Vous êtes sur la page 1sur 60

Half Scale Three-Storey Inlled RC Building; A Comparison of Experimental and Numerical Models

A Dissertation Submitted in Partial Fullment of the Requirements for the Master Degree in

Earthquake Engineering

By Hassan Ali

Supervisors: Dr. Roberto Nascimbene Dr. Rui Pinho

May, 2009

Istituto Univeritario di Studi Superiori Universit` a degli Studi di Pavia

The dissertation entitled Half Scale Three-Storey Inlled RC Building; A Comparison of Experimental and Numerical Model, by Hassan Ali, has been approved in partial fulllment of the requirements for the Master Degree in Earthquake Engineering.

Dr. Roberto Nascimbene

Dr. Rui Pinho

Abstract

ABSTRACT

The present work reports the comparison between the experimental and numerical model of 3D three storey RC inlled building. The experimental model is similar to SPEAR structure but with major differences such as introduction of inll panels and reduced scale (i.e. 1:2). Experimental test was conducted on the Shake Table at European Centre for Training and Research in Earthquake Engineering (EUCENTRE), Pavia Italy. Testing phases consist of the seismic excitation applied to building using a natural accelerogram scaled at different levels of PGA, in order to observe the development of the collapse mode and the overall resistance of the building. SeismoStruct a bre based Finite Element Program has been used for implementation of Numerical Model. The main feature of the current work is the use of inll panel and rigid diaphragm in the model. The hysteric parameters of Inll panel are rst calibrated with experimental test which are then used in the model to accurately perform the Dynamic Time History analysis. Finally the results of the Numerical and experimental model are compared.

Keyword: SPEAR structure; Scaled Structure; SeismoStruct; Inll Panel; Rigid Diaphragm and Calibration of Inll Panel.

Acknowledgements

ACKNOWLEDGEMENTS

I would like to express my deep and sincere gratitude to my supervisor Dr.Rui Pinho. His wide knowledge and logical way of thinking have been a great value for me. I am also deeply grateful to my supervisor Dr.Roberto Nascimbene for his detailed and constructive comments, and for his important support throughout this work. His understanding and personal guidance have provided a good basis for the present thesis. I wish to express my sincere thanks to Dr.Qaisar Ali, Director Earthquake Engineering Center and Dr.Akhtar Naeem Khan, Head of Department, Civil Engineering, N-W.F.P University of Engineering & Technology, Peshawar, Pakistan, who introduced me to the eld of Earthquake Engineering. Their ideals and concepts have a remarkable inuence on my career. Dr.Qaisar Ali has been an inspirational throughout my graduate studies. The nancial support for my entire master program from N-W.F.P University of Engineering & Technology Peshawar, Pakistan, is gratefully acknowledged. Many friends in ROSE School have helped me stay sane through these difcult years. I greatly value their friendship and I deeply appreciate their belief in me. I am thankful to EUCENTRE Staff and administration for providing me all the data required for the present work. Last, I want to thank my parents, without whom I would never have been able to achieve so much. They has been a constant source of love, concern, support and strength all these years. I cordially thank my son for not forgetting me, even though he had so seldomly the opportunity to see me. I especially wish to express my love for my wife who only knows the real price of this dissertation as we suffered and paid it together. I thank for her endless love, patience, and understanding.

ii

Index

TABLE OF CONTENTS

ABSTRACT ACKNOWLEDGEMENTS TABLE OF CONTENTS LIST OF FIGURES LIST OF TABLES 1. INTRODUCTION 1.1 Organization Of the Report . . . . . . . . . . . . . . . . . . . . . . . . . .

i ii iii vi ix 1 2 3 3 4 4

2. PROJECT DESCRIPTION 2.1 2.2 SPEAR Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . EUCENTRE Project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.1 Similitude condition . . . . . . . . . . . . . . . . . . . . . . . . . .

iii

Index

2.2.2 Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.3 Reinforcement Detail for beam and column . . . . . . . . . . . . . . 2.2.4 Additional Masses . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.5 Material Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2.6 Testing Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. LITERATURE REVIEW 3.1 Inll Panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.1 Micro-Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.2 Macro-Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.3 Cyclic Behavior of Inll Panel . . . . . . . . . . . . . . . . . . . . . 3.1.4 Proposed model for the analysis of Inlled frames . . . . . . . . . . 3.1.5 Input Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1.6 Reduction in area due to lateral displacements . . . . . . . . . . . . 3.1.7 Openings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Rigid Diaphragm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 8 11 12 13 15 15 15 16 19 19 24 25 26 27 29 29 30 33 39

4. NUMERICAL MODEL OF CASE STUDY 4.1 4.2 4.3 SeismoStruct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Numerical Model of Structure . . . . . . . . . . . . . . . . . . . . . . . . Calibration of Inll Panel . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3.1 Geometrical Parameters . . . . . . . . . . . . . . . . . . . . . . . .

iv

Index

5. ANALYSIS AND RESULTS 5.1 5.2 Modal Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Non-linear Time History Analysis (NTHA) . . . . . . . . . . . . . . . . .

41 41 42 45 45 46 47

6. CONCLUSIONS AND RECOMMENDATIONS 6.1 6.2 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Future Developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

REFERENCES

List of Figures

LIST OF FIGURES

2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9

Scaled Structure on Shake Table . . . . . . . . . . . . . . . . . . . . . . . . . Typical Plan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Elevation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Beam Column Plan conguration . . . . . . . . . . . . . . . . . . . . . . . . . Typical Beam Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . . Typical Column Reinforcement . . . . . . . . . . . . . . . . . . . . . . . . . . Construction and Reinforcement Detail for Foundation . . . . . . . . . . . . . Typical Column Reinforcement and Construction Detail . . . . . . . . . . . . . Typical Beam & Slab Construction & Reinforcement Detail . . . . . . . . . . .

6 6 7 7 9 9 10 10 11 11 12

2.10 Additional Masses on Scale Structure . . . . . . . . . . . . . . . . . . . . . . 2.11 Additional Masses on Scale Structure, Plan . . . . . . . . . . . . . . . . . . .

vi

List of Figures

2.12 Additional Masses on Scale Structure, Section . . . . . . . . . . . . . . . . . . 2.13 Input Signal [PGA = 0.3g] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.14 Input Signal [PGA = 0.54g] . . . . . . . . . . . . . . . . . . . . . . . . . . . .

12 14 14

3.1 3.2 3.3

Deformation under shear load . . . . . . . . . . . . . . . . . . . . . . . . . . Variation of the ratio


bw dw

16 18

for inll frames as a function of parameter h . . . .

bw as a function of parameter h according to Decanini Variation of the ratio d w and Fantin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

18 19 20 20 21 22 23 26 28

3.4 3.5 3.6 3.7 3.8 3.9

cyclic axial behavior of masonry . . . . . . . . . . . . . . . . . . . . . . . . . Stress Strain Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Modied strut models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Inll panel element conguration . . . . . . . . . . . . . . . . . . . . . . . . . Shear spring conguration . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stress state considered to evaluate the strength of masonry . . . . . . . . . . .

3.10 Variation of area of strut as a function of the axial strain . . . . . . . . . . . . . 3.11 Diaphragm constraint to model rigid oor slab . . . . . . . . . . . . . . . . . .

4.1 4.2 4.3

Material Inelasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rigid Link connection at column C2 . . . . . . . . . . . . . . . . . . . . . . . (a) Displacement-Based Element Formulation, (b) Force-Based Element Formulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

30 32

33 34

4.4

Layout of Inll panel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

vii

List of Figures

4.5 4.6 4.7 4.8 4.9

Reinforcement Detail of Beam and Columns . . . . . . . . . . . . . . . . . . . Force-Displacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Force-Displacement, Trial 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . Force-Displacement, Trial 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . Force-Displacement, Final Calibrated . . . . . . . . . . . . . . . . . . . . . .

34 35 36 37 38

5.1 5.2

Experimental Results for Col 6 . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison of Displacement Time History for Col 6 . . . . . . . . . . . . . .

43 44

viii

List of Tables

LIST OF TABLES

2.1 2.2 2.3

Scale factors for model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Average Compressive Strength of concrete . . . . . . . . . . . . . . . . . . . . Average values of yield strength of specimens . . . . . . . . . . . . . . . . . .

5 13 13

4.1 4.2 4.3 4.4

Default Values for Calibration of Inll Panel . . . . . . . . . . . . . . . . . . . Intermediate Values for Calibration of Inll Panel . . . . . . . . . . . . . . . . Final Values for Calibration of Inll Panel . . . . . . . . . . . . . . . . . . . . Parameters contributing most . . . . . . . . . . . . . . . . . . . . . . . . . . .

36 37 38 39

5.1 5.2

Estimated Vibration Periods . . . . . . . . . . . . . . . . . . . . . . . . . . . Empirical Eq. and SeismoStruct result . . . . . . . . . . . . . . . . . . . . . .

41 42

ix

Chapter 1. Introduction

1.

INTRODUCTION

It has been observed that frame inll panel are mostly used as interior partition walls and external walls in building according to the requirement of the usage. But they are seldom used in numerical calculation. The main reason for not including is complexity of analytical models for inll panel and not adequate knowledge. Recent studies have shown that the use of masonry inll panel has a signicant not only on the strength and stiffness but also on the energy dissipation mechanism of the overall structure. Neglecting the effects of masonry inll can lead to inadequate assessment of structural damage of inlled frame structures subjected to intense ground motions. Current scope of the work is to compare the experimental results with the numerical simulation. The experimental model is a reduced scale multi-storey building with inll panel and openings. Experimental test comprises of a shaking table test at the TREES Lab EUCENTRE, Pavia, Italy. The building is a representation of older building in Southern Europe without earthquake design. It has been designed only for gravity loads. The research at TREES Lab is aimed at verifying efciency and accuracy of the rules for seismic assessment and repair of existing buildings included in the Italian Code OPCM 3274 and recently included in the new Italian Technical Regulations for Construction through the use of numerical simulations and experimental tests on a building representative of the construction in the years 50-60 in Italy. The Finite Element package used for numerical verication is SeismoStruct. The software is a bre-element based program capable of taking into account both geometric non-linearity and material inelasticity. To account for the inelastic deformations of masonry panel, the four node masonry panel element developed by Crisafulli is used. Each panel is represented by six strut members, to carry axial load across two opposite diagonals and shear from top to the bottom of the panel.

Chapter 1. Introduction

The slab of the scaled building is very rigid. It is therefore necessary to correctly distribute the inertia forces through out the structure. Hence use of rigid diaphragm nodal constraint approach that makes use of Lagrange Multiplier is opted for the numerical model. This approach of modeling slab as rigid diaphragm is a reliable way of modeling the presence of very stiff oors in this type of irregular structures (Pinho et al. 2008).

1.1

ORGANIZATION OF THE REPORT

The dissertation is organised into six chapters covering from the theoretical background of inll panel & rigid diaphragm to their implementation in SeismoStruct for analysis. A brief description of each chapter is given below: Chapter 2 provides a brief introduction of original SPEAR project. It includes the main objectives with brief characteristics of the SPEAR building. After the introduction of the SPEAR project a detail overview of the present case study (i.e. EUCENTRE project) is mentioned. The case study is a reduced scale (1:2) of the original SPEAR building tested at EUCENTRE, Pavia Italy. The main difference between this building and SPEAR is; reduce scale and introduction of inll panel. Chapter 3 delves on with the literature review of inll Panel & rigid diaphragm. The inll panel model used in this study is the one developed by Crisafulli and implemented by Blandon in SeismoStruct. Chapter 4 deals with the calibration of inll panel with experimental data. The description of geometrical and material properties used for numerical model are discussed.The Finite element code used for this case study is SeismoStruct, a bre element based program capable of predicting large displacement. Chapter 5 presents the results of two analysis i.e. Eigenvalue & Dynamic Time History Analysis both with displacement based element formulation and force based element formulation. In the end a comparison between numerical and experimental data is presented. Chapter 6 nally presents the summary of the whole work along with the conclusion.

Chapter 2. Project Description

2.

PROJECT DESCRIPTION

2.1

SPEAR PROJECT

A series of experimental test were conducted by European Laboratory for Structural Assessment (ELSA), at ISPRA Italy under the research project nanced by European commission. The project was named as Seismic Performance Assessment and Rehabilitation (SPEAR). Several partners within Europe and around the world participated in the project, with the leading role played by ELSA. The project aim was to develop simple assessment and rehabilitation strategy for existing RC structures which were designed only for gravity loads and are typical of non-seismic design. These comprise a major part of building structure in Europe and many parts of the world. From social and economic point of view, it was not feasible to replace these existing structures with new construction, nevertheless to achieve an acceptable seismic level; these structures must be assessed and retrotted. Under the SPEAR project, a Pseudo Dynamic test of a full scale three-storey reinforced concrete (RC) building (torsionally unbalanced) was tested. The structure is a simplication of an actual three-storey building designed by Fardis (2002). The building is a representative of older construction in Greece, Portugal, Southern Europe and many parts of the world, without provisions for earthquake resistance Pinho et al. (2008). It was designed only for gravity loads using the concrete design code used in Greece between 1954 & 1995. The materials and construction practice are the one used in Southern Europe (Italy and Portugal) in early 70s. The SPEAR structure is regular in elevation but have plan irregularity in both directions, which is a typical conguration of non-earthquake design. Some of the main features of the structure are:

For longitudinal reinforcement smooth bars are used.

Chapter 2. Project Description

Minimum amount of stirrups are provided. Beam-column joint connection is provided without stirrups. Lap splices are located just above the joints.

Inll walls and stairs were not included in the test structure. For detail description of the structure, along with the experimental and numerical activity of the SPEAR project can be found at Negro et al. (2004), Molina et al. (2004), Molina et al. (2005).

2.2

EUCENTRE PROJECT

The present study is focused on the experimental work recently carried out at European Centre for Training and Research in Earthquake Engineering (EUCENTRE, Pavia, Italy). The analyzed experimental model is a three storey two bay RC structure similar to SPEAR structure but with reduced Scale i.e. 1:2. The test is part of the activities programmed in the research projects Numerical and experimental assessment of recommendations for existing RC buildings included in the OPCM 3274 and Use of innovative materials for strengthening and reparation of RC structures in high seismicity areas. The project is nanced by the Italian Civil Protection department and the foundation CARIPLO of Milan Italy. The project intend to verify the seismic assessment and retrotting techniques through numerical simulation by means of conducting experimental test on buildings representative of older construction in Italy. The building is tested with natural accelerograms scaled at different levels of PGA. Although the building is similar to SPEAR but with many differences, the most important differences are: reduce scale (i.e. 1:2), use of inll walls with openings, earthquake will be applied only in one direction (Marazzi et al. 2007).

2.2.1

Similitude condition

For real dynamic test of scaled models on shake table, Cauchy & Froude law must be satised. The Cauchy law is adequate for phenomenon in which restoring forces are derived from stress-strain constitutive relationship, while Froude law applies to cases where gravity forces are important. Thus for the realistic modeling of non-linear dynamic response of structure both similitude laws must be satised Sullivan et al. (2004). The simultaneous satisfaction of Cauchy Ep and Froude similitudes leads to scale factors represented in Table 2.1. (assuming E = 1 i.e. m 4

Chapter 2. Project Description

the model and prototype have the same material properties). The acceleration scale is unity while time scale is square root of geometrical scale . This means that in model the time scale 1 , therefore the accelerogram applied to the structure have shorter is compressed by a factor durations, higher frequency and the same accelerations. Another important consequence of the similitude law, is the increase of the mass of the model relative to the reference prototype ( become inversely proportional to the geometric scale factor ). This relations imposes difculties in terms of either nding non-standard high density material, or adding mass to the model without inuencing its stiffness. In the present study additional masses are applied on the scaled model as discussed in Section 2.2.4. Parameter Length Modulus of Elasticity Specic mass Area Volume Mass Displacement Velocity Acceleration Weight Force Moment Stress Strain Time Frequency Symbol L E A V m d v a w F M t f Scale Factor (Cauchy) Lp /Lm = Ep /Em = e = 1 p /m = = 1 2 3 3 1 1 3 2 3 1 1 1 Scale Factor (Cauchy + Froude) Lp /Lm = Ep Em = e = 1 p /m = = 1 2 3 2 1 2 1 2 2 3 1 1 1 2 1 2

Table 2.1: Scale factors between prototype and model, adapted from Sullivan et al. (2004)

2.2.2

Geometry

An overview of the test building along with plan and elevations are presented in Figure 2.1, Figure 2.2 and Figure 2.3. The scaled model is regular in elevation but non-symmetric in plan with a storey height of 1.5 m between oors. The slab thickness for all oors is 0.08m (80mm). All beams have equal dimensions of 0.125 x 0.25m (125 x 250mm). The cross section dimension of column C2 is 0.125 x 0.375m (125 x 375mm) whilst all remaining columns are 0.125 x 0.125m (125 x 125mm). The plan layout with beam and column conguration is shown in Figure 2.4. 5

Chapter 2. Project Description

The frame was lled by nonstructural masonry inlled walls. Elevation A & C consists of door opening (525 x 1050 mm) and (400 x 1050 mm) at each oor level. Similarly elevation D contains window opening (600 x 650 mm) at each oor. Elevation B contains solid inll, without openings. The general layout of the location and dimension of inll and opening are shown in Figure 2.2 and Figure 2.3.

Figure 2.1: Scaled Structure on Shake Table

Figure 2.2: Typical Plan

Chapter 2. Project Description

Figure 2.3: Elevation

Figure 2.4: Beam Column Plan conguration

Chapter 2. Project Description

2.2.3

Reinforcement Detail for beam and column

Column longitudinal reinforcement consists of 6mm smooth round bars. Lap splicing of column reinforcement is made above oor level which is typical of non-earthquake resistant detailing. Column stirrups are provided by 3mm bars with 90 overlapping hook. Minimum spacing between stirrups is 70 mm. Typical beam reinforcement consists of 6 mm and 10 mm bars, depending on the loading condition. For bottom reinforcement out of four bars two bars are bent up along the column direction while remaining two is anchored in the joint with a 180 hook. Stirrups or transverse reinforcement comprises of 3mm bars with 90 overlapping hook and minimum spacing of 55 mm. Some of the beam directly intersects with other beam which creates a beam to beam joint. The design is based on weak column strong beam, which is opposite to the current seismic design procedure. Also none of the transverse reinforcement from beam and column continues within the joint, which makes the beam-column joint weakest point during earthquake excitation. Typical layout of section of a beam and column along with reinforcing detail is presented in Figure 2.5 and Figure 2.6. Figure 2.7, Figure 2.8 and Figure 2.9 shows typical construction stages of the experimental model. It can be clearly seen from the reinforcement pictures taken at the time of construction that the experimental model lacks seismic resistant reinforcement detailing.

Chapter 2. Project Description

Figure 2.5: Typical Beam Reinforcement

Figure 2.6: Typical Column Reinforcement

Chapter 2. Project Description

Figure 2.7: Construction and Reinforcement Detail for Foundation

Figure 2.8: Typical Column Reinforcement and Construction Detail

10

Chapter 2. Project Description

Figure 2.9: Typical Beam & Slab Construction & Reinforcement Detail

2.2.4

Additional Masses

In order to comply with the similitude condition, additional masses are applied to the specimen. These include permanent and live load modied and increased, as discussed in Section 2.2.1. For the present study additional masses consists of unreinforced concrete having specic mass, = 2400kg/m3 . The position of additional masses are shown in the Figure 2.10, Figure 2.11 and Figure 2.12.

Figure 2.10: Additional Masses on Scale Structure

11

Chapter 2. Project Description

Figure 2.11: Additional Masses on Scale Structure, Plan

Figure 2.12: Additional Masses on Scale Structure, Section

2.2.5

Material Properties

The materials used for the scale specimen such as concrete, steel and inll were chosen such that they fulll the requirements of similitude conditions. The concrete compressive 12

Chapter 2. Project Description

strength test was carried out on cubes of 150mm sides. The results of the test are presented in the following Table 2.2. The table provides average values for each element on each oor. The cylinder strength of concrete are obtained by following relation: fc 1.2 = fcu where fc = Cylinder Compressive Strength fcu = Cube Compressive Strength (2.1)

Compressive Strength Element Column 1st Floor Beam 1st Floor Column 2nd Floor Beam 2nd Floor Column 3rd Floor Beam 3rd Floor Cylinder [MPa] 38 52 39 52 33 44 Cube [MPa] 32 43 33 43 27 36

Table 2.2: Average Compressive Strength of concrete

Steel bars of diameter 3mm, 6mm and 10mm are chosen for the scaled model. Tensile strengths tests were made on the specimen and the results of only the yield values are presented in Table 2.3. For every diameter bar three specimen were tested, average of the three values are chosen for the numerical model. Yield Strength Dia 10mm 06mm 03mm Specimen1 [MPa] 286 366 801 Specimen2 [MPa] 300 360 800 Specimen3 [MPa] 365 801

Table 2.3: Average values of yield strength of specimens

The specimen inll panels are made of cellular concrete Gasbeton RDB. The inlls are constructed after the construction of frame. Thickness of inll panel is 50mm. Geometrical conguration of inll panel is presented in Figure 2.3.

2.2.6

Testing Procedure
13

Chapter 2. Project Description

(a)

Input Signal

The input signal selected for the test is Herceg-Novi record of 1979,

Montenegro earthquake. It was divided into minor, moderate and severe earthquake. Minor intensity excitation (PGA = 0.02g) are designed to access the elastic response of the structure and its dynamic properties through the calculation of the natural frequencies and mode shapes. Moderate (PGA = 0.30g) & severe intensity (PGA = 0.54g) were applied to the structure to study the inelastic response of the structure for assessment. To conform with the time similitude condition, time history of the original accelerogram was compressed by a factor 1/ 2.

Figure 2.13: Input Signal [PGA = 0.3g]

Figure 2.14: Input Signal [PGA = 0.54g]

14

Chapter 3. Literature Review

3.

LITERATURE REVIEW

3.1

INFILL PANEL

It has been observed in the past, that design of buildings is done without taking into account the effect of the inll, mainly because of lack of research and experimental work. The general idea behind ignoring inll panel during analysis was its highly non-linear nature. The most important factors contributing to the nonlinear behavior of inlled frames arise from material nonlinearity, which required complex computational techniques for design. Recent studies have shown that behavior of RC frames lled with masonry inll panel can signicantly increase the stiffness, strength and energy dissipation characteristics of framed structures. In order to fully understand the behavior of inll panel and its mode of failure, several analytical models have been proposed by researchers around the world. These models can be classied into two main groups, namely micro-models (local) and macro-models (simplied).

3.1.1

Micro-Models

These models are represented by using Finite Element method. Mallick and Severn (1967) were the rst one to use this approach. Different elements were used to model this approach such as beam elements for surrounding frame, plane frame element for representing inll and interface element for frame and panel interaction. The benet of using nite element approach is to study in detail all possible modes of failure but its use is limited due to the greater computational effort and time required in analysis & modeling. For further detail on micro-models based on nite element apporach, readers can refer to work of Mallick and Severn (1967), Gooman (1968), Stafford (1962), Riddington and Stafford Smith (1977), KIng and Pandey (1978), Liauw and Kwan (1984), Rivero and Walker (1984), Dhanasekar (1985), Chrysostomou (1991), Shing (1992), Syrmkezis and Asteris (2001) as suggested by Smyrou (2006). 15

Chapter 3. Literature Review

3.1.2

Macro-Models

In order to overcome the complexity and computational requirement using micro-models, research has been done to simplify the modeling of inll panel with a single element. The main idea has been to study the global effects of inll panel on structures under lateral loads. Since rst attempts from Polyakov (1956), analytical and experimental tests have shown that a diagonal strut with appropriate mechanical properties can provide a solution to the problem. Several authors have modied the characteristics of single strut model with multi strut congurations to better understand the effect of,micro-cracking in the corner of the inll panel due to tensile stresses and higher shear strength of the inll panel relative to the frame. A brief review of the expressions developed because of these experimental work is presented below.

Figure 3.1: Deformation under shear load (Paulay and Priestley 1992)

(a)

General Properties of Diagonal Strut

Based on elastic studies Polyakov (1956)

conducted one of the rst analytical studies on inlled frames. He considered the effect of inll in each panel as equivalent to diagonal bracing. In 1961, Holmes took the idea and suggested that inll panel can be replace by an equivalent pin-jointed diagonal strut. He proposed that the diagonal strut to have the same material and thickness as the inll panel. The width of strut was taken equal to one third of the strut length as shown in the Figure 3.1 i.e. dw bw = 3

(3.1) 16

Chapter 3. Literature Review

Stafford (1966) performed a series of test on square steel inlled frames. He observed that contact length between the wall and frame is related to the width of the strut. From the experimental result he proposed the following relation for nding the contact length between the wall and inll frame. z= =
4

(3.2) (3.3)

Em tw sin(2) 4Ec Ic hw

where represents the relative stiffness between the RC frame and the wall. Em is the elastic modulus of the masonry, the angle of the diagonal strut with the beams, Ec and Ic represents elastic modulus and moment of inertia of concrete column respectively. Paulay & Priestley took a conservatively high value for the width of equivalent strut. According to them, a high value of bw will result in a stiffer structure. The relation given by them is as follows: bw = 0.25dw (3.4)

Mainstone (1971) conducted tests on small scale specimens(h = 406mm) diagonally loaded in compression and proposed the following expression:
0 .3 bw = 0.16h dw

(3.5)

Klingner and Berter (1978) based on scale test done by Mainstone (1971) proposed the following the equation: bw = 0.175( h)0.4 dw dw Liauw and Kwan (1984) found the following relation from previous experimental data: bw = 0.95hm cos h (3.6)

(3.7)

In the above equation it was assumed that is equal to 25 and 50 . These values represent the limit values for practical situations. Crisafulli (1997) compared the variation of the parameter h with the ratio of
bw dw

and found out that the ratio bw /dw decreases as h increases. The variation

is presented in the Figure 3.2.

17

Chapter 3. Literature Review

1 Holmes Paulay & Priestley 0.8 Mainstone Liauw & Kwan (25) 0.6 Liauw & Kwan (50)

bw/dw
0.4 0.2 0 0 2 4 6 8 10

Figure 3.2: Variation of the ratio

bw dw

for inll frames as a function of parameter h

Decanini and Fantin (1987) for the rst time considered the fact of cracked and uncracked masonry effect and proposed two sets of equations based on the results from the framed masonry tested under lateral forces. The variation is plotted in the Figure 3.3.
1.0

0.8 Uncracked Cracked

0.6

bw/dw
0.4 0.2 0.0 0 2 4 6 8 10

Figure 3.3: Variation of the ratio Fantin

bw dw

as a function of parameter h according to Decanini and

18

Chapter 3. Literature Review

3.1.3

Cyclic Behavior of Inll Panel

In this section, the model proposed by Crisafulli (1997) for the hysteric behavior of masonry subjected to cyclic loading is described. The model is capable of taking into account the non-linear response of masonry in compression. As the model allows to take into account the variation of struts cross section as a function of the axial deformation experienced by element, it is possible to consider the loss of stiffness due to shortening of the contact length between frame and panel as the lateral load increases. Stress Strain relation for the hysteric model proposed are shown Figure 3.4 and Figure 3.5.

Figure 3.4: General characteristics for cyclic axial behavior of masonry (Crisafulli 1997)

Decanini and Fantin (1987) from experimental results showed that the equivalent width of the strut decreases by about 20% to 50%, due to cracking of masonry panel. The main advantage is that it allows the user to control the variation of stiffness and the axial strength of the masonry strut.

3.1.4

Proposed model for the analysis of Inlled frames

As discussed previously for macro-modeling of inlled frames, equivalent diagonal strut representation is used. Crisafulli (1997) adopted the same approach, considering a multi-strut formulation as shown in the Figure 3.6. An initial study was carried out to see the limitations 19

Chapter 3. Literature Review

Figure 3.5: Stress Strain Curve (Crisafulli 1997)

of single strut model and inuence of different multi strut model on response of the structure. The main focus was on stiffness of the structure and in the actions induced in the surrounding frame.

Figure 3.6: Modied strut models Crisafulli (1997)

20

Chapter 3. Literature Review

Numerical results obtained from three strut models were compared with a nite element micro-model formulation. The area of the equivalent strut was kept constant and static lateral load was applied assuming linear elastic behavior. Nonlinear effects were considered for nite element model to represent the separation of panel frame interface. Results from the test shows that, stiffness of the inlled frame is similar in all cases. It decrease slightly for two and three strut model, however there was signicant change in stiffness for three strut model depending on the contact distance hz , which is function of contact length z. It was also observed from the results that single strut model under-estimated the bending moment, two strut model showed much larger values while three strut model constituted better approximation with the nite element model. It was concluded from the results, although single strut model represent good estimation of stiffness of the inlled frame and axial forces by lateral forces. However, a more rened model is required which could give realistic values for bending moment and shear force in frames.

(a)

Description of proposed model

Crisafulli proposed a new multi-spring model which

is intended to represent the shear failure of masonry. The model accounts separately for the compressive and shear behavior of masonry using a double truss mechanism and a shear spring in each direction. It was assumed that both struts are parallel and are separated by a vertical distance equal to hz .

Figure 3.7: Inll panel element conguration,Crisafulli (1997)

21

Chapter 3. Literature Review

Figure 3.8: Shear spring conguration,Crisafulli (1997)

In order to simplify the application of the proposed model, three different sets of nodes are considered for the development of panel namely, External, Internal and Dummy nodes. External nodes are connected to the frame while internal nodes are dened by a horizontal and vertical offset xoi and yoi . The vertical and horizontal offsets are used to represent the reduction of the dimensions of the panel due to the depth of the frame members. Four dummy nodes are required to dene one end of the strut members, which is not connected to the corners of the panel. The details are shown in Figure 3.7 and Figure 3.8. One spring element is used to describe the shear behavior of the panel element. The spring is connected to two diagonally opposite internal nodes depending on the direction of the shear force. The limitation of the proposed model is that the potential plastic hinges which could develop along the length of the columns can not be considered in the model.

(b)

Compressive Strength fn

The expression for compressive strength of diagonal strut as

proposed by Decanini and Fantin (1987) can be estimated by following expression: Rc = fm Ams (3.8)

where Ams is the area of the equivalent strut i.e. Ams = bw t and fm is the strength of masonry when diagonally loaded at an inclination, . Mann and Muller (1982) developed a failure theory for unreinforced masonry subjected to shear 22

Chapter 3. Literature Review

and compressive stresses based on equilibrium considerations. Crisafulli (1997) modied the theory assuming a linear distribution of stress by considering the shear and normal stresses in the bed joint, and fn . Axial stresses parallel to the bed joint were neglected based on equilibrium consideration. Expressing the failure criterion in terms of principal stresses f1 and f2 , the following equations can be used: fn = f1 sin2 = f1 sin cos (3.9) (3.10)

In the above equation f1 is assumed to be positive. The above equations are derived under the assumption that principal stresses f2 developed in the masonry panel is not signicant.

Figure 3.9: Stress state considered to evaluate the strength of masonry,Crisafulli (1997)

(c)

Element Stiffness

The total element stiffness is distributed in a given proportion

to the strut and to the shear spring. The stiffness of the shear spring Ks can be calculated as a fraction s of the total stiffness of the masonry strut. Since the area of both struts are assumed to be equal such that the combination of the shear spring and the two masonry struts results in a total stiffness equal to the stiffness of the single strut model. The strut stiffness and shear stiffness are computed as follows: KS = s Ams Em cos2 dm Ams Et 2dm (3.11) (3.12)

KA = (1 s )

Tensile Strength, f t : Although analytical and experimental results show that the tensile strength of masonry is usually much smaller than the compressive strength, therefore, f t can be assumed to be zero in the analysis. However, research have shown that a value equal to 10% of the compressive strength can be approximated. It has been introduced in the model to gain generality. 23

Chapter 3. Literature Review

Strain at maximum stress, m : It represents the strain at maximum stress and its value vary from 0.002 to 0.005. It inuences via the modication of the secant stiffness of the ascending branch of stress-strain curve. Ultimate Strain, u : It controls the descending branch of the stress-strain curve. For a large value such as u = 20m , a smooth decrease of the compressive stress is obtained. Closing strain, cl : Closing strain presents the limiting strain at which cracks are closed partially and compressive stresses are resisted. For very large values such as cl = u is not considered in the analysis. Typical value range between 0 and 0.003. Elastic Modulus, Emo : This parameter represents the initial slope of the stress-strain curve. A large variation in its value is reported because of masonry being made up composite material, each have distinct properties. Empirical equations have been proposed for the evaluation of elastic modulus of masonry by various researchers. Most of these equations dene the secant modulus at a stress level between 1/3 and 2/3 of maximum compressive stress. According to (Crisafulli), these values can underestimate the initial stiffness of the inlled frame and to obtain a better approximation for the strength envelope, he assumed the following expression: Emo 2fm m (3.13)

3.1.5

Input Parameters

In order to describe the hysteric behavior of strut, several geometrical, mechanical and empirical parameters were described by (Crisafulli). Crisafulli proposed a range of recommended values obtained from experimental tests. A list of input parameters along with denition and recommended values are presented in this section: Unloading stiffness factor, un : This parameter controls the slope of the unloading branch of the envelope curve. The variation of this parameter modies the internal cycles, and not the envelope. Its value ranges from 1.5 to 2.5, however it is assumed to be un 1.0. Reloading strain factor, re : It denes the point on the strength envelope, where the reloading curve reach the strength envelope. Its value ranges from 0.2 to 0.4. Crisafulli (1997) used a value of 1.5 for non-linear analysis of inlled frames. Strain inection factor, ch : This parameter predicts the strain at which the reloading curve has an inection point. Its value range between 0.1 and 0.7. This parameter controls the fatness of the hysteresis loops. Complete unloading strain factor,a : This parameter denes the auxiliary point used to dene the plastic deformation after complete unloading. Its values typically range between 1.5 and 2.0. Stress inection factor, ch : It denes the stress at which the reloading curve exhibits an in24

Chapter 3. Literature Review

ection point. Its value vary from 0.5 to 0.9. Zero stress stiffness factor, plu : This parameter denes the modulus of the hysteric curve at zero stress after complete unloading in proportion to initial modulus. Its value vary between 0 and 1. Reloading stiffness factor, plr : It denes the reloading stiffness modulus, after complete loading has taken place. Its value is greater than one but typical value range between 1.1 to 1.5 to t experimental data. Plastic unloading stiffness factor, ex1 : It controls the inuence of un in the degradation of the stiffness. Values ranging from 1.5 and 2.0 have been used. Repeated cycle strain factor, ex2 : This parameter increases the strain at which the envelope curve is reached after an envelope curve is reached. Its value may typically vary between 1.0 and and 1.5. It is used to represent cumulative damage inside repeated cycles. Reduction shear factor, alphas : This empirical parameter represents the ratio between the maximum shear stress and the average stress in the panel, and may range between 1.4 and 1.65.

3.1.6

Reduction in area due to lateral displacements

It has been assumed in the proposed model that the area of the equivalent strut varies as a function of axial displacements, as shown in the Figure 3.10. The effect of strength and stiffness degradation due to cracking of the panel was not considered in the previous models. Even though there is insufcient information to estimate the practical values but it is introduced in the model to gain generality. Experimental results of Decanini and Fantin showed that the width of the strut decreases by about 20 % to 50 % due to cracking of the masonry inll panel. However, the results were obtained under the assumption that the modulus E remains constant, whereas the current model considers a variable modulus Crisafulli (1997). In the Figure 3.10, 1 and 2 , the strains for reduced and residual areas dene a linear reduction of the initial area, they can be taken as 1/10(m ) and 1/2 (m ) (Blandon 2005).

25

Chapter 3. Literature Review

Figure 3.10: Variation of area of strut as a function of the axial strain

3.1.7

Openings

Experimental tests have shown that presence of openings signicantly change the behaviour of inlled frames by reduction of strength and resistance. Due to the variability in location and size of openings, exact prediction of strength and stiffness of inlled frame is difcult to predict. (Mosalam and Ayala) observed that inlled frames with openings show more ductile behaviour and less strength then solid inll. He performed experiments on two structures, one having symmetric windows while other having symmetric doors. Initial stiffness was reduced in both the cases but he observed, frame with window openings show no reduction of the maximum strength while for frame with door as opening strength was reduced 20% approximately. Bertoldi et al. proposed a set of expressions for calculation of reduction coefcient rac . The expressions are given below: rac = 0.78e0.322 ln Aa + 0.93e0.762 ln Ac 1 In the above expression, Aa [%] = Ratio of opening area / inll area and Ac [%] = Ratio of opening length / inll length An inll panel with opening can be considered effective, if the following criteria is met: Aa (%) 25 Ac (%) 40 (3.15) (3.16) 26 (3.14)

Chapter 3. Literature Review

Due to a large number of variables and uncertainties involved, there is not much agreement on this topic among the research community. However the effect of openings can be taken into account by reducing the value of strut area A1 as described by (Smyrou 2006) that is,if a given inll panel features openings of 15% to 30% with respect to the area of the panel, good response predictions might be obtained by reducing the value of A1 (i.e. its stiffness) by a value that varies between 30% and 50%.

3.2

RIGID DIAPHRAGM
To enforce certain types of rigid body behaviour constraints are used. They are use to

connect different parts of the model. The advantage of using constraints is that it reduces the number of equations to be solved and will usually result in increased computational efciency. The type of behaviours that can enforced by constraint are, naming a few such as rigid body, rigid diaphragm, etc. A rigid diaphragm constraint causes all of its constrained joints to move together as a planar diaphragm that is rigid against membrane deformation. Effectively all constrained joints are connected to each other by links that are rigid in plane but do not effect out of plane. This constraint can be used to model:

concrete oors in building structures, which typically have very large in-plane stiffness. diaphragm in bridge superstructure.

The use of diaphragm constraint for building structure eliminates the numerical accuracy problem created, when large in-plane stiffness of a oor diaphragm is modeled with a membrane element. It is also very useful in the lateral dynamic analysis of buildings, as it results in a signicant reduction in the size of the eigen value problem to be solved (Computers and Structures. 2004).

27

Chapter 3. Literature Review

Figure 3.11: Diaphragm constraint to model rigid oor slab, Adapted from (Computers and Structures. 2004)

28

Chapter 4. Numerical Model of Case Study

4.

NUMERICAL MODEL OF CASE STUDY

In this chapter numerical model of the scaled building is presented. The non-linear nite element program used for current work is SeismoStruct (SeismoSoft 2008). SeismoStruct is a bre element-based program freely available from Internet. Modeling of any kind of structures is easily done because of its Graphical user Interface. Also the calibration of inll panel and numerical model of the structure in SeismoStruct is presented in detail.

4.1

SEISMOSTRUCT
SeismoStruct is a Finite Element package capable of predicting the large displacement

behavior of space frames under static or dynamic loading, taking into account both geometric nonlinearities and material inelasticity. SeismoStruct accepts static as well as dynamic actions and has the ability to perform eigenvalue, nonlinear static pushover (conventional and adaptive), nonlinear static time-history analysis, nonlinear dynamic analysis and incremental dynamic analysis. To model geometric nonlinearity, both local (beam-column effect) and global (large displacements/rotations effects) sources of geometric nonlinearity are automatically taken into account in SeismoStruct. Since a constant generalised axial strain shape function is assumed in the adopted cubic formulation, it results that its application is only fully valid to model the nonlinear response of relatively short members and hence a number of elements (3-4 per structural member) is required for the accurate modeling of structural frame members. For accurate estimation of structural damage distribution, the spread of material inelasticity along the member length and across the section area is explicitly represented through the employment of a bre modeling approach as shown in Figure 4.1. The sectional stress-strain state of beam-column elements is obtained through the integration of the nonlinear uni-axial 29

Chapter 4. Numerical Model of Case Study

stress-strain response of the individual bers in which the section has been subdivided. If a sufcient number of bres (200-400 in spatial analysis) is employed, the distribution of material nonlinearity across the section area is accurately modeled. Two integration Gauss points per element are then used for the numerical integration of the governing equations of the cubic formulation. If a sufcient number of elements is used (5-6 per structural member) the spread of inelasticity along member length can be accurately estimated.

Figure 4.1: Material Inelasticity

4.2

NUMERICAL MODEL OF STRUCTURE


The software has been used extensively in the past by many authors for verifying the

experimental model with the numerical one, however none of the studies carried out modeling of 3D irregular frames with inll panel and rigid diaphragm. Therefore the current study tries to verify the response of FE model with experimental results. Uni-axial nonlinear constant connement model, that follows the constitutive relationship proposed by (Mander, Priestley, and Park 1988) and the cyclic rules proposed by MartinezRueda and Elnashai (1997) is used for modeling concrete . The connement effects provided by the lateral transverse reinforcement are incorporated through the rules proposed by (Mander, Priestley, and Park 1988) whereby constant conning pressure is assumed throughout the entire stress-strain range. Five parameters must be dened to describe the model completely. These are Compressive strength, tensile strength,strain at peak stress,connement factor and specic weight. The connement factor (kc ) is the ratio between the conned and unconned compressive stress of the concrete. In current study its value is taken equal to 1.0 because transverse 30

Chapter 4. Numerical Model of Case Study

reinforcement of all members is very small to produce an effective concrete connement. The compressive strength values used for the model are presented in Table 2.2. The values for other parameters are given below as they are constant for all sections.

Tensile strength, ft = 0.001 N/mm2 Strain at peak stress, c = 0.002 mm/mm Connement factor, kc = 1.0 Specic weight, = 24 kN/m3

A uni-axial bilinear stress-strain model with kinematic strain hardening is used to for steel. Three parameters must be dened to describe the model completely, modulus of elasticity, yield strength, and strain hardening parameter. Values of yield strength for the model are given in Table 2.3. The values for others parameters are given below:

Elastic Modulus, Es = 200000 N/mm2 Strain hardening, = 0.004

Inelastic frame elements are used to model 3D-beam column elements both with displacement base formulation and force base formulation. The number of section bres for displacement base formulation or integration points for force base formulation used in section equilibrium computations carried out at each of the elements Gauss section needs to be dened. The ideal number of section bers, sufcient to guarantee an adequate reproduction of the stress-strain distribution across the elements cross-section, varies with the shape and material characteristics of the latter, depending also on the degree of inelasticity to which the element will be forced to. For the current study, number of section bers used for displacement base formulation is 300 whereas number of integration points use are 4 for force base formulation. The distributed mass applied to beam and column elements consists of self weight of the frame and additional masses obtained from similitude condition. The total mass of the structure is 51.83 tons. In SeismoStruct geometry of the structure being modeled is a threestep procedure. In the rst step all structural and non-structural nodes are dened, after which nodes are connected through element connectivity and nally the process is then concluded with the assignment of structural restraints, which fully characterise the structures boundary conditions. Centerline dimensions for beams and columns are used to accurately model the 31

Chapter 4. Numerical Model of Case Study

structure. All beams are 125 x 250mm while except column C2 which is 125 x 375mm the remaining columns are 125 x 125mm. Rigid links are used to connect elements at column C2 as shown in the Figure 4.2.

Figure 4.2: Rigid Link connection at column C2

Finally, to correctly distribute the inertia forces in the structure, rigid diaphragm approach is used. In SeismoStruct, nodal constraints are implemented through the use of either Penalty Functions or Lagrange Multipliers. For the present study nodal constraints approach that makes use of Lagrange Multiplier is implemented as suggested by Pinho et al. (2008). A 3D view of the structure both with Displacement base formulation and Force base element formulation is shown in Figure 4.3.

32

Chapter 4. Numerical Model of Case Study

(a)

(b)

Figure 4.3: (a) Displacement-Based Element Formulation, (b) Force-Based Element Formulation

4.3

CALIBRATION OF INFILL PANEL


Due to the complex nature of the masonry inll panel, the model is calibrated with the

experimental results. The experimental test was performed on a single bay single storey 2D reinforced concrete frame with inll panel. It was conducted in the laboratory of Department of Structural Mechanics, University of Pavia. The inll was constructed from masonry blocks of cellular RDB Gasbeston having a height of 250 mm, thickness of 300 mm and length of 625 mm. The layout of the wall along with the reinforcement detail of beam and columns is presented in Figure 4.4 and Figure 4.5, Penna (2006).

33

Chapter 4. Numerical Model of Case Study

Figure 4.4: Layout of Inll panel adapted from Penna (2006)

Figure 4.5: Reinforcement Detail of Beam and Columns,(adapted from Penna (2006))

34

Chapter 4. Numerical Model of Case Study

The cyclic test was performed by repeated cycles of loading until a horizontal displacement of 33mm (1.2 % of panel) was raeched. The results of the test interms of force-horizontal displacement is shown in Figure 4.6.

Force Displacement 500 400 300 200 Force [kN] 100 0 100 200 300 400 500 40 1.20% 1.00% 0.80% 0.60% 0.40% 0.20% 0.10% Interp

30

20

10 0 10 Displacement[mm]

20

30

40

Figure 4.6: Force-Displacement,(adapted from Penna (2006))

To accurately measure the response of the numerical model with the experimental results, Inelastic inll panel are included in the model. A four-node masonry panel element, developed by (Crisafulli 1997) and implemented in SeismoStruct by (Blandon 2005), is used for the modeling of the nonlinear response of inll panels in framed structures. Each panel is represented by six strut members; each diagonal direction features two parallel struts to carry axial loads across two opposite diagonal corners and a third one to carry the shear from the top to the bottom of the panel as shown in Figure 3.7 and Figure 3.8. The axial load struts use the masonry strut hysteresis model, while the shear strut uses a dedicated bilinear hysteresis rule. The empirical parameters as described in Section 3.1.5 were rst calibrated with experimental results. Axial load struts use seventeen parameters to dene strut hysteresis model while shear strut uses four empirical parameters to completely dene the bilinear hysteresis model. Other than these empirical parameters, there are few geometrical and material parameters. A detail overview of both the parameters are given in Section 3.1. 35

Chapter 4. Numerical Model of Case Study

The default values provided in SeismoStruct help and recommended by Smyrou (2006) were used as an initial trial. Initial few trails were run to adjust the value of compressive strength and Initial Young Modulus to control the maximum values of base shear force. The values are presented in Table 4.1 while the results of the analysis are shown in Figure 4.7. Default Values Em m 1 re ch ex1 1900 0.0012 0.0003 0.2 0.9 3.0 0.7 fm u 2 ch plu ex2 max 1.25 0.024 0.0006 0.7 1.0 1.0 1.0 ft cl un a plr o s 0.575 0.003 1.7 2.0 1.1 0.3 1.5

Table 4.1: Default Values for Calibration of Inll Panel

Force Displacement 500 400 300 200 Force [kN] 100 0 100 200 300 400 500 40 Measured SeismoStrcut

30

20

10 0 10 Displacement[mm]

20

30

40

Figure 4.7: Force-Displacement, Trial 1

As mentioned in the study of Smyrou (2006), the most inuential parameters that controls the shape of hysteresis loops, fatness of the loops and the tangent modulus corresponding to the plastic strain are un , ch and ex1 . These three parameters infact decide the amount of energy 36

Chapter 4. Numerical Model of Case Study

dissipated during cyclic loading. Therefore a second trial was done by changing the values of these three parameters in the range give by (Crisafulli 1997). The values are provided in Table 4.2 while the result of the analysis are given in Figure 4.8. Intermediate Values Em m 1 re ch ex1 1900 0.0012 0.0003 0.2 0.9 1.75 0.5 fm u 2 ch plu ex2 max 1.25 0.024 0.0006 0.1 1.0 1.5 1.25 ft cl un a plr o s 0.2 0.003 2.5 1.5 1.5 0.3 1.25

Table 4.2: Intermediate Values for Calibration of Inll Panel

Force Displacement 500 400 300 200 Force [kN] 100 0 100 200 300 400 500 40 Measured SeismoStrcut

30

20

10 0 10 Displacement[mm]

20

30

40

Figure 4.8: Force-Displacement, Trial 2

Analysis results of second trial shows that the unloading part does not match exactly the experimental results and some pinching is still observed during the unloading and reloading part. To remove those pinching, parameters which were less inuential were modied to accurately represent the experimental results, such as closing strain, strut area reduction strain and residual 37

Chapter 4. Numerical Model of Case Study

strut area strain. Out of range value for closing strain,cl = 0.004 is used to better approximate the experimental result. Strut area reduction strain 1 and residual strut area strain 2 values used are 0.0001 and 0.001. Furthermore the value of unloading stiffness factor is change to default value while strain inection factor is taken as 0.1, which is same as supposed in second trail. The nal results are presented in Table 4.3 and results of the analysis are shown in Figure 4.9. Final Values Em m 1 re ch ex1 1900 0.0018 0.0001 0.2 0.9 1.75 0.5 fm u 2 ch plu ex2 max 1.25 0.076 0.001 0.1 1.0 1.5 1.25 ft cl un a plr o s 0.2 0.004 1.7 1.5 1.5 0.3 1.25

Table 4.3: Final Values for Calibration of Inll Panel

Force Displacement 500 400 300 200 Force [kN] 100 0 100 200 300 400 500 40 Measured SeismoStrcut

30

20

10 0 10 Displacement[mm]

20

30

40

Figure 4.9: Force-Displacement, Final Calibrated

The important parameters that contributed most in the calibration of the inll panel are presented in the Table 4.4. 38

Chapter 4. Numerical Model of Case Study

Parameters Contributing Most to Inll Panel Calibration Parameters m m cl 1 2 un ch ex1 Default 0.0012 0.024 0.003 0.0003 0.0006 1.7 0.7 3.0 Intermediate 0.0012 0.024 0.003 0.0003 0.0006 2.5 0.1 1.75 Final 0.0018 0.076 0.004 0.0001 0.001 1.7 0.1 1.75

Table 4.4: Parameters contributing most

4.3.1

Geometrical Parameters

Once the inll panel parameters for experimental results were calibrated in SeismoStruct, they were used for the current analysis. However the geometrical parameters were updated according to the geometry of the structure. These are described below: Thickness of the panel: As described in Section 2.2.5, the inll panels are made up of cellular concrete Gasbeton RDB having thickness 50 mm. Horizontal and vertical offset: The horizontal and vertical offset is calculated from the beam and column depth. Since centerline dimension are used, external nodes are taken as the beam column joints. The vertical offset for all panels is equal to half the beam depth divided by the net height of the inll panel, i.e. 10%. In case of horizontal offset, the value of xoi varies since the bay length is different. This could be divided into three groups depending on the length of the bay i.e. 1.5m, 3m and 2.75m. For simplicity and not too much difference in the results the value of xoi for 3 m and 2.75 m bay length are taken equal, also the xoi value for the 3m bay length governs. Therefore the horizontal offset values for the 3m bay length is 2.2% while for 1.5m bay length is 4.5%. Equivalent contact length, hz : This parameter takes into account, the contact length between the frame and inll panel, introduced as percentage of the vertical height of the panel, effectively yielding the distance between the internal and dummy nodes as shown in Figure 3.7. Values between 1/3 and 1/2 of actual contact length (z) may be used. The values nally implemented for hz are 23% and 21% for 3m and 1.5m bay length respectively.

39

Chapter 4. Numerical Model of Case Study

Area of the Strut, A1 : This parameter is dened as the product of the panel thickness and the equivalent width of the strut (bw ) which normally varies between 10 - 40 % of the diagonal of the inll panel. Area of the strut has a direct inuence on the stiffness of the inll panel and similarly on the structure. Several expressions are developed by researchers for the evaluation of the width of the equivalent strut, for the current study the width of equivalent of strut is taken equal to 12 % of length of the diagonal. A review of these expressions is given in Section 3.1.2. Since the thickness of the panel is known, the area of the struts was computed eaisly from the above consideration. Area of the Strut, A2 : This parameter is taken as a percentage of Strut area A1 . It is assumed that the area varies linearly as function of the axial strain as shown in Figure 3.10. The two strains between which this variation takes place are dened in Section 3.1.6. The values of the two strains used for the current study are 0.0001 and 0.001 respectively.

40

Chapter 5. Analysis and Results

5.

ANALYSIS AND RESULTS

This chapter summarizes the verication of the analysis and results with experimental data. The chapter starts with the Eigenvalue analysis (EVA), results of the rst four modes of the model are presented. The second part of the chapter deals with the Dynamic Time History (DTH) analysis. The analysis is performed both for Displacement based element formulation and Force based element formulation. Finally the results of the analysis are compared with the experimental results.

5.1

MODAL ANALYSIS
To check the model reliability, elastic periods of vibrations and the corresponding de-

formed shapes are computed. Seismostruct uses the efcient Lanczos algorithm for the evaluation of the structural natural frequencies and mode shapes. The experimental results are obtained by applying a low level PGA (0.02g) to the structure and calculating the natural frequencies and mode shapes. The experimental data is available only for rst two fundamental frequencies while higher frequencies were difcult to obtain from signals. Table 5.1 shows the comparison of vibration period obtained from numerical model and experimental test. In the Table 5.1, longitudinal direction refers to the direction of accelerogram both for numerical and experimental model while transverse direction refers transverse to the applied accelerogram. Vibration Periods (Sec) Numerical 0.19 0.20
Table 5.1: Estimated Vibration Periods

Direction Longitudinal Transverse

Experimental 0.15 0.20

41

Chapter 5. Analysis and Results

Fundamental period dominates in both the cases. It can be observed that the period of vibration in transverse direction matches well with the experimental result, however the difference in the values of the longitudinal direction. The experimental model seems to be more stiff in the longitudinal direction. Furthermore the results were checked with the Height wise empirical equation such as given by Crowley and Pinho (2006). The results of the numerical model are in close agreement with the empirical equation, Table 5.2 shows the comparison of results from empirical equation and modal analysis. Deformed shapes of rst four modes are given in Figure ??. Reference Crowley and Pinho (2006) Numerical Model Formula Ty = 0.043H (H in meters) SeismoStruct Value (Sec) 0.19 0.19

Table 5.2: Empirical Eq. and SeismoStruct result

5.2

NON-LINEAR TIME HISTORY ANALYSIS (NTHA)

Once the numerical model is veried with EigenValue analysis, the second step is to check the results of the shake table test with the Non-linear Time History Analysis (NTHA) of the numerical model. The numerical model is subjected to acceleration record of PGA 0.3g and 0.54g. The Accelerogram applied to the numerical model is shown in Section (a) and it is applied in one direction only. The displacements obtained for each oor from seismic records are compared with those measured during the shake table test.

42

Chapter 5. Analysis and Results

Figure 5.1: Experimental Results for Col 6

43

Chapter 5. Analysis and Results

Figure 5.2: Comparison of Displacement Time History for Col 6

44

Chapter 6. Conclusions and Recommendations

6.

CONCLUSIONS AND RECOMMENDATIONS

6.1

CONCLUSIONS

Effect of Inll panel have widely been ignored in the non-linear analysis of inlled frame structures, mainly due to the complexity and uncertainty involved in the analysis procedure. However recent studies have shown that inclusion of masonry inll in non-linear structural analysis greatly inuence not only on the strength and stiffness but also on the energy dissipation mechanism of the structure. The objective of this analytical investigation was to compare the response of a numerical model having masonry inll and rigid diaphragm with the experimental results. The experimental building is an irregular half scale three-storey inlled reinforced concrete building tested at TREES Lab EUCENTRE. The building is typical of older construction without earthquake resistant provision and designed only for gravity loads. It is similar to the SPEAR structure but with major differences, such as half scale (1:2) and provision of inll panel with openings and earthquake excitation using shaking table. SeismoStruct a bre element-based program is used for the for the numerical modeling of the building. The numerical model is done both for Force-Based element formulation and Displacement-Based element formulation. Inll panel is modeled using the model developed by Crisafulli 1997 and implemented by Blandon 2005 in SeismoStruct. The 4-node masonry panel element takes into account both the compressive and shear behaviour of panel, using two parallel struts and a shear spring in each direction. The conguration of the model adequately considers the lateral stiffness of the panel and masonry strength of panel. The inll panel model is rst calibrated with the experimental results. It requires 19 parameters to fully calibrate the model, however some default values have been used which corresponds to better approximation of experimental and numerical results.

45

Chapter 6. Conclusions and Recommendations

One of the main features of the numerical model is the use of rigid diaphragm. Rigid diaphragm is implemented in the model using nodal constraints with Lagrange multipliers, this option provides better response of numerical results than the usual truss model as shown by Pinho et al. 2008. This approach of modeling slab as rigid diaphragm is a reliable way of modeling the presence of very stiff oors in this type of irregular structures. The result of the both eigenvalue analysis and non-linear time history analysis shows that the response of numerical model is in good agreement with the experimental results. Though certain discrepancies are observed in the eigenvalue analysis but it has to bear in mind the complex nature of masonry inll panel model.

6.2

FUTURE DEVELOPMENTS

There are many issues which need to be addressed in future research work. A few of them are listed below:

The consideration regarding to take into account the effect openings in the calculations is very important. Lack of recommendations in this regard greatly effect on the response of the structure. Calibration of inll panel model with experimental results shows that the model works good for the loading and unloading part, however it still needs to be improved for the reloading part. Force-Based Element formulation recently implemented in SeismoStruct is used for analysis, which accounts for less computation time and showed good results. Verication of the formulation with different geometry structure especially with masonry inll panel will be of great interest.

46

References

REFERENCES

Bertoldi, S., L. Decanini, P. Santinti, and G. Via (1994). Analytical model in inlled frames. Proceedings of the 10th European Conference on Earthquake Engineering Aienna, Austia. Blandon, C. (2005). Implementation of an inll masonry model for seismic assessment of existing buildings. Individual study, Pavia, Italy. Chrysostomou, C. (1991). Effects of Degrading Inll Walls on the Non-Linear Seismic Response of Two-Dimensional Steel Frames. Phd thesis, Cornell, United States. Computers, S. and Structures. (2004). CSI Analysis Reference Manual for SAP2000, ETABS & SAFE. C&S, Inc. Wiley & Sons, USA. Crisafulli, F. (1997). Seismic Behaviour of Reinforced Structures with Reinforced Masonry Inills. Phd thesis, University of Canterbury, New Zeland. Crowley, H. and R. Pinho (2006, September). Simplied equations for estimating the period of vibration of existing buildings. In First European Conference on Earthquake Engineering and Seismology, Geneva, Switzerland, pp. 1122. Decanini, L. and G. Fantin (1987). Modelos simplicados de lamamposteria incluida en porticos. caracteristicas de rigidez y resitencia lateral en estado limite. Jornadas Argentinas de Inteneieria Estructural Buenos Aires, Argentina, 817836. Dhanasekar, M. (1985). Behaviour of brick masonry under biaxial stress with particular reference to inlled frames. Proceedings of the 7th International Brick Masonry Conference Melbourne, Australia. Fardis, M. (2002). Design of an irregular building for the spear project-Description of the 3-storey structure. Technical report, University of Patras, Greece. Gooman, R. (1968). A model for the mechanics of jointed rock. American society of Civil Engineers Journal of Soil Mechhanics(94), 637659. Holmes, M. (1961). Steel frames with brickwork and concrete inlling. Proceedings of the Institute of Civil Engineers 19, 473478.

47

References

KIng, G. and P. Pandey (1978). Analysis of inlled frames using nite elements. Proceedings of the Institute of Civil Engineers 65, 749760. Klingner, R. and V. Berter (1978). Earthquake resistance of inlled frames. ASCE Journal of the Structural Division Vol. 100, pp. 973989. Liauw, T. and K. Kwan (1984). Nonlinear behaviour of non-integral inlled frames. Computers and Structures 18, 551560. Mainstone, R. (1971). On the stiffnesses and strengths of inlled frames. Proceedings of the Institute of Civil Engineers Supplement IV, 5790. Mallick, D. and R. Severn (1967). The behavior of inlled frames under static loading. Proceedings of the Institute of Civil Engineers 49, 193209. Mander, J., M. Priestley, and R. Park (1988). Theoretical stress strain model for conned concrete. Journal of Structural Engineering 114, 18041826. Mann, W. and H. Muller (1982). Failure of shear-stresses masonry - an enlarge theory, tets and application to shear walls. Proceedings of the British Ceramic Society 30, 139149. Marazzi, F., A. Pavese, L. Petrini, and R. Nascimbene (2007, Jun). Comparison between experimental and numerical models of a full scale rc-3 storey structure. Proceedings of the Thematic Conference on Computational Methods in Structural Dynamics and Earthquake Engineering Rethymno, Crete, Greece. Martinez-Rueda, J. and A. Elnashai (1997). Conned concrete model under cyclic load. Material and Structures 30(197), 139147. Molina, F., P. Bauchet, G. Magonette, O. Hubert, and P. Negro (2004, August). Bidirectional pseudodynamic technique for testing a three storey reinforced concrete building. Proceedings of 13th World Conference on Earthquake Engineering Vancouver, Canada(Paper N.75). Molina, F., P. Bauchet, G. Magonette, O. Hubert, and P. Negro (2005, April). Full scale bidirectional psd testing of the torsionally unbalanced spear structure: method, algorithm and experimental set-up. Proceedings of the International Workshop on SPEAR project Ispra, Italy. Mosalam, K.M, P. R. and A. Ayala (1996). Effect of geometrical conguration on the seismic response of inlled frame. In Elsevier Science Ltd, 11th World Conference on Earthquake Engineering, Number 1675, Mexico. Negro, P., F. Molina, and G. Magonette (2004, August). Full scale psd tesing of a torsionally unbalanced three-storey non-seismic rc frame. Proceedings of 13th World Conference on Earthquake Engineering Vancouver, Canada(Paper N.968).

48

References

Paulay, T. and M. Priestley (1992). Seismic Design of Reinforced Concrete and Masonry Buildings. New York, United States: John Wiley & Sons. Penna, A. (2006). Campagna sperimentale su telai in calcestruzzo armato con tamponamenti in calcestruzzo cellulare e diverse soluzioni di rinforzo. Technical report, EUCENTRECentro Europeo di Formazione e Ricerca in Ingegneria Sismica. Pinho, R., C. Bhatt, S. Antoniou, and R. Bento (2008, Oct 1217). Modelling of the horizontal slab of a 3D irregular building for nonlinear static assessmen. Proceedings of the 14th World Conference on Earthquake Engineering Beijing, China. Polyakov, S. (1956). Masonry in framed buildings. Gosudarstvennoe Izdatel Stvo Literatury Po Stroitel Stvui Arkitektuze Moscow. Translated by Crains G.L. Riddington, J. and B. Stafford Smith (1977). Analysis of inlled frames subject to racking with design recommendations. American society of Civil Engineers Journal of Structural Engineering(52), 263268. Rivero, C. and W. Walker (1984). An analytical study of the interaction of frames and inll masonry walls. Proceedings of the 8th World Conference on Earthquake Engineering San Franciso, United States, 551560. SeismoSoft (2008). SeismoStruct - A Computer Program for Static and Dynamic Nonlinear Analysis of Framed Structures. [URL: www.seismosoft.com]. Shing, P. (1992). Finite element analysis of shear resistance of masonry wall panels with and without conning frames. Proceedings of the 10th World Conference on Earthquake Engineering Madrid, Spain. Smyrou, E. (2006). Implementation and verication of a masonry panel model for non-linear dynamic analysis of inlled rc frame. Master thesis, European School for Advance Studies In Reduction of Seismic Risk, ROSE SCHOOL, Pavia, Italy. Stafford, S. B. (1962). Lateral stiffness of inlled frames. Proceedings of the American society of Civil Engineers Journal of Structural Division(ST6), 183191. Stafford, S. B. (1966). Behaviour of squared inlled frames. ASCE Journal of Structural Division 92(ST1), 381403. Sullivan, T., R. Pinho, and A. Pavese (2004). An introduction to structural testing techniques in earthquake engineering. Educational Report IUSS Press(ROSE 2004/01). Pavia, Italy. Syrmkezis, C. and P. Asteris (2001). Masonry failure criterion under biaxial stress state. ASCE Journal of Materials in Civil Engineering 13, 5864.

49

Vous aimerez peut-être aussi