3-D FINITE ELEMENT MODELING OF REINFORCED CONCRETE BEAM-COLUMN CONNECTIONS – DEVELOPMENT AND COMPARISON TO NCHRP PROJECT 12-74 A Project Presented to the faculty of the Department of Civil Engineering California State University, Sacramento Submitted in partial satisfaction of the requirements for the degree of MASTER OF SCIENCE in Civil Engineering by Harpreet Singh Hansra FALL 2012 © 2012 Harpreet Singh Hansra ALL RIGHTS RESERVED ii 3-D FINITE ELEMENT MODELING OF REINFORCED CONCRETE BEAM-COLUMN CONNECTIONS – DEVELOPMENT AND COMPARISON TO NCHRP PROJECT 12-74 A Project by Harpreet Singh Hansra Approved by: __________________________________, Committee Chair Eric E. Matsumoto, Ph.D., P.E. __________________________________, Second Reader Mark Schultz, M.S., P.E., S.E. ____________________________ Date iii Student: Harpreet Singh Hansra I certify that this student has met the requirements for format contained in the University format manual, and that this project is suitable for shelving in the Library and credit is to be awarded for the project. ________________________________, Department Chair ____________________ Kevan Shafizadeh, Ph.D., P.E., PTOE Date Department of Civil Engineering iv Abstract of 3-D FINITE ELEMENT MODELING OF REINFORCED CONCRETE BEAM-COLUMN CONNECTIONS – DEVELOPMENT AND COMPARISON TO NCHRP PROJECT 12-74 by Harpreet Singh Hansra This report, 3-D Finite Element Modeling of Reinforced Concrete Beam-Column Connections – Development and Comparison to NCHRP 12-74, investigates the use of finite element modeling (FEM) to predict the structural response of the cast-in-place (CIP) reinforced concrete bent cap-column test specimen reported in NCHRP Report 681 – Development of a Precast Bent Cap System for Seismic Regions. Analysis was performed using LS-DYNA as the finite element processor. The Karagozian & Case Damaged Concrete model, material MAT_072, was used as the constitutive model for all concrete elements and material MAT_003, a plastic kinematic model, was used as the constitutive model for the reinforcing steel. Strain-hardening effects of steel were neglected for this analysis. Boundary conditions on the FE model were identical to the vertical and horizontal restraints used on the CIP specimen during testing. The FE model only considered a monotonic push loading sequence, whereas the CIP specimen was subjected to reverse cyclic loading. To account for the difference in v loading, the FE model results were compared to the hysteretic envelope from the CIP specimen. The lateral load-lateral displacement response of the FE model (Model 1) compared reasonably well to the actual and theoretically predicted response of the CIP specimen. For lateral displacements less than that corresponding to a displacement ductility of 4.1, the FE model showed a larger stiffness than the actual CIP response. The model stiffness degraded as a greater number of concrete elements in the column plastic hinging region accumulated damage. The degradation and lateral load-displacement response matched the predicted response within 5% for a displacement ductility larger than 2.0; however, the model degradation was not as severe as that observed for the CIP specimen. Concrete damage in the FE model correlated reasonably well with observed cracking and spalling of the CIP specimen. Significant damage was observed in the column of the FE model, near the joint, reflecting flexural cracking. Initial yielding of column longitudinal bars in the FE model occurred at a displacement ductility 26% larger than the CIP specimen. Based on contours of concrete damage and principal stress vectors, the primary shear crack formed diagonally through the joint of the FE model at a lateral load 6% higher than that of the CIP specimen. Joint rotation for the FE model was significantly less than that of the CIP specimen, approximately half of the specimen values. Conclusions include: 1) finite element modeling using appropriate constitutive models and element formulation can accurately capture the nonlinear behavior of vi reinforced concrete beam-column connections, including flexural cracking, joint shear cracking, steel reinforcement yielding and overall stress distribution; 2) element size for concrete and steel reinforcement significantly impacts the overall response and accuracy of results and therefore must be carefully selected for convergence; 3) the Karagozian & Case damaged concrete model, material MAT_072, can accurately capture the cracking of concrete using limited inputs (f ’c and aggregate size). Recommendations include: 1) additional analysis should be performed to appropriately incorporate a strain hardening model for the reinforcing steel; 2) strain distribution of the steel reinforcement in the joint (longitudinal reinforcement, joint hoops, and joint stirrups) should be further investigated as well as the hoop strain distribution in the column plastic hinge region; 3) a concrete constitutive model capable of reverse cyclic loading should be investigated; 4) a bar slip model for bond between the concrete and reinforcing steel should be investigated. _______________________, Committee Chair Eric E. Matsumoto, Ph.D., P.E. _______________________ Date vii TABLE OF CONTENTS Page List of Tables ............................................................................................................................ x List of Figures ........................................................................................................................ xii Chapter 1. INTRODUCTION ............................................................................................................... 1 1.1 Background ............................................................................................................ 1 1.2 Project Objective.................................................................................................... 3 1.3 Significance of Project Results .............................................................................. 3 1.4 Literature Review .................................................................................................. 4 1.5 Project Approach ................................................................................................... 7 1.6 Scope of Report ..................................................................................................... 8 2. DEVELOPMENT OF FINITE ELEMENT MODEL ...................................................... 11 2.1 CIP Specimen Information ................................................................................... 11 2.2 Finite Element Model Geometry ......................................................................... 12 2.3 Element Formulation ........................................................................................... 13 2.4 Constitutive Models ............................................................................................. 14 2.5 Mesh Development .............................................................................................. 16 2.6 Boundary Conditions ........................................................................................... 18 2.7 Loading Application ............................................................................................ 18 3. FINITE ELEMENT ANALYSIS RESULTS ................................................................... 34 3.1 Lateral Load-Lateral Displacement Response ..................................................... 34 3.2 Concrete Damage Parameter................................................................................ 36 3.3 Principal Stress Vectors ....................................................................................... 39 viii 3.4 Joint Shear Stress ................................................................................................. 39 3.5 Joint Rotation ........................................................................................................ 40 3.6 Longitudinal and Principal Stress Distribution ..................................................... 41 3.7 Response of Reinforcing Steel .............................................................................. 43 4. COMPARISON OF RESULTS TO SPECIMEN DATA ................................................. 63 4.1 Lateral Load-Lateral Displacement Response ...................................................... 63 4.2 Joint Shear Stress ................................................................................................. 66 4.3 Joint Rotation ....................................................................................................... 67 4.4 Principal Stress .................................................................................................... 67 4.5 Stages of Concrete Cracking ................................................................................ 68 4.6 Reinforcing Steel Stain Profiles ........................................................................... 69 5. SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS ..................................... 84 5.1 Summary .............................................................................................................. 84 5.2 Conclusion ........................................................................................................... 86 5.3 Recommendations ................................................................................................ 87 References ............................................................................................................................... 89 ix LIST OF TABLES Page Table 2.1 A) Summary of CIP Concrete Properties B) Comparison of GD and CIP properties .................................................................................... 19 Table 2.2 Summary of Reinforcement for Bent Cap, Joint, and Column ....................... 20 Table 2.3 Loading Sequence for FEM ............................................................................ 21 Table 3.1 Summary of Stresses in Concrete Elements ................................................... 45 Table 3.2 Summary of Stresses in Reinforcing Steel...................................................... 45 Table 4.1 Comparison of Lateral Load-Lateral Displacement Response between Model 1, CIP Specimen, GD Specimen, and Specimen Predictions .............. 71 Table 4.2 Comparison of Lateral Load-Lateral Displacement Response between Model 2, CIP Specimen, GD Specimen, and Specimen Predictions .............. 71 Table 4.3 Comparison of Average Joint Shear Stress between Model 1, CIP Specimen, and the GD Specimen.................................................................... 71 Table 4.4 Comparison of Joint Rotation between Model 1, the CIP Specimen, and the GD Specimen ..................................................................................... 72 Table 4.5 Comparison of Specimen Cracking – Including 38 kip Axial Load and Self-Weight ..................................................................................................... 72 Table 4.6 Strain Profile Table – Column Longitudinal Rebar (LC8), CIP Specimen ................................................................................................ 73 Table 4.7 Strain Profile Table – Stirrups in Bent Cap (Mid-height) and Joint (Top), CIP Specimen.............................................................................. 74 x Table 4.8 Strain Profile Table – Stirrups in Bent Cap (Mid-height) and Joint (Bottom), CIP Specimen ................................................................................. 75 Table 4.9 Strain Profile Table – Hoops in Column and Joint (East), CIP Specimen ................................................................................................. 75 xi LIST OF FIGURES Page Figure 1.1 Prototype Structure from NCHRP Project 12-74.............................................. 9 Figure 1.2 Portion of Prototype Used for Testing and Modeling ..................................... 10 Figure 2.1 Elevation of CIP Specimen Dimensions ......................................................... 22 Figure 2.2 Cross Sections of CIP Specimen Showing Rebar ........................................... 23 Figure 2.3 Column with Instrumentation in Place Prior to Casting of Bent Cap ............. 24 Figure 2.4 Bent Cap Formwork Placed Over Column ..................................................... 25 Figure 2.5 Test Set-up ...................................................................................................... 26 Figure 2.6 Individual Bent Cap and Column Components .............................................. 27 Figure 2.7 Dimensions of FE Model ................................................................................ 27 Figure 2.8 A) Rebar configuration in FE Model, B) CIP Specimen Bent Cap Rebar Cage ................................................................................................................ 28 Figure 2.9 Half of Bent Cap Cut-away (with Stirrups Removed) to Show Longitudinal Rebar Connecting the Bent Cap and Column ........................... 29 Figure 2.10 Mesh of Solid Elements .................................................................................. 29 Figure 2.11 3-D Isometric View of Meshed Structure ....................................................... 30 Figure 2.12 Close-up of Meshed Beam Elements in Joint Region..................................... 31 Figure 2.13 Strength Model for Concrete (Malvar et al., 1997): (a) Failure Surfaces in Concrete Model; (b) Concrete Constitutive Model ................. ….31 Figure 2.14 Comparison of #3 Hoop Steel Element Length in Model 1and Model 2 ....... 32 Figure 2.15 Vertical and Horizontal Boundary Conditions ............................................... 32 Figure 2.16 Boundary Conditions for Axial Load and Lateral Displacement .................... 33 xii Figure 3.1 Model 1 Lateral Load-Lateral Displacement Curve ....................................... 46 Figure 3.2 Model 2 Lateral Load-Lateral Displacement Curve ....................................... 46 Figure 3.3 Comparison of Lateral Load-Lateral Displacement Response for Models 1 and 2................................................................................................ 47 Figure 3.4 Relationship between Damage and Strength (compression plotted as positive stress to show relationship for representative element in column plastic hinging region) .................................................................................... 47 Figure 3.5 Relationship between Tensile Strength and Damage (representative element in column plastic hinging region)...................................................... 48 Figure 3.6 Concrete Damage Parameter Contours at 0.045 in (μ0.1) of Lateral Displacement (Initial Flexural Cracking of Column) ..................................... 49 Figure 3.7 Concrete Damage Parameter Contours Showing Initial Flexural Cracking in Bent Cap at 0.165 in (μ0.35) of Lateral Displacement (Bottom View) ............................................................................................... 49 Figure 3.8 Development of Cracking in Joint Region at 0.42 (μ0.9) in of Lateral Displacement .................................................................................................. 50 Figure 3.9 Plot of Damage Parameter Contours at 4 in (μ8.6) of Lateral Displacement .................................................................................................. 50 Figure 3.10 Principal Stress Vectors at 0.27 in (μ0.58) of Lateral Displacement .............. 51 Figure 3.11 Close-up of Principal Stress vectors at 0.27 in (μ0.58) of Lateral Displacement .................................................................................................. 52 Figure 3.12 Principal Stress Vectors Overlaid on Damage Parameter Contours at 1.0 in (μ2.15) of Lateral Displacement ........................................................... 52 xiii Figure 3.13 Maximum Shear Stress at 0.35 in (μ0.75) of Lateral Displacement, prior to Development of Cracking in Joint ..................................................... 53 Figure 3.14 Close-up of Maximum Shear Stress in Joint Region at 0.35 in (μ0.75) of Lateral Displacement, prior to Development of Cracking in Joint ............ 53 Figure 3.15 Maximum Shear Stress Distribution at 4 in (μ8.6) of Lateral Displacement .................................................................................................. 54 Figure 3.16 Close-up of Maximum Shear Stress in Joint Region at 4 in (μ8.6) of Lateral Displacement ...................................................................................... 54 Figure 3.17 Average Joint Shear Stress vs. Column Lateral Displacement ....................... 55 Figure 3.18 Joint Rotation vs. Column Lateral Displacement ........................................... 55 Figure 3.19 Joint Rotation at 4 in (μ8.6) of Lateral Displacement (scaling factor of 50) .................................................................................................... 56 Figure 3.20 Longitudinal (X-Stress) Distribution under Gravity and 38 kip Axial Load on Column ............................................................................................. 56 Figure 3.21 Longitudinal Stress Distribution at 4 in (μ8.6) of Lateral Displacement ........ 57 Figure 3.22 Minimum Principal Stress Contours under Gravity and Column Axial Load ................................................................................................................ 57 Figure 3.23 Minimum Principal Stress Contours at 4 in (μ8.6) of Lateral Displacement .................................................................................................. 58 Figure 3.24 Maximum Principal Stress Contours under Gravity and Column Axial Load ................................................................................................................ 58 Figure 3.25 Maximum Principal Stress Contours at 4 in (μ8.6) of Lateral Displacement .................................................................................................. 59 xiv Figure 3.26 Stress in Column and Bent Cap Longitudinal Rebar under Gravity and Axial Column Load ........................................................................................ 59 Figure 3.27 Stress in Column and Bent Cap Longitudinal Rebar at 4 in (μ8.6) of Lateral Displacement ...................................................................................... 60 Figure 3.28 Plot of Stress versus Strain for #5 Column Longitudinal on the South Side of the Bent Cap and Column Joint .......................................................... 60 Figure 3.29 Plot of Compression Stress vs. Compression Strain for #5 Column Longitudinal on the North Side of Column .................................................... 61 Figure 3.30 Plot of Tension Stress vs. Tension Strain for #5 Column Longitudinal on the South Side of Column for an Element Located at Mid-depth of Bent Cap ......................................................................................................... 61 Figure 3.31 Stress in Stirrups and Hoops at 4 in (μ8.6) of Lateral Displacement .............. 62 Figure 3.32 Plot of Tensile Stress vs. Tensile Strain for #3 Stirrup in Column near the Bent Cap ................................................................................................... 62 Figure 4.1 Model 1 and CIP specimen Lateral Load-Lateral Displacement Response Comparison ..................................................................................................... 76 Figure 4.2 Model 2 and CIP Specimen Lateral Load-Lateral Displacement Comparison ..................................................................................................... 76 Figure 4.3 Model 1 and GD Specimen Lateral Load-Lateral Displacement Comparison ..................................................................................................... 77 Figure 4.4 Model 2 and GD Specimen Lateral Load-Lateral Displacement Comparison ..................................................................................................... 77 xv Figure 4.5 Comparison of Average Joint Shear Stresses – Model 1 and CIP and GD Specimens ................................................................................................ 78 Figure 4.6 Comparison of Joint Rotation - Model 1 and CIP and GD Specimens........... 78 Figure 4.7 Maximum Principal Joint Stresses from Model 1and CIP and GD Specimens ....................................................................................................... 79 Figure 4.8 Minimum Principal Joint Stress - Model 1 and CIP and GD Specimens ....... 79 Figure 4.9 Comparison of Concrete Cracking at 48 kips of Lateral Load – Model 1 and CIP Specimen ............................................................................ 80 Figure 4.10 Comparison of Concrete Cracking at 1.4 in (μ3) of Lateral Displacement – Model 1 and CIP Specimen .................................................. 80 Figure 4.11 Comparison of Concrete Cracking at 2.8 in (μ6) of Lateral Displacement – Model 1 and CIP Specimen .................................................. 80 Figure 4.12 Comparison of Concrete Cracking at 3.2 in (μ6.9) of Lateral Displacement – Model 1 and CIP Specimen .................................................. 81 Figure 4.13 Section of Column Shows Location of LC8 Rebar ........................................ 81 Figure 4.14 Location of Longitudinal Rebar Strain Gauges on the CIP Specimen ........... 82 Figure 4.15 Location of Stirrup Strain Gauges on the CIP Specimen................................ 83 xvi 1 Chapter 1 INTRODUCTION 1.1 Background This report, 3-D Finite Element Modeling (FEM) of Reinforced Concrete Beam- column Connections – Development and Comparison to National Cooperative Highway Research Program (NCHRP) Project 12-74, investigates the use of finite element modeling to predict the structural response of the cast-in-place and grouted duct test specimen from NCHRP Report 681 – Development of a Precast Bent Cap System for Seismic Regions. Bridges throughout the United States are in need of immediate repair or replacement because of being deemed structurally deficient or obsolete. Accelerated Bridge Construction techniques are sought as a viable option to quickly replace or rehabilitate structures while minimizing effects on traffic flow. Significant research has been conducted to develop constructible details with reliable performance but use of these details has been limited in seismic regions (Matsumoto, 2009). The prototype structure studied in NCHRP Project 12-74 is a two-span nonintegral three-column precast bridge bent. Figure 1.1 shows the prototype structure. All test specimens were a 42% scale version of the center column and center bent cap region of the prototype structure as shown in Figure 1.2. The cast-in-place (CIP) test specimen, specifically the joint, will be modeled and analyzed in this report. FEM is widely used throughout civil engineering as a method for analyzing complex systems, especially in structural engineering. A finite element model is a 2 mathematical representation of a physical problem and the results of an analysis depend on the type of analysis, element type, element aspect ratio, mesh density, load application and rate, boundary conditions, and material models used in the analyses (Mills-Bria, 2006). In structural engineering applications, the two common finite element analysis (FEA) methods are implicit and explicit analysis. Both of these methods can solve many different types of nonlinear, static, or dynamic problems. The implicit procedure relies on a stiffness matrix and a known set of forces that result in a set of linear equations that can be solved for displacements. The explicit procedure relies on kinematic relationships to solve for accelerations. Force being applied to a structure causes movement and elements in the model strain at certain rates and resist loads (Mills-Bria, 2006). LS-DYNA is an explicit finite element analysis program capable of performing highly nonlinear and dynamic analysis, but this report only considers a quasi-static loading case. This program includes several constitutive models for concrete developed by different researchers for a variety of applications (Predictive Engineering, 2011). One of the most important aspects of creating a reliable model is selecting an appropriate constitutive model. Constitutive models for steel have become reliable and relatively simple to apply to an analysis. Nonlinear constitutive modeling of concrete, however, has proven to be more difficult and still has limitations on its applicability. It is always up to the engineer to verify the accuracy and understand the limitations of the constitutive models being utilized. An accurate prediction of the behavior of reinforced concrete structures requires that the concrete constitutive model simulate known behavior at smaller specimen sizes 3 up to full-scale tests (Wu et al, 2012). Consequently, having a test specimen to compare FEA results against provides an increased level of confidence in the results. Having a well-calibrated model allows engineers to confidently analyze variations of the test specimen and using a FE (finite element) model carry to out alternate loading scenarios. 1.2 Project Objective The overall objective of this project is to create a finite element model that can accurately replicate the structural response of the CIP and GD bent cap systems from NCHRP 12-74. The first goal is to replicate the lateral load-lateral displacement response because it is a strong indicator of overall structural response. Since testing on the specimen was conducted using cyclic loading, the results of the FE model are compared to the lateral load-displacement hysteretic envelope in the positive displacement region under “push” loading. Other structural parameters examined in this report are deflected shape, cracking of concrete at various stages, shear stress in the joint, magnitude and orientation of principal stresses in the joint, overall flexural stress distribution, and stress distribution throughout steel reinforcement. 1.3 Significance of Project Results FEA results are only meaningful if the model correlates well with test data and accurately predicts structural behavior. Ultimately, the goal is to have a well-calibrated and reliable model that can allow for further study on effects of geometric and material modifications. More directly, the FEA results will be used to develop a strut-and-tie model of the structure. The orientation of principal stresses and the flow of forces allows 4 engineers to determine proper magnitude and location of compression struts, tension ties and nodes. 1.4 Literature Review Several analyses have been published on the nonlinear finite element analysis of reinforced concrete using LS-DYNA. LS-DYNA includes at least eight different constitutive models that can be used to model concrete. This analysis uses the Karagozian and Case Concrete Damage Model Release 3 (K&C model) (LSTC, 2007). Much of the literature available for this model is based on blast loading applications, but it has proven to be an effective model for simulating static and quasistatic loading scenarios as well. The K&C model was selected because numerous studies, referenced throughout this report, have shown excellent response under various confining pressures, damaged conditions, and load capacity. Release 3 of the K&C Concrete Damage Model brings several improvements over previous versions. Magallanes et al. presented the formulation, improvement, and results for the K&C model. The first improvement is the automatic input capability for generating the model parameters. This addition makes the model much easier to use, especially when limited information is available on the concrete properties. The second improvement relates to methods embedded into the model to reduce dependencies on the mesh due to strain-softening. Lastly, guidance is provided on properly modeling strain rate effects and discusses the effects of the strain rate parameter. The U.S. Department of the Interior Bureau of Reclamation published the handbook, “State-of-Practice for the Nonlinear Analysis of Concrete Dams at the Bureau 5 of Reclamation.” The focus of the handbook is on analysis of concrete dams; however, some of the general information it provides can be applied to many other concrete structures. Information specifically reviewed for this report was on contacts between different surfaces, damping, material properties, and boundary conditions. Malvar and Simons (1996) discusses the development of a Lagrangian finite element code with explicit time integration for analyzing structures. It explains the formulation for a damaged concrete constitutive model and the basis of its formation. Tensile cutoff, volumetric damage, damage accumulation, strain rate effects, and pressure cutoff are presented along with their formulas. Wu et al (2007) presented their results of comparing various LS-DYNA constitutive concrete models in a presentation paper titled “Performance of LS-DYNA Concrete Constitutive Models.” This report studied the K&C model, Winfrith Concrete model, and the Continuous Surface Cap model. The models are compared against each other under an unconfined uniaxial compression test, tension tests, triaxial compression tests, and response to a blast load. Concluding remarks from this report show that the K&C model is capable of capturing key concrete behaviors including post-peak softening, shear dilation, confinement effect, and strain rate effect. The K&C model is also suitable for quasi-static loads, even though primarily developed for blast and impact loading. Sandia National Laboratories also presented their results of comparing four concrete damage constitutive models in their report titled “Survey of Four Damage Models for Concrete”. They compared results from the K&C concrete model, the Riedel- 6 Hiermaier-Thoma model, the Brannon-Fossum model, and the Continuous Surface Cap model. The parameters against which these models were compared include strength surfaces, strain-rate dependence, damage accumulation, plastic update, and shear and bulk moduli. Schwer and Malvar (2005) compared the results of the K&C model against the well-characterized unconfined compression strength concrete from the Geotechnical and Structures Laboratory of the US Army Engineering Research & Development Center. Areas of study included compaction, compressive shear strength, and extension and tension of concrete. Data from this study shows that the K&C model is capable of capturing the complex behavior of concrete, especially when only a minimal amount of information is known (i.e. compressive strength) about the concrete. Many of the internal parameters have been calibrated to extensive test data and relationships based on compressive strength were developed. Sritharan et al. (2000) performed similar nonlinear finite element analysis of beam-column connections in order to incorporate effects of bar slip. It was found that bar slip appeared to have a significant effect on the stress and strain contours as well as cracking in the joint region. Sritharan et al. (2000) modeled the nonlinearity of cracked concrete using a smeared-concrete constitutive model in ABAQUS with ANAMAT. Confinement of concrete was accounted for by forcing the concrete model to follow the Mander model stress- strain curve for confined concrete. The K&C model differs from this because it is capable of self-generating the effects of confinement. This paper also 7 provides stress distributions and corresponding strut and ties of the model. These results were studied as a verification of results. The end goal for the results of this project is to develop a strut-and-tie model of the beam column connection. Sritharan (2005) presents a strut-and-tie model based approach to designing concrete bridge joints. Sritharan (2005) discusses the locations and magnitudes of struts, ties, and nodes. 1.5 Project Approach This project began with a review of the NCHRP Project 12-74 CIP test set-up, testing procedure, and analysis of relevant test data. This information was an excellent starting point for developing an FEA scheme with the appropriate geometry, boundary conditions, and material properties for the test. LS-DYNA was selected as the most suitable finite element analysis software after extensive review of available constitutive models, nonlinear modeling capabilities, and case histories of the software. LS-DYNA was advantageous over other FEA programs such as ABAQUS because of its ability to handle highly nonlinear analysis using an explicit solving scheme. The nonlinear constitutive models available in LS-DYNA have been extensively researched and tested, thus providing greater confidence in results. In total, seven different models were developed and analyzed in order to select the most appropriate modeling approach. Each of the seven models varied with one or more of the following: mesh density, boundary conditions, constitutive models, and element type. Of the seven models, only the two that provided the most comparable data will be discussed in this report. The results have been post-processed and presented in the following chapters. 8 1.6 Scope of Report This report will serve as a basis for future FEA on other NCHRP 12-74 precast bent cap models and strut-and-tie models. This report includes extensive comparisons of FEA results to NCHRP 12-74 results, with a primary focus on the CIP specimen and with limited comparisons to the GD specimen. This report includes the following chapters: 1.0 Introduction 2.0 Development of Finite Element Model 3.0 Finite Element Analysis Results 4.0 Comparison of Results to Specimen Data 5.0 Summary, Conclusions, and Recommendations 9 Figure 1.1: Prototype Structure from NCHRP Project 12-74 (Matsumoto, 2009) 10 Figure 1.2: Portion of Prototype Used for Testing and Modeling (Matsumoto, 2009) 11 Chapter 2 DEVELOPMENT OF FINITE ELEMENT MODEL This chapter begins with a description of the CIP specimen being modeled and then transitions to the geometry, element formulation, constitutive models, boundary conditions, and the mesh formulation. 2.1 CIP Specimen Information The CIP specimen was made up of two components, a bent cap and a column (see Figure 2.1). The bent cap used a 25 in x 25 in cross section and a length of 12 feet. The steel reinforcing in the bent cap consists of 12-#5’s (0.65%) at top and bottom for flexural reinforcement and #3’s at 6 in for shear reinforcement. Dimensions and reinforcement placement are detailed in Figures 2.1 and 2.2. The concrete mix had a compressive strength of 4,553 psi. The column has a circular cross section with a 20 in diameter. It includes 16-#5’s (1.58%) for longitudinal reinforcement and #3 hoops at 2 in. The concrete in the column had a compressive strength of 6,178 psi. Concrete properties are summarized in Table 2.1. The joint region reinforcement consists of 4-leg #3 stirrups at 5 in (with two sets of #3 cross ties through depth) adjacent to each side of the joint. Hoop reinforcement through the column consisted of #3s at 5 in. Average yield strength measured for the #5 rebar was 61.3 ksi and 68.2 ksi for the #3 rebar. Reinforcement quantity and strengths are summarized in Table 2.2 (Matsumoto, 2009). The CIP specimen was fabricated as two separate components; the column was cast first, as shown in Figure 2.3, and then the bent cap was cast over the column as 12 shown in Figure 2.4. The fabrication and assembly process was intended to replicate the field process as much as possible in order to predict any constructability issues. After concrete had cured to an adequate strength, the structure was inverted for testing. The bent cap now formed the bottom of the structure and the column formed the top of the structure. The bent cap was simply supported, with a pin support at the north end and a vertical “roller” support at the south end, as shown in Figure 2.5. A force controlled 38 kip axial load was applied to the top of the column followed by a force controlled and displacement control sequence of lateral load or displacement applied to the center of the column stub (Matsumoto, 2009). 2.2 Finite Element Model Geometry Similar to the CIP specimen, the FE model consists of two individual components, a bent cap and a column (Figure 2.6), with the steel reinforcing. The FE model is based on the design drawings of the CIP specimen, not the as-built dimensions, as obtained from Figures 2.1 and 2.2. The differences between the design and as-built dimensions are minor and were therefore neglected. Minor changes were made to rebar lengths in order to simplify modeling. For example, the test specimen had clear cover on the ends of the longitudinal rebar in both the column and bent cap of approximately 1 inch. The FE model assumes the rebar extends to the end of column, leaving no clear cover. Since the ends of the longitudinal bars were not in any of the critical regions of the model, this change had no direct effect on the overall results. Figure 2.7 shows the dimensions of the FE model and Figure 2.8 shows the geometry of the rebar compared to the CIP rebar cage. 13 The concrete elements in the bent cap are not bonded or merged in any way to the concrete elements in the column. The two components are connected by the longitudinal rebar extending through the column and all the way through the height of the bent cap. The connection is assumed to be a cold joint. There is a contact surface with a coefficient of friction of 0.6 (AASHTO requirement for CIP structures) defined between the two components. Figure 2.9 shows a partial cutaway of the concrete elements in the bent cap to show the longitudinal rebar connecting the two parts. 2.3 Element Formulation The two concrete parts of the FE model use eight-node hexahedron elements. The constant stress solid element formulation was used with varying mesh sizes ranging from 1 in to 1.5 in. The smaller elements were used in the critical joint region and larger elements outside of the joint. Element size was varied in order to reduce element quantity in non-critical regions, which decreased computation time. Figure 2.10 shows the relative mesh size for the solid elements of the entire structure. Figure 2.11 shows a 3-D view of the meshed structure. The steel reinforcement is modeled using circular Hughes-Liu beam elements. Figure 2.12 shows a close-up of the meshed beam elements in the joint region. The FE model consists of 55,860 solid concrete elements and 4,976 beam elements, for 60,836 total elements. The concrete and steel elements require a good coupling mechanism in order to achieve interaction between the two parts. This analysis uses the CONSTRAINED_LAGRANGE_IN_SOLID formulation to achieve a proper interaction relationship. Nodes from the rebar beam elements couple with the surrounding concrete 14 element nodes and therefore strain between the two elements is coupled. Consequently, this technique implies that the concrete and rebar are fully bonded for the entire length of the rebar and there is no occurrence of bar slip. There is no development length for the rebar, which would not happen in a real specimen. Although the lack of development length is not expected to have any significant impact on the results, the stress distribution in the rebar can show whether or not there is considerable stress near the ends of the rebar and whether or not refinement is needed to rebar modeling. If stress was present on the ends of the rebar in the development length region, the coupling technique would need to be modified. If there is no stress in that region, then it can be assumed that the bar was not strained in that region. The Lagrange Constraint command is advantageous over other coupling techniques because it does not require rebar nodes to coincide with the concrete nodes (Bermejo et al, 2011). This allows for more flexibility with placing rebar in the model and makes the task of modeling rebar much quicker. 2.4 Constitutive Models The solid concrete elements are modeled using LS-DYNA material type MAT_072R3, the MAT_CONCRETE_DAMAGE_REL3 model. Karagozian & Case (Glendale, CA) developed this model for blast loading and quasi-static loading applications using lightweight and normal weight concrete. Release 3 of the K&C concrete model is a three invariant model, uses three shear failure surfaces, and includes damage and strain-rate effects and is based on Material Type 16, which is the PseudoTENSOR Model (LSTC, 2007). 15 This constitutive model is able to generate input parameters based solely on the unconfined compressive strength of the concrete. It relies on the concrete strength to obtain other parameters by using relationships that correlate compressive strength to tensile strength and bulk modulus. The deviatoric strength is calculated using simple functions to characterize three independent failure surfaces that define the yield strength, maximum strength, and the residual strength of the material (Magallanes et al, 2010). Hardening of the material is captured by interpolating the plasticity surface between yield and maximum surfaces based on the value of an internal damage parameter. Softening follows a similar procedure but interpolation is performed between the maximum and residual surfaces. Figure 2.13 shows the strength model of the concrete with the three failure surfaces. This model also includes a tension softening parameter that is scaled using simple relationships for concrete. The tension softening parameter controls the strain softening and also forms the basis for determining the fracture energies. This model provides the option to manually enter the fracture energy or allow it to internally generate the parameter based the unconfined compressive strength and maximum aggregate size. Parameters for strain rate effects are also self generated by the model in order to capture inertial effects on the concrete (Magallanes et al, 2010). For this analysis, the aggregate size and unconfined compressive strength were defined in the model, leaving all other parameters to be self-generated. Concrete for the bent cap used a compressive strength of 4,553 psi and the concrete for the column used a compressive strength of 6,178 psi. Both the column and bent cap usd ¾ - inch aggregate. 16 Limitations of this model, relevant to this analysis, are its performance under cyclic loading. Elements accumulate damage and are unable to sustain stress after the maximum damage parameter of 2.0 has been reached. The damage accumulates in this model based on strain rate, strain, and volumetric strain parameters (Markovich et al, 2011). If an element becomes fully damaged from tensile stress, its ability to carry a compressive load is greatly diminished. This presents an issue under cyclic loading, where an element may be subjected to many compression and tension cycles. The reinforcing steel is modeled using MAT_003, known as MAT_PLASTIC_KINEMATIC. This model is well suited for isotropic and elastic behavior. It has the option of including strain-hardening effects using a linear relationship for plastic behavior (LSTC, 2007). This analysis does not consider strain-hardening effects of rebar because of issues with obtaining a steady response of the overall FEM. The stress-strain curve follows a bilinear relationship that has a zero slope after yield. Further analysis is needed to refine the concrete constitutive model definition and meshing of the structure in order to include strain-hardening effects to provide more reliable results. 2.5 Mesh Development Finite element analysis is reliant on the size and quality of the mesh. Typically, a greater number of elements results in increased accuracy of analysis and more a refined distribution of stresses. The drawback to increasing the number of elements is that computation time tends to increase significantly. 17 For this project, analyses were performed with increasing mesh density (i.e., reduced element size) until results were acceptable. Several analyses were discarded because of unsteady response of lateral load as a function of displacement. Through iterations of mesh size, the response of the structure became increasingly steady. The increased mesh density of the concrete, especially in the joint region allowed for concentrated damage through the joint with distinct diagonal cracking. Analyses with a course mesh were not detailed enough to isolate specific cracking, but rather just the overall region where the cracking occurs. Increasing the mesh density in the column resulted in a more accurate distribution of vertical stress and vertical reactions. After several analyses with varying mesh density, the length of beam elements used to model steel reinforcement appeared to have a greater impact on the stability of the results than the concrete element size. Overall results such as concrete damage patterns and stress distributions were all much more realistic after a suitable mesh was settled upon. This analysis will discuss two of the models that were developed, Model 1 and Model 2. Model 1 will be the primary focus of discussion but Model 2 will be used to compare the lateral load-displacement response. The only difference between Model 1 and Model 2 is the length of the beam elements used to model the reinforcing hoops in the column. Model 1 uses the same length elements for every hoop in the column and the joint region. Model 2 has a relatively coarse mesh for the hoops in the column but the hoops in the joint region have the same element length as in Model 1. Except for the lengths of the hoop elements just described, Model 1 and Model 2 were identical. Figure 18 2.14 compares a column hoop from Model 1 and Model 2. The course hoop from Model 2 has an element length of 3.44 in and the finer meshed hoops from Model 1 have an element length of 1.58 in. 2.6 Boundary Conditions Boundary conditions selected in the FE model are consistent with conditions from the actual specimen test set-up. Figure 2.5 is a photograph of the test set-up and shows vertical and horizontal restraints at “N” (north) end and a vertical restraint at “S” (south) end. An axial load and lateral displacement were applied to the top of the column. Figure 2.15 shows boundary conditions applied to the FE model to replicate the vertical and horizontal restraints applied to the test specimen. Figure 2.16 shows the nodes used to apply axial load and lateral displacement on the column. 2.7 Loading Application In order to minimize dynamic effects in the analysis and simulate the quasi-static test that was actually conducted, loading was applied in a sequence over a ten-second time interval. Gravity was applied first by linearly ramping from 0 to 100% over a two second interval. Second, the 38-kip axial load was applied to the column stub over a twosecond interval. Finally, 4.5 inches of lateral displacement was applied over a 6-second interval. Table 2.3 summarizes the loading sequence. 19 Table 2.1: A) Summary of CIP Concrete Properties B) Comparison of GD and CIP properties (Matsumoto, 2009) Parameter Design Actual Slump 5½'' +/- 2½'' < 3 in, cap and column Unit Weight 143.9 pcf N/A Cap: 4553 psi (137 days) Column: 6178 psi (194 days) Cap: 361 psi (138 days) Column: 452 psi (195 days) Compressive Strength 4000 psi (28 day) Tensile Strength (Split Cylinder) N/A Table 2.1.A Parameter GD CIP f'c Steel Rebar Strength Cap and Column: 4557 psi Yield Tensile Cap: 4553 psi Column: 6178 psi Yield Tensile #3 (Bent cap stirrups; Column hoops) 64.1 99 68.2 95.5 #5 (Bent cap longitudinal; Column longitudinal) 64.5 95.2 64.5 90 Grout Compressive Strength (Bedding layer and ducts) 8026 psi (6421 psi, equivalent cylinder strength) Table 2.1.B N/A 20 Table 2.2: Summary of Reinforcement for Bent Cap, Joint, and Column (Matsumoto, 2009) CIP Prototype Design Similitude [or Design] Requirement Test Specimen Specimen to Similitude [or Design] Ratio #5 (0.63) 16 64.5 0.0158 222 48.0 1.06 1.14 — 1.14 0.82 0.80 #3 (0.38) 2.0 68.2 0.0125 1.20 [1.11] — 0.90 #5 (0.63) 12 64.5 0.0065 454 50.2 1.06 — — 1.27 0.94 1.14 #3 (0.38) 6.0 68.2 1.20 [0.71] — Column Longitudinal Reinforcement Bar Size (diameter, in) No. of bars fy ksi ρ Mn* K∙ft Mn/Dc3 K∙ft/ft3 #11 (1.41) 16 66.0 0.0138 3760 58.7 Bar Size (diameter, in) Spacing in fy ksi ρ #6 (0.75) 3.0 66.0 0.0139 0.59 14 — 0.0138 272 58.7 Transverse Reinforcement 0.31 [1.8] — 0.0139 Bent Cap Longitudinal Reinforcement Bar Size (diameter, in) No. of bars fy ksi ρ Mn* K∙ft Mn/bh2 K∙ft/ft3 #11 (1.41) 12 66.0 0.0051 6680 44.2 Bar Size (diameter, in) Spacing in fy ksi #6 (0.75) 12.0 66.0 0.59 — — 0.0051 483 44.2 Transverse Reinforcement 0.31 [8.4] — Joint Inside Joint Transverse Reinforcement (ρs) Bar Size (diameter, in) Spacing in ρs/ρmin fy ksi Side Face Reinforcement (Assf) No. of bars - Bar Size As/Acap in2/ in2 fy ksi Construction Stirrups No. of bars - Bar Size Area in2 fy ksi #6 (0.75) 3.0 5.56† 66.0 — — — — #3 (0.38) 5.0 1.22 68.2 — — — — 8 - #6 0.19 60.0 — [0.10] — 4 - #3 0.12 68.2 — [1.20] — 2 - #6 0.88 66.0 — — — 2 - #3 0.22 68.2 — — — Adjacent to Joint Vertical Stirrups (Asjv) No. of bars - Bar Size Spacing in Asjv/Ast in2/ in2 fy ksi Horizontal Ties (Asjh) No. of bars - Bar Size Spacing in Asjh/Ast in2/ in2 fy ksi 5 - #6 6.0 0.35 66.0 — — [0.20] — 3 - #3 5.0 0.27 68.2 — — [1.33] — 4 - #6 12.0 0.35 66.0 — — [0.10] — 2 - #3 8.0 0.13 68.2 — — [1.33] — 21 Time 0 to 2 seconds 2 to 4 seconds 4 to 10 seconds Table 2.3: Loading Sequence for FEM Loading Gravity applied over a linear ramp. Gravity is ramped to 100% over 2 seconds and stays constant for remainder of analysis. 38-kip axial load applied as a ramp load over this interval and remains constant for remainder of analysis. 4.5 inches of displacement applied to the column stub over this interval. 22 Figure 2.1: Elevation of CIP Specimen Dimensions (Matsumoto, 2009) 23 Figure 2.2: Cross Sections of CIP Specimen Showing Rebar (Matsumoto, 2009) 24 Figure 2.3: Column with Instrumentation in Place Prior to Casting of Bent Cap (Matsumoto, 2009) 25 Figure 2.4: Bent Cap Formwork Placed Over Column (Matsumoto, 2009) 26 Figure 2.5: Test Set-up (Matsumoto, 2009) 27 Joint Region S Support N Support Bent Cap Column Column Stub Figure 2.6: Individual Bent Cap and Column Components 144 in 25 in 3 in diameter 20 in 55 in Figure 2.7: Dimensions of FE Model 28 #3 stirrups #3 hoops at 5 in 16-#5 #3 hoops at 2 in 18 in O.D. A B Figure 2.8: A) Rebar configuration in FE Model, B) CIP Specimen Bent Cap Rebar Cage (Matsumoto, 2009) 29 Bent Cap Column Figure 2.9: Half of Bent Cap Cut-away (with Stirrups Removed) to Show Longitudinal Rebar Connecting the Bent Cap and Column Elevation View Figure 2.10: Mesh of Solid Elements End View 30 Figure 2.11: 3-D Isometric View of Meshed Structure 31 Joint Region Column Figure 2.12: Close-up of Meshed Beam Elements in Joint Region Figure 2.13: Strength Model for Concrete (Malvar et al., 1997): (a) Failure Surfaces in Concrete Model; (b) Concrete Constitutive Model 32 1.58 in Model 1 3.44 in Model 2 Figure 2.14: Comparison of #3 Hoop Steel Element Length in Model 1 and Model 2 Vertical and horizontal restraints Vertical restraint N S Figure 2.15: Vertical and Horizontal Boundary Conditions Axial load 33 Lateral displacement N S Figure 2.16: Boundary Conditions for Axial Load and Lateral Displacement 34 Chapter 3 FINITE ELEMENT ANALYSIS RESULTS This chapter presents results from the FE analysis of the CIP specimen models of NCHRP 12-74 described in Chapter 2. The lateral load-displacement response is presented first. The lateral load-displacement relationship is the most important area of study for this analysis because it is the best indicator of overall structural response. The second result presented is the concrete damage parameter that is internal to the K&C concrete constitutive model. This parameter is an indicator of cracking in the structure at various stages. This is followed by an analysis of the orientation of principal stresses, joint rotation, joint shear stress, overall (flexural) stress distribution, and stress distribution in the reinforcing steel. Each of the parameters described are analyzed at relevant key points of interest, which include one or more of the following: a. Initial cracking of bent cap and column b. Initial cracking of joint region b. Initial yielding of longitudinal rebar in column d. Maximum lateral displacement of column 3.1 Lateral Load-Lateral Displacement Response The lateral load-lateral displacement response for both FE models was developed from the FEA by acquiring the lateral displacement at the center of the column stub and the lateral reaction corresponding to a specific displacement (shown in Figure 3.1). Displacement ductility, μ, was calculated by normalizing displacement by the displacement at effective yield. Effective yield was calculated as described in 35 Matsumoto (2009). Both curves start with elastic behavior which transitions to plastic response between 0.2 and 0.5 in of lateral displacement. Model 1 reaches a peak lateral load of 65.5 kips at a lateral displacement of 4 in (μ8.6). Model 2 reaches a peak lateral load of 57 kips at a lateral displacement of 2.63 in (μ5.7). Based on the lateral load-displacement response, Model 1 (shown in Figure 3.1) appears to have a steadier response relative to Model 2 (shown in Figure 3.2) through four inches of lateral displacement. The lateral load-displacement response for Model 2 has two sudden dips in force of approximately 8.6 kips at 0.18 in (μ0.34) and 0.465 in (μ1) of lateral displacement. These dips correspond with times when large areas of concrete accumulate damage over a lateral displacement interval of less than 0.015 in (μ0.03). Elements exceed a damage parameter value of 1.0 and LS-DYNA interpolates strength between the maximum strength surface and the residual strength surface. Since the damage parameter goes from zero to two over a very small displacement, the transition from the maximum strength surface to the residual surface is abrupt. A drastic change in the active surface, from maximum to residual, causes a drop in the force at the displacement where damage has occurred. Since the residual strength surface is weaker and less stiff than the other two strength surfaces (as shown in Figure 2.13), the transition to the residual strength translates to a less stiff response in the lateral load-displacement following the dip. The “softening” of the response can also be observed on the lateral load displacement curve for smaller dips that occur at various displacements. Model 1 has a series of relatively smaller dips on the lateral load-displacement response between displacements of 0.40 in (μ0.86) and 0.735 in (μ1.6). 36 These dips occur for the same reason as described for Model 2, except the magnitude of the drop in load is smaller. Figure 3.3 shows the lateral load-displacement curves for both Model 1 and Model 2 for comparison. Relatively large areas of concrete elements accumulating significant damage simultaneously were not observed as in Model 2. Damage to the elements is distributed out over a longer interval of lateral displacement. The smoother response of Model 1 is attributed to the smaller element size of the hoop steel reinforcing in the column. The increased number of elements in the hoop steel provides a more accurate confinement of the elements, which limits the simultaneous damage accumulation. Since the hoop steel element length of Model 2 is longer than Model 1, when an element reaches the yield stress and begins plastic strain, a greater length of element yields, which results in less confinement. With the finer rebar mesh of Model 1, the yielding can be isolated more accurately to a smaller length of rebar. The smaller length elements therefore produce smaller strains, consequently resulting in higher confinement. Analysis from this point on is carried out on Model 1, unless otherwise specified, as it is the more reliable and accurate model. 3.2 Concrete Damage Parameter The K&C concrete constitutive model determines the strength of the concrete based on an internal damage parameter that ranges from 0.0 to 2.0. The active strength surface is based on an interpolation between two of the three strength surfaces in the model. If the damage parameter is less than 1.0, it is interpolating strength between the yield surface and the maximum strength surface. When the parameter is equal to 1.0, it has reached the maximum strength surface. When the damage parameter is greater than 37 1.0, strength is being interpolated between the maximum strength surface and the residual strength surface. In this analysis, most concrete elements do not reach the fully damaged state (i.e., a damage parameter equal to 2.0). The reinforcing steel provides confinement, which increases the failure strength of the concrete and ultimately does not allow elements to become fully damaged in compression. In this analysis, elements that accumulated damage would typically reach a maximum damage value that approached 2.0 but was always slightly less (approximately 1.97). For elements under compression, the damage parameter was close to 2.0 and the damaged element strength is based was essentially based on the residual strength curve. Figure 3.4 shows a plot of the vertical compressive stress (Z-stress) and the damage accumulated for an element in the column. This figure shows the short transition period of the model interpolating between the various surfaces. After reaching the maximum stress value, the stress drops rapidly towards the residual strength curve. This corresponds to the damage parameter going from 1.0 and approaching 2.0 over a short amount of time. The model does not have a residual strength curve for elements that have failed in tension, so after reaching the maximum tension strength the stress in the element rapidly degrades to zero. Figure 3.5 is a plot of the vertical tensile stress (Z-Stress) and the damage parameter. It shows that as the damage parameter approaches 2.0, the tensile strength goes to zero. Contours of this damage parameter are plotted in Figures 3.6 through 3.9. The damage contours range between blue and red; blue representing nearly no damage accumulation and red representing damage values approaching 2.0. Figure 3.6 shows the 38 plot of the damage parameter at 0.045 inches (μ0.01) of lateral displacement, which corresponds to approximately 20 kips of lateral load. The damage parameter for the concrete elements on the tension side the column near the bent cap and column connection approaches 2.0, which appears to be the formation of initial cracking due to flexure. Figure 3.7 shows the damage parameter approaching 2.0 on the bottom south side of the bent cap at 0.165 (μ0.35) in of displacement, which corresponds to approximately 36 kips of lateral load. Figure 3.8 shows the initial development of a crack through the joint at 0.42 (μ0.9) in of lateral displacement, which corresponds to a lateral load of 51 kips. Lastly, Figure 3.9 shows the damage accumulation at 4.0 in (μ8.6) of lateral displacement, which corresponds to the maximum lateral load of 65 kips. At 4.0 in (μ8.6) of lateral displacement, there is significant damage accumulation in the column. Most of the elements on the compression side of the column near the joint have transitioned toward the residual strength curve at this stage because the damage parameter approaches 2.0. No attempt was made at making a correlation between the damage parameter and concrete spalling on the column because of the relatively large size of concrete elements in the column. Larger elements made it difficult to isolate spalling. A cross section of the center of the bent cap shows that interior joint damage was minimal and remained in the top 17 in of the bent cap in a 5 in diameter area surrounding the column longitudinal rebar connecting to the joint. 39 3.3 Principal Stress Vectors The principal stress vectors are studied to determine where the maximum stresses are forming and to locate areas of cracking. For this analysis, the focus was on the principal stresses in the joint region of the bent cap. These stresses are used to predict cracking through the joint and are compared against cracking predicted by the K&C model using the damage parameter. Figure 3.10 shows the maximum (tensile) principal stress vectors at 0.27 inches (μ0.58) of displacement, which just before the development of cracking in that area. The plot shows principal stress vectors forming at a diagonal through the joint. Figure 3.11 zooms in on the joint to highlight the vectors through the joint. The principal vectors show that tension causes the primary diagonal crack through the joint. Figure 3.12 shows the principal stress vectors overlaid with the contours of the concrete damage parameter. The concrete becomes damaged along the same diagonal formed by the maximum principal vectors. 3.4 Joint Shear Stress Shear stress in the joint region of the bent varies across a large range with small areas of very high stress concentration. Typical shear stress values range between 100 psi (1.48√𝑓𝑐′ ) to 500 psi (7.41√𝑓𝑐′ ). Figures 3.13 and 3.14 show the contour plots of the maximum shear stress at 0.345 in (μ0.74) of lateral displacement, which is prior to cracking in the joint. A small zone of shear stress concentration, approximately 2x2 in, is seen where the beam contacts the compression side of the column. Shear stress values in this region reach a maximum value of 1,041 psi (15√𝑓𝑐′ ) at 4 in (μ8.6) of lateral displacement. Figures 3.15 and 3.16 show the maximum shear stress values at 4 in (μ8.6) 40 of lateral displacement. The damage parameter of the concrete elements in the shear stress concentration zone approaches 2.0 at initial joint cracking. Since the element strength is calculated based on the residual strength surface, the elements reach high stress values. Maximum stress values are summarized in Table 3.1. The average shear stress of the joint is estimated by averaging the shear stress values from twelve elements through the center of the joint. The average values from the center were comparable with other sections of the joint. The averaging was limited to twelve elements because of the time-consuming process of post-processing data from several elements. Selecting every other element through the center of the joint resulted in twelve elements. Figure 3.17 shows the average shear stress plotted against the lateral displacement of the column. This plot exhibits a similar trend as the lateral loaddisplacement plot and relatively small stresses. 3.5 Joint Rotation Joint rotation in the FE model was measured using a similar method as from Matsumoto (2009). The instrumentation set up from Matsumoto (2009) is shown in Figure 2.5. Joint rotation on the FE model was calculated using nodal displacement data at the nodes corresponding to locations of the cap rotation instrumentation on the CIP specimen. The following equation from Matsumoto (2009) was used to calculate the joint rotation angle: 𝜃𝑗𝑟 = 𝛿𝑗𝑟,𝑛 − 𝛿𝑗𝑟,𝑠 𝐷𝑐 where, 𝛿𝑗𝑟,𝑛 = 𝐽𝑜𝑖𝑛𝑡 𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡, 𝑛𝑜𝑟𝑡ℎ 41 𝛿𝑗𝑟,𝑠 = 𝐽𝑜𝑖𝑛𝑡 𝑟𝑜𝑡𝑎𝑡𝑖𝑜𝑛 𝑑𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡, 𝑠𝑜𝑢𝑡ℎ 𝐷𝑐 = 𝐷𝑖𝑎𝑚𝑒𝑡𝑒𝑟 𝑜𝑟 𝑑𝑒𝑝𝑡ℎ 𝑜𝑓 𝑐𝑜𝑙𝑢𝑚𝑛 𝑖𝑛 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑙𝑜𝑎𝑑𝑖𝑛𝑔 Figure 3.18 shows a plot of the joint rotation against the lateral displacement. The maximum joint rotation was calculated to be 1.0x10-3 radians at a lateral column displacement of 4 in (μ8.6). The exaggerated deflected shape of the bent cap is shown in Figure 3.19 with a scaling factor of fifty applied at 4 in of lateral displacement. The column is removed from Figure 3.19 for clarity. 3.6 Longitudinal and Principal Stress Distribution The two primary stresses discussed in this section are the longitudinal stresses acting in the horizontal direction, the X-stresses, and the principal stresses. The Xstresses in the bent cap are a result of flexure and axial loading caused by the lateral displacement of the column and the 38 kip axial load on the column. Figure 3.20 shows the X-stress contours acting on the bent cap after gravity and axial loading on the column stub is applied (lateral displacement equals zero at this point). This stress distribution reflects a simply supported beam: tension on the bottom face and compression on the top face. By observation, the neutral axis runs through mid-height of the bent cap since the concrete is uncracked and the reinforcing is symmetric. Figure 3.21 shows the stress distribution of the entire structure at 4 in (μ8.6) of lateral displacement of the column. Results are only presented at 4 in (μ8.6) of lateral displacement because following lateral loading, the distribution does not change significantly, but only the magnitude does. As lateral displacement is incrementally applied to the column stub, the stress distribution begins to change and forms a diagonal compression strut through the joint. Distinct 42 tensile zones form near the joint and the lateral restraint (as shown in Figure 3.21), which suggest possible locations of tension-ties. Stress distribution in the column behaves as expected, with compressive stress acting on the side opposite of where the push load is applied. Tension stresses develop on the same side as the lateral load application due in part to the horizontal restraint at the pin. The moment generated in the column by lateral loading of the column stub is transferred to the joint region of the bent cap via tension and compression of longitudinal column reinforcement and compression of the column concrete against the bent cap. The moment that develops in the joint causes a compression zone on the top face of the right (S) side of the bent cap and a tension zone on the left (N) side of the bent cap, as shown in Figure 3.21. The maximum tension stress on the top face of the bent reached approximately 280 psi (4√𝑓𝑐′ ). The maximum compressive stress on the top face of the bent reached approximately 700 psi (0.15𝑓𝑐′ ). These values do not include stress concentrations near the bent-column contact, where stress levels reached higher values. The concrete elements in the stress concentration area are reaching higher values based on their residual strength surface. Figure 3.22 shows the minimum principal stress contours after gravity and the axial load on the column have been applied. This figure shows how compressive stresses are transferred from the column to the bent cap, and then transferred to the vertical restraints. Figure 3.23 is a plot of the principal compressive stress contours at 4 in of lateral displacement. This plot clearly identifies the location and direction of a 43 compression strut through the joint. Another compression strut can be seen going from the top left (N) side of the joint to the left (N) restraint. Figure 3.24 shows the maximum principal stress contours after gravity and axial loading on the column have been applied. Tensile stresses are observed on the bottom half of the bent cap and the stress can be observed being transferred to the vertical restraints. Figure 3.25 shows the maximum principal stress vectors at 4 in of lateral displacement. Tension stress is observed on the bottom of the right side of the bent cap beneath where the compressive stresses act. Figure 3.25 also shows significant tension in the joint region, which contributed to the development of a shear crack. Maximum principal stress vectors overlaid on the damage parameter contours verifies this result. 3.7 Response of Reinforcing Steel The stress distribution in the longitudinal rebar under gravity and the axially loaded column is shown in Figure 3.26. Figure 3.27 shows the stress in the longitudinal bars at 4 in of lateral displacement. The longitudinal #5 rebar in the column yielded in the plastic hinging region adjacent to the column connection, and indicates strain penetration into the joint. Near the joint, column longitudinal rebar yielded at 60.6 ksi and remained constant at that value for the remaining analysis because strain-hardening effects were neglected. Rebar in the column yielded due to both compression and tension, although the zone of tension yielding area is much larger. Longitudinal rebar on the bottom right (N) side of the bent cap adjacent to the joint reached a maximum stress of 27.4 ksi (45.2% of yield). 44 Figure 3.28 is a stress-strain plot of the #5 column longitudinal rebar on the north side of the column. Figure 3.29 is a stress-strain plot of the #5 column longitudinal rebar on the south side of the column, in the compression region. Figure 3.30 is a stress-strain plot for an element of #5 column longitudinal rebar 12.5 in into the bent cap on the north side of the joint. Figure 3.30 shows a stress strain plot for an element of rebar that did not reach yield. Hoop reinforcing along the length of the column yielded at 68.4 ksi due to damage and subsequent dilation of concrete elements. The hoops yielded in tension on the north side of the column. Stress and strain values at key locations on the hoops is discussed further in Chapter 4. Stirrups in the bent cap did not reach the yielding stress. Stirrups in the joint region reached a stress of 15% of yield, but outside of this region, they had negligible stress values. Figure 3.31 shows the distribution of stress throughout the column hoops and the stirrups in the bent cap. Figure 3.32 shows the tensile stress-strain plot of a stirrup on the compression side of the column that yielded. 45 Table 3.1: Summary of Stresses in Concrete Elements Stress Loading State Component Gravity + 38 kip axial on column 4 in of lateral displacement (μ8.6) Maximum Principal (psi) Minimum Principal (psi) Maximum Shear (psi) Joint Region 335.13 469 21 Column - 245 105 Joint Region 434 3472 1566 Column - 6944 2558 Table 3.2: Summary of Stresses in Reinforcing Steel Loading State Component Gravity + 38 kip axial on column Bent Cap 4 in of lateral displacement (μ8.6) Bent Cap Column Column Bar Size Yield Stress (ksi) #5 #3 #5 #3 #5 #3 #5 #3 60.6 68.4 60.6 68.4 60.6 68.4 60.6 68.4 Tension Max. % of Stress Yield (ksi) 2.3 3.8 2.0 3.0 27.2 44.9 26.6 38.9 60.6 100 68.4 100 Compression Max. % of Stress Yield (ksi) 2.55 4.2 1.78 2.6 3.96 6.5 9.67 16.0 4.67 6.8 60.6 100 31.5 46.1 46 70 Lateral Force (kips) 60 Longitudinal rebar in column yields at 0.45 in 50 Peak lateral load of 65.5 kips 40 30 Concrete damage on compression side of column near joint 20 10 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 3.1: Model 1 Lateral Load-Lateral Displacement Curve 60 Lateral Force (kips) 50 Peak lateral load of 57 kips 40 Concrete damage on compression side of column near joint 30 20 10 0 0 0.5 1 1.5 2 2.5 3 3.5 Lateral Displacement (in) Figure 3.2: Model 2 Lateral Load-Lateral Displacement Curve 4 47 70 Lateral Force (kips) 60 50 40 30 Dip in response results in lower stiffness 20 Model 1 10 Model 2 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 3.3: Comparison of Lateral Load-Lateral Displacement Response for Models 1 and 2 2 Z-Stress Damage Paramter Damage approaches 2.0 2500 1.5 Damage of 1.0 corresponds to maximum stress (2,520 psi) 2000 1500 1 1000 0.5 500 Damage Parameter Vertical Compressive Stress (psi) 3000 Residual Strength 0 0 0 2 4 6 8 10 Analysis Time (seconds) Figure 3.4: Relationship between Damage and Strength (compression plotted as positive stress to show relationship for representative element in column plastic hinging region) 48 2 Analysis Time (seconds) 4 6 8 500 400 Z-stress (psi) 300 200 100 2 Maximum tensile stress (468 psi) Damage parameter approaches 2.0 Compressive stress due to gravity and axial load -200 1.5 1 0 -100 10 Damage Parameter 0 0.5 Z-stress Damage Parameter 0 Figure 3.5: Relationship between Tensile Strength and Damage (representative element in column plastic hinging region) 49 Figure 3.6: Concrete Damage Parameter Contours at 0.045 in (μ0.1) of Lateral Displacement (Initial Flexural Cracking of Column) Direction of lateral displacement Figure 3.7: Concrete Damage Parameter Contours Showing Initial Flexural Cracking in Bent Cap at 0.165 in (μ0.35) of Lateral Displacement (Bottom View) 50 Introduction of joint shear crack development Figure 3.8: Development of Cracking in Joint Region at 0.42 (μ0.9) in of Lateral Displacement Significant damage accumulation in column Shear crack through joint region Damage due flexure in bent cap Figure 3.9: Plot of Damage Parameter Contours at 4 in (μ8.6) of Lateral Displacement 51 Principal stress vectors indicating diagonal crack through joint Figure 3.10: Principal Stress Vectors at 0.27 in (μ0.58) of Lateral Displacement 52 Close-up of principal vectors through the joint Figure 3.11: Close-up of Principal Stress vectors at 0.27 in (μ0.58) of Lateral Displacement Figure 3.12: Principal Stress Vectors Overlaid on Damage Parameter Contours at 1.0 in (μ2.15) of Lateral Displacement 53 Units in psi Figure 3.13: Maximum Shear Stress at 0.35 in (μ0.75) of Lateral Displacement, prior to Development of Cracking in Joint Units in psi Figure 3.14: Close-up of Maximum Shear Stress in Joint Region at 0.35 in (μ0.75) of Lateral Displacement, prior to Development of Cracking in Joint 54 Concentration of shear stress Units in psi Figure 3.15: Maximum Shear Stress Distribution at 4 in (μ8.6) of Lateral Displacement Units in psi Figure 3.16: Close-up of Maximum Shear Stress in Joint Region at 4 in (μ8.6) of Lateral Displacement 55 Average Joint Shear Stress (psi) 350 300 250 Peak average shear stress (299 psi) 200 150 Concrete elements in stress concentration zone transition to residual strength after approaching a damage parameter of 2.0 100 50 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 3.17: Average Joint Shear Stress vs. Column Lateral Displacement 0.0012 Joint Rotation (rad) 0.001 0.0008 0.0006 0.0004 0.0002 0 0 0.5 1 1.5 2 2.5 3 Lateral Displacement (in) Figure 3.18: Joint Rotation vs. Column Lateral Displacement 3.5 4 56 Figure 3.19: Joint Rotation at 4 in (μ8.6) of Lateral Displacement (scaling factor of 50) Units in psf Figure 3.20: Longitudinal (X-Stress) Distribution under Gravity and 38 kip Axial Load on Column 57 Compressive stress caused by lateral displacement of column Tensile stress across entire section Units in psf Figure 3.21: Longitudinal Stress Distribution at 4 in (μ8.6) of Lateral Displacement Transfer of compressive stresses to vertical restraints Units in psf Figure 3.22: Minimum Principal Stress Contours under Gravity and Column Axial Load 58 Compression strut through joint Units in psf Figure 3.23: Minimum Principal Stress Contours at 4 in (μ8.6) of Lateral Displacement Units in psf Figure 3.24: Maximum Principal Stress Contours under Gravity and Column Axial Load 59 Units in psf Figure 3.25: Maximum Principal Stress Contours at 4 in (μ8.6) of Lateral Displacement Bent cap longitudinal bars Peak compressive stress of 3,333 psi (54% of yield) Units in psi Peak tensile stress of 2,308 psi (38% of yield) Figure 3.26: Stress in Column and Bent Cap Longitudinal Rebar under Gravity and Axial Column Load 60 Rebar reached yield stress Units in psi Tensile stress dissipated through joint Figure 3.27: Stress in Column and Bent Cap Longitudinal Rebar at 4 in (μ8.6) of Lateral Displacement 70000 Tensile Stress (psi) 60000 50000 Dip in stress possibly corresponding to damaged concrete 40000 30000 20000 10000 0 0 0.002 0.004 0.006 0.008 0.01 Tensile Strain Figure 3.28: Plot of Stress versus Strain for #5 Column Longitudinal on the South Side of the Bent Cap and Column Joint 61 0 0.002 0.004 0.006 0.008 0.01 0 Compression Stress (psi) -10000 -20000 -30000 -40000 -50000 -60000 -70000 Compression Strain Figure 3.29: Plot of Compression Stress vs. Compression Strain for #5 Column Longitudinal on the North Side of Column 45000 40000 Tensile Stress 35000 30000 25000 20000 15000 10000 5000 0 0 0.0002 0.0004 0.0006 0.0008 0.001 0.0012 0.0014 0.0016 Tensile Strain Figure 3.30: Plot of Tension Stress vs. Tension Strain for #5 Column Longitudinal on the South Side of Column for an Element Located at Mid-depth of Bent Cap 62 Hoops reached yield stress of 68,480 psi Units in psi Figure 3.31: Stress in Stirrups and Hoops at 4 in (μ8.6) of Lateral Displacement 70000 Tensile Stress (psi) 60000 50000 40000 30000 20000 10000 0 0 0.001 0.002 0.003 0.004 0.005 Tensile Strain Figure 3.32: Plot of Tensile Stress vs. Tensile Strain for #3 Stirrup in Column near the Bent Cap 63 Chapter 4 COMPARISON OF RESULTS TO SPECIMEN DATA This chapter compares results from the FEA with the results from the CIP and GD specimen. The lateral load-lateral displacement response of the FE models was compared against the response from of the CIP specimen, GD specimen, and the predicted responses of both specimens. Comparisons including Model 2 were limited to the lateral load-displacement response and all other comparisons were made using data from Model 1. Joint shear stress from Model 1 was compared with the CIP and GD specimen using a Table of values and a plot of the stress vs. lateral displacement of the column. Similar comparisons were made for joint rotation, cracking of concrete, the strain profiles of reinforcing steel. 4.1 Lateral Load-Lateral Displacement Response The lateral load-displacement response of both FE models compared reasonably well to the response of the CIP and GD specimens and specimen predictions. Model 2 is only considered in this section for comparison of lateral load-displacement response. Figure 4.1 shows the lateral load-displacement response of Model 1 plotted with the hysteretic envelope of the CIP test specimen and the predicted lateral loaddisplacement response for the CIP specimen. Model 1 closely matches the predicted CIP response through 4 in (μ8.6) of lateral displacement, except from 0.4 in to 1.0 in. It does not compare as closely against the hysteretic push envelope of the CIP specimen, as it predicts higher loads at most displacements greater than 0.5 in (μ1.07). The initial stiffness (disp. < 0.1 in) of Model 1 is, on average, twice as stiff as the predicted response 64 and 20% larger than the hysteretic envelope of the CIP specimen. Significant change in the stiffness of Model 1 begins when concrete elements begin accumulating damage on the south side of the column near the plastic hinging region at 0.36 in (μ0.77) of lateral displacement, which corresponds to 54 kips of lateral load. A similar change in stiffness is observed in the predicted response and the hysteretic envelope at approximately the same lateral displacement but with much different values of force. The predicted response shows a change in stiffness at approximately 58 kips and the hysteretic envelope shows stiffness change at approximately 48.2 kips. At displacements greater than 2 in (μ4.17), the force values for the CIP hysteretic envelope begin to decrease due to column cover spalling and eventual column bars buckling in the plastic hinging region. This was not replicated in either Model 1 or the predicted response. The lateral load-displacement response of Model 1 and the predicted envelope therefore continued to increase through 4 in (μ8.6) of lateral displacement. For the CIP predicted values, the increase in load is caused by the strain hardening of the rebar. The increase in load in the FE model may have been related to confinement effects. Figure 4.2 shows the load-displacement curve of Model 2 plotted with the hysteretic envelope of the CIP test specimen and the CIP predicted values. The lateral load-displacement response for Model 2 compares well to the hysteretic envelope through a 2.0 in (μ4.17) displacement of the test specimens. Model 2 has higher stiffness values for displacement less than 0.179 in (μ0.38), which is where the first dip in the lateral load-displacement response of Model 2 appears. After the dip, damaged concrete decreases the stiffness and appears to match the stiffness more accurately for the 65 hysteretic envelope. The second dip in the lateral load-displacement response again changes the stiffness even more significantly and this allows Model 2 to continue matching the response of the CIP test specimen. As previously explained, Model 2 does not capture the decrease in stiffness after 2.0 inches (μ4.17) of displacement, but it increases in capacity to approximately 56 kips at 4 in (μ8.6) of lateral displacement. The predicted lateral load-displacement response of the GD specimen is essentially the same as the predicted response from the CIP specimen. A similar comparison is made between Model 1 and the predicted response as was done for the CIP specimen. The lateral load-displacement response of the GD specimen is plotted with Model 1 on Figure 4.3. The lateral load-displacement response of Model 2 is compared against the GD specimen in Figure 4.4. The hysteretic envelope of the GD specimen compares reasonably well with Model 1 for lateral displacement less than 1 in and with Model 2 while lateral displacement is less than 2.2 in (μ4.73). Model 1 and Model 2 are, on average, within 5% of the hysteretic envelope while lateral displacement is less than 0.4 in. After 0.4 in of lateral displacement, Model 1 continues to be in close agreement with the hysteretic envelope through 1 in (μ2.15) of lateral displacement because it has a higher load capacity than Model 2. For a lateral displacement less than 1 in, Model 1 is within 5% of the hysteretic envelope. Both models exhibited the same trends as the CIP comparison. Tables 4.1 and 4.2 summarize and compare lateral load-displacement response at lateral displacements of 0.25, 0.5, 1.0, 2.0, and 3.0 in. 66 4.2 Joint Shear Stress Figure 4.5 shows a plot of the average shear stress in the joint of the CIP specimen, GD specimen, and Model 1. Overall response of Model 1 in predicting average shear stress in the joint is in agreement with the CIP specimen and the GD specimen. Table 4.3 provides a summarized comparison of joint shear stress values for Model 1, the CIP specimen, and the GD specimen. Model 1 has higher shear stresses compared to the CIP specimen for a lateral displacement of less than 0.3 in (μ0.65). As concrete elements begin to accumulate damage, the shear stress values begin to decrease. Average shear stresses in Model 1 are less than those in the CIP specimen for displacements greater than 0.3 in (μ0.65). Shear stresses from Model 1 are within 17% of the CIP specimen for displacements greater than 0.5 in (μ1.1). The difference eventually begins to drop below 10% as lateral displacement increases. Average shear stresses in Model 1, in general, agree more closely with the shear stress data of the GD specimen. For lateral displacement less than 2.6 in (μ5.6), Model 1 generally has lower shear stress values that compare more closely with the GD data than with the CIP data. Model 1 has higher shear stresses when the displacement is larger than 0.6 in (μ1.3). Shear stresses in the GD specimen begin to decrease relatively quickly after 2.25 in (μ4.8) of lateral displacement compared to the CIP specimen and Model 1. This decrease is due to the loss of load corresponding to spalling and buckling of column longitudinal bars in the plastic hinging region. The shear stresses in Model 1 remain relatively constant after 1.6 in (μ3.4) of lateral displacement. The lower shear stress is partially attributed to the non-strain 67 hardening steel model used for the rebar. After yielding, the steel does not pick up any additional load, which does not put additional tensile force on the joint. Increases in stress beyond this point are caused by the confinement effects as mentioned earlier. 4.3 Joint Rotation Figure 4.6 compares the joint rotation for increasing displacement from Model 1, the CIP specimen hysteretic envelope, and the GD hysteretic envelope. The joint rotation from Model 1 is significantly less than both CIP and GD specimens, indicating a much stiffer response in the joint. Table 4.4 summarizes and compares joint rotation data from the Model and both specimen at various displacements, revealing that Model 1 reached on average 65% of the CIP rotation and 45% of the GD rotation. 4.4 Principal Stress Figure 4.7 shows a plot of the average maximum principal joint stress with increasing displacement in Model 1, the CIP specimen, and the GD specimen. Maximum principal stresses in Model 1 do not compare well with those of the CIP or GD specimens. For a displacement greater than 0.2 in (μ0.43), principal stresses in Model 1 are up to 40% less than those in the CIP and GD specimens. For displacements greater than 2.0 in (μ4.3), stresses in both specimens decrease because of spalling and buckling of longitudinal rebar in the plastic hinge region of the column, but Model 1 stresses remain relatively constant. However, Figure 4.8 shows that the minimum principal joint stresses for Model 1 compare reasonably well with CIP and GD specimen data. For lateral displacement between 0.2 in (μ0.43) and 2.5 in (μ5.4), Model 1 principal stresses are within 15% of 68 those from the CIP and GD specimens. For displacements larger than 2.5 in (μ5.4), Model 1 stresses continue increasing whereas those from the CIP and GD specimens decrease. Principal stresses in Model continued to increase because of confinement effects and decreased in the CIP and GD specimens because of a spalling and buckling of longitudinal column reinforcement in the plastic hinge region. 4.5 Stages of Concrete Cracking Table 4.5 summarizes and compares Model 1 to the CIP specimen for flexural cracking of the bent cap, diagonal shear cracking in the joint, and flexural cracking of the column. Concrete elements south of the joint on the bottom face of the bent cap in model show cracking, based on the damage parameter contours, at 0.15 inches (μ0.32) of lateral displacement, which corresponded to a lateral load of 35 kips. Similar cracking was observed in the CIP specimen at 13 kip of lateral load, which is 63% less than Model 1. Based on the damage parameter contours, a diagonal crack in the joint of Model 1 forms at a lateral displacement of 0.345 inches (μ0.75), which corresponds to a lateral force of 51 kips. Shear cracking in the CIP specimen joint was observed at 48 kips of lateral force, which is 6% less than Model 1. Figures 4.9 through 4.12 show the damage contours from Model 1 and pictures of the CIP specimen at the same displacement, to compare cracking. Figure 4.9 compares cracking of Model 1 to the CIP specimen at 48 kips of lateral load, which is prior to the development of the primary diagonal joint crack in Model 1. Figure 4.10 compares cracking of Model 1 to that of the CIP specimen at 1.4 inches (μ3) of lateral displacement. At this displacement, the diagonal shear crack through the joint is visible on Model 1 and the CIP specimen. Cracking of Model 1 and 69 the CIP specimen at 2.8 in (μ6) of lateral displacement is shown in Figure 4.11. At this lateral displacement, the diagonal shear crack in Model 1 has formed through the length of the joint and significant damage is observed in the column. At 3.2 in (μ6.88) of lateral displacement, Figure 4.12 shows significant damaged concrete in Model 1 and the CIP specimen, and it appears to correlate well with one another. Much of the damage, in both Model 1 and the CIP specimen, was limited to the column plastic hinge region, with the exception of the predicted diagonal joint shear crack. It should be emphasized that cracking on the test specimen is due to reverse cyclic loading, whereas as damage in Model 1 is due to monotonic (push) load. 4.6 Reinforcing Steel Strain Profiles Table 4.6 compares the steel strain profile in the column longitudinal reinforcement located in the plastic hinge region and through the joint of Model 1 and the CIP specimen. A cross section of the CIP column showing rebar ID LC8, which is the bar analyzed in Table 4.6, is shown in Figure 4.13. Because this comparison considers the push load case, LC8 is the extreme tensile rebar. Figure 4.14 shows the three locations of the strain gauges: 6 in into the column, 1 in into the joint, and 6 in into the joint. CIP specimen strain gauge locations for bent cap stirrups and column hoops are shown in Figure 4.15. Table 4.7 compares the strain profile of the stirrups in the bent cap at six locations: mid-height of two stirrups adjacent to the south side of the joint, upper side of two stirrups in the joint, and mid-height of two stirrups adjacent to the north side of the joint. Strain values from Model 1 and the CIP specimen were well below yield strain and in many cases so small that a meaningful comparison was difficult. In some 70 cases, the Model 1 strain values were negative of the CIP specimen. Table 4.8 was formulated similarly to Table 4.7 with the exception of using the bottom of joint stirrups instead of the top. Table 4.9 compares the strain profile on the east side of five hoops: three located in the bent cap and two in the column (Figure 4.15). Model 1 exhibited an unusual strain pattern in column hoops that was not in agreement with the CIP specimen. Maximum strain in reinforcing hoops was expected on the transverse face of the stirrups, as seen on the CIP specimen. However, in Model 1, the maximum strain occurred on the south side of the column. Because of the unusual strain distribution in Model 1, strain was not expected to be comparable with the CIP specimen. CIP hoops achieved large strains beyond yield, hoops strains the same location on Model 1 were significantly less, and reversed in sign in some cases. 71 Table 4.1: Comparison of Lateral Load-Lateral Displacement Response between Model 1, CIP Specimen, GD Specimen, and Specimen Predictions Lateral Disp. (in) Disp. Ductility, μ 0.25 0.5 1 2 3 0.54 1.08 2.15 4.30 6.45 Lateral Force (kips) Model 1 CIP 45 52.1 58.5 63 63.9 GD Ratio Specimen Model 1/ Predicted CIP 39 40 49 51 55.8 56.4 55.9 55.5 52 - 48 58.5 59 62 64.4 Model 1/ GD Model 1/ Predicted 1.13 1.02 1.04 1.14 - 0.94 0.89 0.99 1.02 0.99 1.15 1.06 1.05 1.13 1.23 Table 4.2: Comparison of Lateral Load-Lateral Displacement Response between Model 2, CIP Specimen, GD Specimen, and Specimen Predictions Lateral Force (kips) Lateral Disp. (in) Disp. Ductility (μ) Model 2 0.25 0.5 1 2 3 0.54 1.08 2.15 4.30 6.45 41.2 46.1 54.3 56.7 65.8 CIP GD 39 40 49 51 55.8 56.4 55.9 55.5 52 - Ratio Predicted Model 2/ CIP Model 2/ GD Model 2/ Predicted 48 58.5 59 62 64.4 1.06 0.94 0.97 1.01 1.27 1.03 0.90 0.96 1.02 - 0.86 0.79 0.92 0.91 1.02 Table 4.3: Comparison of Average Joint Shear Stresses between Model 1, CIP Specimen, and the GD Specimen Average Joint Shear Stress (psi) Ratio Lateral Displacement (in) Disp. Ductility (μ) Model 1 CIP GD Model 1/ CIP Model 1/ GD 0.25 0.5 1 2 3 0.54 1.08 2.15 4.30 6.45 176 236 276 285 284 140 264 325 322 303 197 274 309 303 233 1.26 0.89 0.85 0.89 0.94 0.89 0.86 0.89 0.94 1.22 72 Table 4.4: Comparison of Joint Rotation between Model 1, the CIP Specimen, and the GD Specimen Lateral Displacement (in) Disp. Ductility (μ) Model 1 CIP GD 0.25 0.5 1 2 3 0.54 1.08 2.15 4.30 6.45 0.000401 0.000635 0.000778 0.000908 0.000958 0.00045 0.00093 0.00129 0.00162 0.00183 0.00091 0.00143 0.00186 0.00221 0.00183 Rotation (rad) Ratio Model 1/ Model 1/ CIP GD 0.89 0.68 0.61 0.56 0.52 0.44 0.44 0.42 0.41 0.53 Table 4.5: Comparison of Specimen Cracking – Including 38 kip Axial Load and SelfWeight Lateral Load (kips) Stage Bent Cap - Flexural Column - Flexural Joint Shear Ratio Predicted CIP Test Model 1 Model 1/ Predicted Model 1/ CIP 8.9 14 36.5 13 20 48 35 25 50 3.93 1.79 1.37 2.69 1.25 1.04 73 Table 4.6: Strain Profile Table – Column Longitudinal Rebar (LC8), CIP Specimen Location C1 B1 B2 Bar Strain (με) Lateral Load/Ductility Disp. (in) 13 kips 20 kips 30 kips 48 kips μ1 μ1.5 13 kips 20 kips 30 kips 48 kips μ1 μ1.5 13 kips 20 kips 30 kips 48 kips μ1 μ1.5 0.034 0.072 0.154 0.397 0.406 0.533 0.034 0.072 0.154 0.397 0.406 0.533 0.034 0.072 0.154 0.397 0.406 0.533 Ratio Model 1 CIP Model 1/ CIP 91.5 577 907 1620 1590 1890 415 580 1220 1850 1850 2002 80.4 124 517 1550 1570 1700 389 999 1875 3249 1757 1711 1391 2228 2457 3019 5589 5595 97 553 1080 1387 813 1192 0.235 0.578 0.484 0.499 0.905 1.105 0.298 0.260 0.497 0.613 0.331 0.358 0.829 0.224 0.479 1.118 1.931 1.426 74 Table 4.7: Strain Profile Table – Stirrups in Bent Cap (Mid-height) and Joint (Top), CIP Specimen Lateral Location Load/Displacement Ductility SS5E SS1E-T SN1E-T SN3E SN5E Bar Strain (με) Disp. (in) Ratio Model 1 CIP Model 1/ CIP 13 kips 20 kips 30 kips 48 kips μ1 0.034 0.072 0.154 0.397 0.406 1.24 2.23 0.154 -6 -6.56 -15 -14 -26 -91 -55 -0.083 -0.159 -0.006 0.066 0.119 μ1.5 13 kips 20 kips 30 kips 48 kips μ1 0.533 0.034 0.072 0.154 0.397 0.406 1.72 9.24 19.5 11.5 77.5 99.9 -85 63 175 273 690 504 -0.020 0.147 0.111 0.042 0.112 0.198 μ1.5 13 kips 20 kips 30 kips 48 kips μ1 0.533 0.034 0.072 0.154 0.397 0.406 336 -12.4 -9.78 29.2 -29.8 -36.3 630 -1 35 57 94 104 0.533 12.400 -0.279 0.512 -0.317 -0.349 μ1.5 13 kips 20 kips 30 kips 48 kips μ1 0.533 0.034 0.072 0.154 0.397 0.406 -41.8 -14.7 -19.2 -1.95 -75.8 -71.5 151 -13 -151 -189 -307 -131 -0.277 1.131 0.127 0.010 0.247 0.546 μ1.5 13 kips 20 kips 30 kips 48 kips μ1 0.533 0.034 0.072 0.154 0.397 0.406 -93.5 -7.36 -10.5 -23.7 -139 -137 -147 -40 -48 -34 84 40 0.636 0.184 0.219 0.697 -1.655 -3.425 μ1.5 0.533 -143 96 -1.490 75 Table 4.8: Strain Profile Table – Stirrups in Bent Cap (Mid-height) and Joint (Bottom), CIP Specimen Lateral Location Load/Displacement Ductility SS5E SS3E SS1E-B SN1E-B SN3E SN5E μ2 μ2 μ2 μ2 μ2 μ2 Bar Strain (με) Disp. (in) 0.684 0.684 0.684 0.684 0.684 0.684 Ratio Model 1 CIP Model 1/ CIP 3.78 121 511 -47.64 -89.4 -145 -81 -10 -205 3307 -180 112 -0.047 -12.100 -2.493 -0.014 0.497 -1.295 Table 4.9: Strain Profile Table – Hoops in Column and Joint (East), CIP Specimen Lateral Location Load/Displacement Ductility HC4-E HC1-E HB3-E HB4-E HB5-E 30 kips μ2 30 kips μ2 30 kips μ2 30 kips μ2 30 kips μ2 Bar Strain (με) Disp. (in) 0.154 0.684 0.154 0.684 0.154 0.684 0.154 0.684 0.154 0.684 Ratio Model 1 CIP Model 1/ CIP 50.1 498 -58.3 4.3 5.4 -3.5 37.3 9.2 67.6 159 619 1465 295 2234 11 37 42 1288 161 542 0.081 0.340 -0.198 0.002 0.493 -0.095 0.888 0.007 0.420 0.293 76 70 Lateral Force (kips) 60 50 Model 1 continues to follow predicted response through 4 in 40 Significant change in stiffness occurs at similar displacement but different forces 30 20 Model 1 CIP Test Specimen Predicted Response 10 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 4.1: Model 1 and CIP specimen Lateral Load-Lateral Displacement Response Comparison 70 Lateral Force (kips) 60 50 40 CIP load decreases due to spalling and buckling of rebar in plastic hinge Significant change in stiffness of Model 2 and continues to follow CIP specimen 30 20 Slight decrease in stiffness of Model 2 following concrete damage 10 Model 2 CIP Test Specimen Predicted 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 4.2: Model 2 and CIP Specimen Lateral Load-Lateral Displacement Comparison 77 70 Lateral Force (kips) 60 50 40 Model 1 is, on average, within 5% of GD Specimen through 1 in of displacement 30 20 Model 1 Predicted Response GD Specimen 10 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 4.3: Model 1 and GD Specimen Lateral Load-Lateral Displacement Comparison 70 Lateral Force (kips) 60 50 40 Smaller force than Model 1 but is still within 10% of GD Specimen 30 20 Model 2 Predicted Response GD Specimen 10 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 4.4: Model 2 and GD Specimen Lateral Load-Lateral Displacement Comparison 78 350 Average Joint Stress (psi) 300 250 200 150 100 Model 1 CIP Specimen GD Specimen Larger initial shear stress in Model 1 50 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 4.5: Comparison of Average Joint Shear Stresses – Model 1 and CIP and GD Specimens 0.0024 Joint Rotation (rad) 0.002 0.0016 0.0012 0.0008 Model 1 CIP Specimen GD Specimen Concrete damage in column causes a jump in displacement but minor joint rotation 0.0004 0 0 0.5 1 1.5 2 2.5 3 3.5 Lateral Displacement (in) Figure 4.6: Comparison of Joint Rotation - Model 1 and CIP and GD Specimens 4 79 400 Maximum Principal Stress (psi) 350 300 250 200 150 100 Model 1 CIP Specimen GD Specimen Initial stress from gravity and axial load in column 50 0 0 0.5 1 1.5 2 2.5 3 3.5 4 Lateral Displacement (in) Figure 4.7: Maximum Principal Joint Stresses from Model 1and CIP and GD Specimens 400 Stresses continue to increase due to confinement effects Minimum Principal Stress (psi) 350 300 250 200 150 100 Initial stress from gravity and axial load in column Model 1 CIP Specimen GD Specimen 50 0 0 0.5 1 1.5 2 2.5 3 3.5 Lateral Disaplacement (in) Figure 4.8: Minimum Principal Joint Stress - Model 1 and CIP and GD Specimens 4 80 Figure 4.9: Comparison of Concrete Cracking at 48 kips of Lateral Load – Model 1 and CIP Specimen Figure 4.10: Comparison of Concrete Cracking at 1.4 in (μ3) of Lateral Displacement – Model 1 and CIP Specimen Figure 4.11: Comparison of Concrete Cracking at 2.8 in (μ6) of Lateral Displacement – Model 1 and CIP Specimen 81 Figure 4.12: Comparison of Concrete Cracking at 3.2 in (μ6.9) of Lateral Displacement – Model 1 and CIP Specimen Figure 4.13: Section of Column Shows Location of LC8 Rebar (Matsumoto, 2009) 82 Figure 4.14: Location of Longitudinal Rebar Strain Gauges on the CIP Specimen (Matsumoto, 2009) 83 Figure 4.15: Location of Stirrup Strain Gauges on the CIP Specimen (Matsumoto, 2009) 84 Chapter 5 SUMMARY, CONCLUSIONS, AND RECOMMENDATIONS 5.1 Summary This report, 3-D Finite Element Modeling of Reinforced Concrete Beam-Column Connections – Development and Comparison to NCHRP 12-74, investigates the use of finite element modeling (FEM) to predict the structural response of the cast-in-place (CIP) reinforced concrete bent cap-column test specimen reported in NCHRP Report 681 – Development of a Precast Bent Cap System for Seismic Regions. The response of the FE model was compared to the response of the CIP bent capcolumn test specimen using several different parameters, including the following: Column Lateral load-lateral displacement response Joint shear stress Joint rotation Magnitude and orientation of principal stresses Stages of concrete cracking Yielding pattern of steel reinforcement in bent and column Overall distribution of flexural and compressive stresses The FE model was based on the same dimensions as the 42% scale inverted tee CIP test specimen, as provided in drawings of NCHRP Report 681, and includes two solid components, a bent cap and a column, and associated steel reinforcement. The primary FE model, Model 1, consists of 55,860 solid concrete elements and 4,976 beam elements, for 60,836 total elements. The solid concrete elements are eight node hexahedron 85 elements with constant stress solid element formulation and the steel reinforcement elements are circular beam elements. The concrete and rebar is coupled using the constrained Lagrange card. The Karagozian and Case damaged concrete model (MAT_072), was used as the constitutive model for the concrete. Rebar was modeled with a plastic kinematic constitutive model, without strain hardening. The analysis included monotonic push loading on the FE model (due to various limitations of the constitutive models) instead of reverse cyclic loading. Therefore FEA results were compared to the hysteretic envelope from the cyclic test results of the test specimen. To minimize inertial effects, loading on the FE model was applied over a 10 second interval: gravity and a 38 kip axial load were applied first, followed lateral displacement applied to the column stub. The lateral load-lateral displacement response of the FE model (Model 1) compared reasonably well to the actual lateral load-displacement response of the CIP specimen and the GD specimen. For lateral displacements less than those corresponding to a displacement ductility of 4.1, the FE model showed a larger stiffness relative to the specimen response. Stiffness of Model 1 degraded as an increasing number of concrete elements accumulated damage in the plastic hinging region, but this degradation was not a severe as that of the CIP specimen. For displacement ductility greater than 2, the lateral load-displacement response matched the predicted response within 5%. Average joint shear stresses in the FE model and the CIP specimen compared reasonably well with one another. For lateral displacements less than a displacement ductility of 1.3, the FE model closely matched the average joint shear stress response of 86 the CIP specimen. For lateral displacements larger than a displacement ductility of 1.3, the FE model, on average, had 10% less average joint shear stress. Joint rotation in the FE model was on average 60% of the CIP specimen, for lateral displacement larger than a displacement ductility of 1.1. Concrete damage parameter contours of the FE model correlated reasonably well with cracking and spalling in the CIP specimen. The FE model developed damage in the bent cap due to flexure at 35 kips of lateral (includes 38 kip axial load on column and self weight) load, whereas the CIP specimen developed cracks at 13 kips of lateral load, which is 63% less than the FE model. Diagonal joint shear damage developed in the FE model at 51 kips, compared to 48 kips in the CIP specimen, which is 6% less than the FE model. Concrete damage contours at a lateral displacement ductility of 8.6 compared reasonably well with cracking observed in the CIP specimen. The FE model had minimal damage in the joint region, with the exception of a predicted diagonal shear crack, and significant damage in the column near the plastic hinging region. The damage reflects the design mode of failure, which was plastic hinging of the column. Principal stress vectors in the FE model, prior to joint shear damage, show the orientation and magnitude of tensile stress development along the diagonal. The orientation of maximum principal stress vectors is in agreement with flexural damage observed from the damage parameter contours. 5.2 Conclusion Based on the results of this analysis, explicit nonlinear finite element analysis is suitable for predicting the response of reinforced concrete beam-column connections. 87 Particularly, LS-DYNA combined with proper constitutive models has proven to provide accurate and reliable data when compared against CIP test specimen data. The following conclusions were made: a. Finite element modeling, using appropriate constitutive models and element formulation, can accurately capture the nonlinear behavior of reinforced concrete beam-column connections. A well-developed model can reasonably approximate the lateral load-lateral displacement response, flexural cracking, joint shear cracking, steel reinforcement yielding, and overall stress distribution, as compared to the CIP specimen. b. Element size for concrete and steel reinforcement significantly impacts the overall response and accuracy. Ideally, several iterations with decreasing element size should be performed in order to more accurately determine the effect of mesh size. For this analysis, decreasing the element size resulted in a more steady response of stress distribution, displacement, and more accurate concrete damage contours. c. The Karagozian & Case damaged concrete model (MAT_072) can accurately capture the cracking of concrete using a limited number of inputs (f ’c and aggregate size) into the model. Flexural and shear cracking patterns correlated closely with CIP test specimen data when using this constitutive model. 5.3 Recommendations Recommendations for further analysis include the following: 88 a. Incorporate the use of a strain hardening in the constitutive model for steel reinforcement in order to replicate the CIP test specimen more closely. This would also provide further analysis between strain hardening and confinement effects of concrete. b. Strain distribution of steel reinforcement in the joint (longitudinal, joint hoops, and joint stirrups) and plastic hinging region should be further investigated to address issues with unrealistic distributions observed in some cases. c. Incorporate a concrete constitutive model capable of accurately modeling reverse cyclic loading. Incorporating a damaged model capable of coupling the effects of tension and compression damage would allow the FE model to be analyzed more closely to specimen testing. d. Calibration of constitutive model parameters should be performed when attempting to match specimen data. This would require a more detailed and rigorous analysis but would result in close match between actual specimen data and FEA results. e. Incorporate the effects of bar slip by utilizing a bar slip model for the bond between concrete and steel reinforcement. f. Further investigation of the effects of beam element lengths of hoop steel as it affects the overall response of the structure. 89 REFERENCES Restrepo, Jose I., Matthew J. Tobolski, and Eric E. Matsumoto. Development of a Precast Bent Cap System for Seismic Regions. Publication no. 681. National Cooperative Highway Research Program, 2011. Print. Matsumoto, Eric E., Emulative Precast Bent Cap Connections for Seismic Regions: Component Tests – Cast-in-place Specimen (Unit 1), ECS Report No. ECS-CSUS-200901, California State University, Sacramento, CA (November 2009) 89 pp. Mills-Bria, Barbara. State-of-practice for the Nonlinear Analysis of Concrete Dams at the Bureau of Reclamation. [Denver, Colo.]: U.S. Dept. of the Interior, Bureau of Reclamation, 2006. Print. Predictive Engineering. “LS-DYNA Analysis for Structural Mechanics.” Predictive Engineering, Inc. Presentation. 2011. 11 November 2012 http://www.predictiveengineering.com/Solutions/products/training/docs/ls-dyna/LSDYNA_Analysis_for_Structural_Mechanics-Partial_Note_Set.pdf Wu, Youcai, John E. Crawford, and Joseph M. Magallanes. Performance of LS-DYNA Concrete Constitutive Models. Proc. of 12th International LS-DYNA Users Conference, Hyatt Regency Dearborn, Dearborn, Michigan. Print. Livermore Software Technology Corporation (LSTC). LS-DYNA Keyword User’s Manual. [Livermore, CA.]: Livermore Software Technology Corporation, 2007. Print. 90 Joseph M. Magallanes, Wu, Youcai, L. Javier Malvar. Recent Improvements to Release III of the K&C Concrete Model. Proc. of 11th International LS-DYNA Users Conference, Hyatt Regency Dearborn, Dearborn, Michigan. Print. Malvar, L. Javier, and Don Simons. Concrete Material Modeling in Explicit Computations. Proc. of Workshop on Recent Advances in Computational Structural Dynamics and High Performance Computing, USAE Waterways Experiment Station. April 24, 1996. Web. 11 Nov. 2012. Brannon, Rebecca M., and Seubpong Leelavanichkul. Survey of Four Damage Models for Concrete. Rep. no. SAND2009-5544. Livermore, CA: Sandia National Laboratories, 2009. Print. Schwer, Leonard E., and L. Javier Malvar. Simplified Concrete Modeling with *MAT_CONCRETE_DAMAGE_REL3. Proc. of JRI LS-DYNA User Week 2005. N.p., n.d. Web. 12 Nov. 2012. <http://www.geomaterialmodeling.com/pdf/88briefly.pdf>. Sritharan, S., M.J. Nigel Priestley, and Frieder Seible. "Nonlinear Finite Element Analyses of Concrete Bridge Joint Systems Subjected to Seismic Actions." Finite Elements in Analysis and Design 36 (2000): pp. 215-233. Elsevier. Web. 12 Nov. 2012. Sritharan, Sri. "Strut-and-Tie Analysis of Bridge Tee Joints Subjected to Seismic Actions."Journal of Structural Engineering, American Society of Civil Engineers, Reston, VA, Vol. 131, No. 9 (2005) pp.1321-1333 91 Sritharan, Sri. "Improved Seismic Design Procedure for Concrete Joints."Journal of Structural Engineering, (2005) American Society of Civil Engineers, Reston, VA, Vol. 131, No. 9 (2005) pp.1334-1344 Bermejo, Mario, Jose M. Goicolea, Felipe Gabaldon, and Anastasio Santos. Proc. of 3rd ECCOMAS Thematic Conference on Computational Methods in Structural Dynamics and Earthquake Engineering, Corfu, Greece. Web. 12 Nov. 2012. <http://oa.upm.es/12227/1/INVE_MEM_2011_99522.pdf>. Matsumoto, Eric E., Emulative Precast Bent Cap Connections for Seismic Regions: Component Tests – Grouted Duct Specimen (Unit 1), ECS Report No. ECS-CSUS-200902, California State University, Sacramento, CA (November 2009) 102 pp.