
REFERENCES
12. Canadian Standards Association. Code for
1. American Petroleum Institute. the Design Construction and Installation of
Recommended Practice for Planning, Fixed Offshore Structures. S473, June 1992.
Designing and Constructing Fixed Offshore
Platforms - Working Stress Design. API 13. CEN (European Committee for
RP2A-WSD, 20th Edition, 1993. Standardisation). Eurocode 3: Design of
Steel Structures - Part 1.1: General Rules
2. UK Health and Safety Executive. Offshore and Rules for Buildings. European
Installations: Guidance on Design, Prestandard ENV 1993-1-1, April 1992.
Construction and Certification. 4th Edition,
1990. (Formerly issued by the Department of 14, Lalani, M. Post-yield and Post-peak
Energy). Behaviour of Tubular Joints in Offshore
Structures, Tubular Structures V, E & FN
3. Assessment Criteria, Reliability and Reserve Spon, 1993.
Strength of Tubular Joints - Invitation to
Participate, MSL Engineering Limited, Dec.
Ref. P168PO01, October 1993. TABLE 1: Number of screened results

Brace Load
4. Dexter, E., Haswell. J., and Lee, M.K. A
Comparative Study on Out-of-plane Moment
Capacity of Tubular Joints, Tubular m
Structures V, E & FN Spon, 1993.
Compression 83 123 438

5. Healy, B and Zettlemoyer, N. In-plane Tension 21 44 60


Bending Strength of Circular Tubular Joints,
In-plane Bending 16 ~ 33 47
Tubular Structures V, E & FN Spon, 1993. I I
Out-of-plane Bending 17 20 30
6. Fessler, H., Hyde, T.H. and Khalid, Y.,
The Static S~en@ of T-joints Under Total 137 220 575
I I 1 1 I

Bending in Different Directions, Tubular


Structures
V,E &FNSpon,1993.

945
Purposes OTC 7799

?*.

&a

.*
2,W
. . .
.
. .
s..

& .. :
.
*
L/
.*....
.
. .
r .
i

am tom Xc-a lam mm mu! ,002 row


0

ii ?rtdkrcd I.. d .<mdi~ 1. Arl

m-

eu . . ...*- , 1.

2W

1.$0

8
..

.
* x

f{..~ -
0$4

.nca .mca .v3.c4

fl

b] ?rcdi<tcd [od .c,ord!. ~ ,. MSL

Fig. 1: Measured/Predicted Loads for K joints

r
Pu .._ -L1---.l 2 = CRACK INITL4TIC)N
/ 3 3 = UL.mmTE LOAD
/

I
/ p (OtiPRISSION

I II NSION

[ .- .- -.

or FORHATION.

Fig. 2: Typical Load/Deformation curves

946
CJ
n

1500
50 I 1
. -.
.
,.
2
o
*O _
. 2020
P
-30
u
~
Q
G 20
c
F
.-
.
:.
10
I
3
.. Iwo
5

I
0
0 0.2 04 0.6 0.8 [
BETA so 0

00 J
0,0 >0 100 1>0 200 150 100 1$0 100 4s.0
Odwmztim [m]

Fig. 3: Normalised Yield Load Against Beta

Fig. 5: Sixth Order Polynomial fit to Typical Test Curve

)03 o

.
.

0
0 I
del[ar&D/Pr(x IOE-:)
4 5

r
Fig. 4: Normalised Deformation at Yield Against Beta 4
Fig. 6: Fit of New Formulation to Typical Test Curve
- . ___

_ _- .
. _ ___+ __ ___ ___ ,_ _. __ _.._= ==:. ,. . -.- - ..-=- .. ..... ..... ... .- -c- -. _ ._ __==. -_
-m

-.-.. . - . ..===. *..=.Q


.- ._. :__- -. ..:= .._>* ,
=_.._ ._- =+<.
>T=-z-_.=.. .._ .. .
___. .3-.. --

~ . . .. . . T._-=;._=_: <-= -:.=: =_-_,>;= :-::: : ->. _ ._ .=.= -. z_..-_ .


>. . .. .=._ .. . .. .. . . . . . .. _=_..= ...
----- =___
G._

-- ..7 -. : . ..+ ~ .. ~.=..=__ ,= .-. ... _ .=,

.- -= -. ., .. .---:... .7. . .= _.. _ ~-__ ~: ~

.__- _

_% ___ ._.. _. __ .== ___+_ _.


_. _A ___
.. =. _= - _._ . . .- .. _ .= ._ _= ~_:_ ._j ___ T__ T..
.= .-.
~_ ..__. _.. Q= _._._~ _. ..= ~. ~., ___ _._ .,

.<. .
_____ ~.. . __. .- =
.= ___ .-: :.:. _- =
. . =-- .
. = _:. _= =. .. _=Q .-__._r __ _

- ___ _______ __ ___ : __. .
______ _____ __ -. _. . _=. =- _. ,. _: _ _


.

..-- _. ...= _- _______ ___ =:_= = =_ _. _=___ .:-L_ =, . .. .
-_ ___ _- .. --_-;
_.___ _. ._ . ._=. _, ._=

=. = :=-
.

.
. ___

Vous aimerez peut-être aussi