Evaluating the use of OpenSees for lifetime seismic performance assessment of steel frame structures
- 1. Evaluating the use of OpenSees for Lifetime Seismic Performance Assessment of Steel Frame Structures John Hickey Prof. Brian Broderick Terence Ryan Department of Civil, Structural and Environmental Engineering
- 2. Performance of CBFs Concentrically Braced Frames (CBFs) ‒ Diagonal bracing members resist lateral load though axial tension and compression ‒ Energy dissipated through buckling and tensile yielding of braces ‒ High lateral stiffness ‒ Unlikely to suffer collapse when designed to modern codes ‒ Can still suffer significant losses » Due to damage to structural & and particularly non- structural components Concentrically Braced Frame Steel Structures
- 3. Overall Aim Examine the influence of the behaviour factor (q) on the lifetime seismic performance of CBFs ‒ q used to account for nonlinear behaviour in design while avoiding the need for nonlinear analysis ‒ Influence of q on lifetime costs? Department of Civil, Structural and Environmental Engineering Research Goal
- 4. Performance Assessment PEER Equation - ‒ 4 Probability Distributions » Seismic Hazard Assessment – number of different Hazard levels » Structural Analysis – Peak drift & floor acceleration » Damage Assessment – Structural & Non-structural Components » Consequence Estimation – Probable repair costs, downtime, casualties etc. Department of Civil, Structural and Environmental Engineering US Geological Survey http://geohazards.usgs.gov/hazardtool/application.php OpenSees (McKenna, 1997) PACT (FEMA P-58) (ATC, 2012) Components of Lifetime Performance Assessment Procedure Seismic Hazard Ground Motion Records OpenSees – Time History Analysis Performance Metrics Engineering Demand Parameters
- 5. OpenSees in Performance Assessment Structural analysis for a given seismic hazard Calculate Engineering Demand Parameters (EDPs) ‒ Peak inter-storey drift ‒ Peak absolute floor acceleration Question ‒ How accurately do OpenSees models represent CBF response? Role of OpenSees in Lifetime Performance Assessment
- 6. BRACED ‒ Shake table test program examining CBF response (Broderick et al., 2015) ‒ Various structural configurations at 50%, 10% and 2% in 50 year intensities BRACED Experimental Test Program
- 7. Model Accuracy Results from to OpenSees models can be compared to BRACED results ‒ Allowing model accuracy to be evaluated Approach ‒ Using ‘conventional‘ modelling procedures, what is the level of uncertainty associated with OpenSees models compared to experimental results for EDPs of interest? How do conventionally modelled CBFs compare to experimental result
- 8. OpenSees Model Material ‒ Steel02 – Giuffré-Menegotto-Pinto Model Braces ‒ Uriz et al., 2008 » 2 nonlinearBeamcolumn elements » Initial camber 0.1% of brace length at midpoint » 3 Integration points per element » Fibre Section ‘Conventional’ Modelling Assumptions
- 9. OpenSees Model Gusset Plates ‒ Hsiao et al., 2012 » Out of plane nonlinear rotational springs Beams & Columns ‒ Nonlinear BeamColumn elements » 3 Integration points per element » Fibre Section Analysis ‒ Ground motion = recorded shake table motions ‒ 3% Rayleigh damping at 1st and 3rd mode ‘Conventional’ Modelling Assumptions
- 10. OpenSees Model Illustration of OpenSees model geometry
- 11. OpenSees Model Illustration of OpenSees model geometry superimposed on test frame
- 12. OpenSees vs Experimental EDPs Peak Drift ‒ Underestimated at all intensity levels » Typically model predicts about 60% of experimental value ‒ Attributable to: » Model underestimating flexibility in connections Comparison between numerical and experimental results
- 13. OpenSees vs Experimental EDPs Peak Acceleration ‒ Overestimated at 50% in 50 year intensity level » On average model predicts 135% of experimental value ‒ Good estimates at 10% and 2% in 50 intensity level » Peak acceleration limited by yield strength Comparison between numerical and experimental Results
- 14. OpenSees vs Experimental EDPs Summary ‒ OpenSees model underestimates peak elastic and inelastic drift (about 60%) ‒ OpenSees model overestimates peak elastic acceleration (about 135%) ‒ Predicts peak inelastic acceleration reasonably well (±10%) Question ‒ How do these inaccuracies impact on the performance assessment procedure? Comparison between numerical and experimental Results
- 15. Impact on Performance Assessment Influence of modelling inaccuracies on Performance Assessment Procedure Case Study Building ‒ Single storey CBF with BRACED frame as seismic resisting element » Performance assessment using PACT 1. Experimental EDPs 2. Numerical EDPs ‒ Allows us to see how different EDP values impact on performance metrics
- 16. Impact on Performance Assessment Influence of modelling inaccuracies on Performance Assessment Procedure Case Study Building ‒ Performance metrics with Numerical EDPs less than those with Experimental EDPs
- 17. Impact on Performance Assessment Influence of modelling inaccuracies on Performance Assessment Procedure Dealing with Modelling Uncertainty in PACT ‒ Factor βm ‒ Increases covariance of EDP distributions to account for uncertainty
- 18. Impact on Performance Assessment Influence of modelling inaccuracies on Performance Assessment Procedure Performance Metrics – Including Modelling Uncertainty ‒ Including factor to account for uncertainty for numerical EDPs ‒ Using PACT recommended value of βm ‒ Performance metrics from numerical and experimental EDPs are very similar
- 19. Impact on Performance Assessment Influence of modelling inaccuracies on Performance Assessment Procedure Summary ‒ Performance metrics from numerical and experimental EDPs similar when uncertainty accounted for along with numerical values ‒ Level of modelling uncertainty found agrees with that anticipated in PACT model ‒ Can proceed with performance assessment with some level of confidence
- 20. Impact of q on Lifetime Performance Frames Analysed Frames Analysed ‒ Perimeter CBFs ‒ 2 & 5 storey CBFs ‒ q = 1, 2, 3, 4, 5 ‒ 10 storey CBFs » q = 2, 3, 4 ‒ Case study site in Oakland, CA. ‒ Modelled in OpenSees as before ‒ Time history analysis performed using ground motions records matching the Conditional Spectrum at the site
- 21. Impact of q on Lifetime Performance Sample EDP Results – Peak Inter-storey Drift
- 22. Impact of q on Lifetime Performance Sample EDP Results – Peak Floor Acceleration
- 23. Impact of q on Lifetime Performance Expected Annual Losses Sample Results – Expected Annual Losses ‒ PACT used to calculate performance metrics using EDPs from OpenSees ‒ Losses in $/m2 ‒ Losses represented as financial losses due to repair and downtime ‒ Estimated losses increase with the behaviour factor
- 24. Impact of q on Lifetime Performance Expected Annual Losses – Percentage of Initial Costs Sample Results – Expected Financial Losses – Including Initial Costs ‒ Losses as percentage of initial costs ‒ Initial costs estimated as $250 per ft2 ‒ Initial costs increase as behaviour factor reduces - estimated by cost of extra weight of steel
- 25. Conclusion Summary • Investigated the level of uncertainty in OpenSees models using experimental results • Verified that the assumed level of modelling uncertainty in the performance assessment is in line with this • Examined the impact of the behaviour factor on lifetime seismic performance for case study CBFs • Shown expected losses increase with the behaviour factor Summary of work discussed