• Federal Highway Administration >
• Publications >
• Research Publications >
• 04096 >
• Nondestructive Evaluation >
• Evaluation of LS-DYNA Wood Material Model 143

Evaluation of LS-DYNA Wood Material Model 143

PDF files can be viewed with the Acrobat® Reader®

View Table of Contents

FOREWORD

This report documents the evaluation of a wood material model that has been implemented into the dynamic finite element code LS-DYNA, beginning with version 970. This material model was developed specifically to predict the dynamic performance of wood components used in roadside safety structures when undergoing a collision by a motor vehicle. This model is applicable for all varieties of wood when appropriate material coefficients are inserted. Default material coefficients for two wood varieties, southern yellow pine and Douglas fir, are stored in the model and can be accessed for use.

This report is one of two that completely documents this material model. The first report, Manual for LS-DYNA Wood Material Model 143 (FHWA-HRT-04-097), completely documents this material model for the user. The second report, Evaluation of LS-DYNA Wood Material Model 143 (FHWA-HRT-04-096), completely documents the model’s performance and the accuracy of the results. This performance evaluation was a collaboration between the model developer and the model evaluator. Regarding the model’s performance evaluation, the developer and the evaluator were unable to come to a final agreement regarding the model’s performance and accuracy. These disagreements are itemized and thoroughly discussed in chapter 17 of the second report.

This report will be of interest to research engineers associated with the evaluation and crashworthy performance of roadside safety structures, particularly those engineers responsible for the prediction of the crash response of such structures when using the finite element code LS-DYNA.

Michael F. Trentacoste, Director
Offices of Safety Research and Development

Preface

The goal of the work performed under this program, Development of DYNA3D Analysis Tools for Roadside Safety Applications, is to develop wood and soil material models, implement the models into the LS-DYNA finite element code,(¹) and evaluate the performance of each model through correlations with available test data.

Two reports are available for each material model. One report is a user’s manual; the second report is a performance evaluation. The user’s manual, Manual for LS-DYNA Wood Material Model 143,(²) thoroughly documents the wood model theory, reviews the model input, and provides example problems for use as a learning tool. It is written by the developer of the model. This report, Evaluation of LS DYNA Wood Material Model 143, comprises the performance evaluation for the wood model. It documents LS-DYNA parametric studies and correlations with test data performed by the model developer, and by a potential end user. The reader is urged to review the user’s manual before reading this evaluation report. A user’s manual(³) and evaluation report(⁴) are also available for the soil model.

The development of the wood model was conducted by the prime contractor. The associated wood model evaluation effort to determine the model‘s performance and the accuracy of the results was a collaboration between the developer and the potential end user, with the user’s evaluation intended to be independent of the developer’s evaluation. The developer partially evaluated the wood model. The potential end user performed a second independent evaluation of the wood model, provided finite element meshes for the evaluation calculations, and provided static post and bogie impact test data for correlations with the model.

Regarding the second independent evaluation of the wood model, the developer and evaluator were unable to come to a final agreement regarding several issues associated with the model‘s performance and accuracy. These issues are itemized and thoroughly discussed by the developer in chapter 17 of this evaluation report.

Throughout this report, the developer of the wood material model is referred to as the developer. The potential end user of the wood material model is referred to as the user. The developer’s calculations and conclusions are given in chapters 1 through 8 of this report. The user’s calculations and conclusions are given in chapters 9 through 16 of this report.

SI* (Modern Metric) Conversion Factors

Table of Contents

Page

1. Developer's Introduction

   1.1 Model Theory
   1.2 Model Input
   1.3 Limitations of Laboratory Material Property Data
   1.4 Evaluation Process
   1.5 Verification and Validation

2. Single-Element Simulations

3. Timber Compression Test Correlations

4. Timber-Bending Test Correlations

5. Quasi-Static Post Test Correlations

   5.1 Southern Yellow Pine Bending Test Data
   5.2 LS-DYNA Correlations
   5.3 LS-DYNA Parametric Studies

6. Dynamic Post Test Correlations

   6.1 Bogie Impact Test Data
   6.2 LS-DYNA Correlations
   6.3 Filtering and Sampling Issues
   6.4 LS-DYNA Parametric Studies

7. Additional Evaluation Calculations

   7.1 Plasticity Algorithm Iterations
   7.2 Fully Integrated Elements
   7.3 Erosion Criteria
   7.4 Post-Peak Hardening Parameter

8. Developer's Summary and Recommendations

9. User's Introduction

10. Validation Criteria for the Wood Material Model

    10.1 NDOR Tests: Performance Envelopes

11. Verification of Results on Different Computer Platforms

    11.1 Single-Element Models
    11.2 Dynamic Post Test Simulation: Bogie Model
    11.3 Dynamic Post Test Simulation: Fast Bogie Model

12. Single Element: Tension Parallel to the Grain

    12.1 Stress-Strain Behavior: *MAT_WOOD_PINE
    12.2 Volume of Element: *MAT_WOOD_PINE
    12.3 Material Properties: *MAT_WOOD_PINE

13. Static Wood Post Test Simulations

    13.1 Static Post Model
    13.2 Baseline Model Versus Test Comparison
    13.3 Baseline Versus Refined-Mesh Comparison
    13.4 Parameter Study

14. Static Wood Post Test Simulations

    14.1 Dynamic Post Model
    14.2 Vaporization and Time Step
    14.3 Sharp Edge Contacts
    14.4 Bending
    14.5 Further Analysis

15. Element Formulation: Hourglassing

16. User's Conclusions

17. Developer's Comments on User's Evaluation

    17.1 Table of Wood Model Topics
    17.2 Discussion of Wood Model Topics
    17.3 Instabilities in Dynamic Analyses

18. References

List of Figures

• Figure 1. LS-DYNA simulations of southern yellow pine showing brittle behavior in tension and shear, and ductile behavior in compression.
• Figure 2. Good correlation between LS-DYNA simulations (dashed lines) and measured clear wood data (solid lines) for southern yellow pine in tension parallel to the grain.
• Figure 3. Good correlation between LS-DYNA simulations (dashed lines) and measured clear wood data (solid lines) for southern yellow pine in compression perpendicular to the grain.
• Figure 4. Schematic of FPL timber compression test setup.
• Figure 5. These comparisons of the model with test data were used to set the hardening behavior of the southern yellow pine model in parallel-to-the-grain compression.
• Figure 6. These comparisons of the model with the parallel-to-the-grain timber-bending test data demonstrate the need for different quality factors in tension and compression.
• Figure 7. Quasi-static post test setup.
• Figure 8. Details for the rigid frame used in the quasi-static post test setup.
• Figure 9. Deformed configurations of posts in static tests.
• Figure 10. Closeup of damage observed in break region of posts in static tests.
• Figure 11. Measured load-deflection histories exhibit sudden drops in force as the post fails in the tensile region.
• Figure 12. Quasi-static (blue solid line) and dynamic (pink dashed line) performance envelopes developed and plotted by the user.
• Figure 13. Grades 1 and 1D static post test measurements exhibit substantial scatter.
• Figure 14. Good correlations between the LS-DYNA calculations and quasi-static performance envelopes determine the default quality factors for grade 1 southern yellow pine of Q_T = 0.47 with Q_C = 0.63.
• Figure 15. Good correlations between the LS-DYNA calculations and quasi-static performance envelopes determine the default quality factors for DS-65 southern yellow pine of Q_T = 0.80 with Q_C = 0.93.
• Figure 16. The parallel-to-the-grain post fracture calculated in the post just below the top of the rigid support is in agreement with the location of the damage observed in the tests.
• Figure 17. Three geometric models were set up for performing parametric calculations.
• Figure 18. The force-deflection curve calculated with the fast-running, rigid-wall model is similar to that calculated with the full-support structure, making it useful for performing parametric calculations.
• Figure 19. Pinning a truss to the bolt to simulate the vertical force applied by the pulley system increases the peak force by about 4 percent.
• Figure 20. The peak force increases with the applied velocity until convergence is attained at 0.25 mm/ms.
• Figure 21. Quasi-static calculations are filtered for easier comparison with the unfiltered test data.
• Figure 22. The energy required to break the post increases as the value of the parallel-to-the-grain fracture energy increases.
• Figure 23. Calculations with Q_C > Q_T soften more rapidly than calculations with Q_C = Q_T.
• Figure 24. The calculated peak force increases with increasing values of the softening parameter B.
• Figure 25. The softening parameter B determines the shape of the softening curve in these single-element simulations for tension parallel to the grain.
• Figure 26. Type 4 stiffness control reduces hourglassing better than type 3 viscous control.
• Figure 27. The peak force calculated with type 4 stiffness control agrees with the fully integrated element calculation better than the type 3 viscous calculation.
• Figure 28. Less final hourglass energy is calculated with type 4 stiffness control than with type 3 viscous control.
• Figure 29. Decreasing the moisture content by 6 percent increases the post-peak force by 80 percent.
• Figure 30. Post setup for dynamic bogie impact tests.
• Figure 31. Processed data from the user's bogie impact tests.
• Figure 32. Damage in the breakaway region of the posts, just below ground level.
• Figure 33. These correlations between the LS-DYNA calculations and the performance envelopes were used to adjust the parallel fracture energy for grade 1 pine to 50 times the perpendicular fracture energy.
• Figure 34. These correlations between the LS-DYNA calculations and the performance envelopes were used to adjust the parallel fracture energy for DS 65 pine to 85 times the perpendicular fracture energy.
• Figure 35. These correlations between the LS-DYNA calculations and the energy-deflection data were used to confirm that grade 1 Douglas fir can be simulated with the same parallel fracture energy and rate-effect parameters as grade 1 southern yellow pine, but with different quality factors.
• Figure 36. These correlations between the LS-DYNA calculations and the energy-deflection data were used to set the parallel fracture energy for frozen grade 1 pine to five times the perpendicular fracture energy.
• Figure 37. The simulated post breaks just below ground level, in agreement with the failure location observed in the tests.
• Figure 38. User's geometric bogie model used in the developer's calculations.
• Figure 39. The filtered bogie acceleration history sampled at 3200 Hz is significantly different than the history sampled at 10,000 Hz.
• Figure 40. Force-deflection histories calculated from bogie nodal accelerations depend significantly on the bogie type and output sampling frequency.
• Figure 41. Force-deflection histories calculated from the wood post cross-sectional forces are not significantly affected by the details for the bogie model or sampling frequency.
• Figure 42. The internal energy calculated to break the wood post is less than the kinetic energy reduction of the bogie.
• Figure 43. These deformed configurations at 180 mm of deflection indicate that adjustments in the quality factors affect the deflection at which the grade 1 pine post breaks in two.
• Figure 44. Although the grade 1 pine post breaks earlier in time with Q_T = 0.40 and Q_C = 0.70, the calculated energy and velocity histories are similar.
• Figure 45. The default number of plasticity algorithm iterations is set to one because these static load-deflection curves are insensitive to the number of iterations.
• Figure 46. Erosion affects the fully integrated element curves, but not the under-integrated element curves, indicating that the fully integrated elements erode while still carrying load.
• Figure 47. Deformed configurations and fringes of damage calculated with fully integrated elements (eight points) are similar to those calculated with under-integrated elements (one point).
• Figure 48. Energy-deflection and bogie velocity-reduction histories are not strongly influenced by the type of element formulation (eight points or one point) modeled in the breakaway region.
• Figure 49. The breakaway region calculated with perpendicular erosion in these static bending simulations looks more realistic than that calculated without perpendicular erosion; however, perpendicular erosion is not recommended for practical use.
• Figure 50. Load-deflection curves calculated with perpendicular erosion are slightly more brittle than those calculated without perpendicular erosion.
• Figure 51. The breakaway region calculated with perpendicular erosion in this bogie impact simulation looks more realistic than that calculated without perpendicular erosion.
• Figure 52. Dynamic load-deflection and bogie velocity-reduction curves calculated with perpendicular erosion are nearly identical to those calculated without perpendicular erosion.
• Figure 53. Use of perpendicular erosion causes excessive erosion to be calculated in the impact region in this preliminary bogie impact calculation.
• Figure 54. These single-element simulations demonstrate post-peak hardening in compression with positive values of G_hard.
• Figure 55. Inclusion of post-peak hardening, both parallel and perpendicular to the grain, prevented this calculation from aborting at a large deflection.
• Figure 56. Performance envelopes from 1995 NDOR testing.
• Figure 57. Static post setup.
• Figure 58. Dynamic post setup.
• Figure 59. Typical static post test results.
• Figure 60. Typical dynamic post test results.
• Figure 61. Other dynamic post test results.
• Figure 62. Impact sequence of post simulation.
• Figure 63. Damage (stored as effective plastic strain in d3plot files).
• Figure 68. Contact penetrations caused locking of parts.
• Figure 69. Sequence of fast bogie simulations.
• Figure 70. Energy of post parts for fast bogie simulation.
• Figure 71. Section forces through post just below impact: Fast bogie simulation.
• Figure 72. Stress-strain behavior of single elements.
• Figure 73. Volumetric behavior of single elements.
• Figure 74. Static post models.
• Figure 75. Rounded edge on brace.
• Figure 76. Force deflection: Baseline versus test.
• Figure 77. Energy deflection: Baseline versus test.
• Figure 78. Force deflection: Baseline versus refined mesh.
• Figure 79. Energy deflection: Baseline versus refined mesh.
• Figure 80. Variations by mesh size: Deformed geometry.
• Figure 81. Force-deflection behavior as a function of grade, moisture content, and temperature.
• Figure 82. Energy-deflection behavior as a function of grade, moisture content, and temperature.
• Figure 83. Variation by grade: Deformed geometry.
• Figure 84. Variation by moisture content: Deformed geometry.
• Figure 85. Variation by temperature: Deformed geometry.
• Figure 86. Dynamic wood post test model.
• Figure 87. Post vaporization.
• Figure 88. Contact at sharp corner.
• Figure 89. Post bending.
• Figure 90. Contact penetrations.
• Figure 91. Improved contact.
• Figure 92. Element vaporization at bottom of post.
• Figure 93. Volume expansion of 75 percent in an element near the bottom of post.
• Figure 94. Vaporization with a time step of 0.0001 ms.
• Figure 95. Highly distorted elements sometimes do not erode.
• Figure 96. Damage of highly distorted element.
• Figure 97. Southern yellow pine, grade 1: Hourglass control type 1, qm = 0.1.
• Figure 98. Southern yellow pine, grade 1: Hourglass control type 3, qm = 0.1.
• Figure 99. Southern yellow pine, grade 1: Hourglass control type 4, qm = 0.005.
• Figure 100. Southern yellow pine grade, DS-65: Hourglass control type 4, qm = 0.005.
• Figure 101. Schematic demonstration of how a finite element treats fracture prior to erosion.
• Figure 102. Demonstration of void formation and crushing of wood specimens.
• Figure 103. User's grades 1 and 1D simulation compared to the performance envelopes, using G_{f II} = 50 G_{f ⊥}.
• Figure 104. Most of the grades 1 and 1D static post test measurements exhibit brittle behavior.
• Figure 105. Good correlation is achieved between the simulation and the performance envelopes by increasing the fracture energy above the default value (to G_{f II} = 250 G_{f ⊥}).
• Figure 106. DS-65 pine post modeled with simple pinned boundary conditions.
• Figure 107. Developer's static post simulations using hourglass stiffness type 4 with a reduced (0.03) coefficient.
• Figure 108. Finite element meshes used in demonstration problems.
• Figure 109. Stable behavior in the developer's single-flap calculation.
• Figure 110. Unstable behavior in the developer's double-flap calculation.
• Figure 111. Unstable behavior in the user's single-flap calculation.
• Figure 112. LS-DYNA MESSAG file which shows that the stable time step is exceeded.
• Figure 113. Stable behavior is achieved by reducing the time step to that needed for stable contact surface behavior (default grade 1 saturated pine properties without rate effects).
• Figure 114. LS-DYNA MESSAG file with diagnostics for the stable time step.
• Figure 115. Unstable behavior of an elastic element in the neoprene liner.
• Figure 116. The liner penetrates the post when the post is modeled with either elastic or wood material models.
• Figure 117. The post breaks earlier when modeled with post-peak hardening (default saturated grade 1 pine properties with G_hard = 0.5).
• Figure 118. The post breaks by 15 ms in this Douglas fir simulation (default saturated grade 1 properties without rate effects).
• Figure 119. Bogie impact at 29.5 m/s for a grade 1 wood post modeled without rate effects (grade 1 default saturated pine properties).
• Figure 120. Bogie impact at 29.5 m/s for a saturated grade 1 wood post modeled with rate effects (default pine properties).
• Figure 121. Fringes of damage for bogie impact at 29.5 m/s (default properties for saturated grade 1 pine with rate effects).
• Figure 122. Unstable behavior in the developer's rigid-sleeve calculation.
• Figure 123. Unstable behavior in the user's rigid-sleeve calculation.
• Figure 124. LS-DYNA diagnostics indicate that a contact surface is improperly positioned and requires movement of nodes prior to running the simulation.
• Figure 125. The liner does not penetrate the post in calculations conducted with LS-DYNA, version 960.

List of Tables

• Table 1. Average of static post test data by grade.
• Table 2. Summary of bogie impact tests on posts by grade.
• Table 3. Approximate tensile strength ratios versus strain rate for saturated pine.
• Table 4. Model Cfa: Uniaxial Compression in Parallel Direction
• Table 5. Model Cfe: Uniaxial Compression in Perpendicular Direction
• Table 6. Model Tfa: Uniaxial Tension in Parallel Direction
• Table 7. Model Tfe: Uniaxial Tension in Perpendicular Direction
• Table 8. Parameters based on wood grade.
• Table 9. Parameters based on moisture content.
• Table 10. Parameters based on temperature.
• Table 11. Developer's response to user's review of wood model.

Next