eurjap2004pres

Seismic Performance ofDissipative DevicesMartin WilliamsUniversity of Oxford : Seismic Performance of Dissipative Devices Martin Williams University of Oxford Japan-Europe Workshop on Seismic Risk Bristol, July 2004


Outline : Outline Introduction to knee bracing Optimisation of the knee element design: Full-scale experiments on knee elements Finite element modelling Seismic design and analysis of knee braced frames Conclusions and future work Acknowledgements: Tony Blakeborough, Denis Clément, Neil Woodward


Introduction to knee braced frames : Introduction to knee braced frames Seismic energy dissipated through yielding/hysteresis of knee elements


Knee bracing : Knee bracing Knee element requirements: Early yield Large energy dissipation – shear vs flexure Stable under large non-linear excursions – web buckling Easily replaceable – no damage to ends Pursued via testing and FE analysis Focus on standard section types Flexural hinge: Shear yield in web:


Knee element designs : Knee element designs Column sections provide high lateral stability Different stiffener patterns explored to prevent plastic web buckling Perforation of webs explored as a way of giving a designer greater flexibility over choice of shear yield load


Test set-up : Test set-up


Loading regimes : Loading regimes Slow cyclic: Real-time loading:


Under-stiffened element : Under-stiffened element Failure mode Hysteresis:


Well-stiffened section : Well-stiffened section Failure mode: Hysteresis


Perforated web : Perforated web Failure mode: Hysteresis:


Thermal monitoring system : Thermal monitoring system Plastic strain distributions during tests could be deduced from measurements of the knee element temperature Thermal imaging system: Typical images:


Thermal analysis results : Thermal analysis results Amplitude = 20 mm 30 mm Energy: Plastic strain: Von Mises stress:


Summary of experimental findings : Summary of experimental findings Full scale cyclic loading gives responses representative of a real earthquake Yielding in shear is optimal UC sections are are less prone to lateral instabilities To prevent buckling, web stiffeners are required at a spacing approximately equal to the section depth At a realistic design deflection the load on a knee element is approximately 1.7 times the yield load Perforating the web was unsuccessful Thermal imaging is an effective method for identifying the energy dissipation areas and tracking the spread of yielding


FE analysis of knee elements using ABAQUS : FE analysis of knee elements using ABAQUS Cyclic analysis with three different hardening laws: Cyclic + thermal analysis – comparison of temperature rise in one half-cycle with test:


Buckling analysis : Buckling analysis Over-predicted buckling load of unstiffened web by 20% Unable to model buckling of stiffened web


Summary of FE results : Summary of FE results An accurate hardening law is essential for realistic cyclic analysis Thermal analysis showed reasonable agreement with thermal imaging results It was not possible to build a model that agreed with all aspects of behaviour - shear forces, axial forces, moments and thermal dissipations Buckling analysis overestimated the critical load by 20% for an unstiffened knee element and was unable to predict the failure mode for knee elements with stiffeners


Design of a knee braced frame : Design of a knee braced frame 5-storey building designed to EC8, for earthquake with peak ground acceleration 0.35g


Design using pushover analysis : Design using pushover analysis Designed using EC8 pushover approach Also FEMA 356 approach, ATC 40 capacity spectrum method Key difference is idealisation of pushover curve:


Comparison with time-history analysis : Comparison with time-history analysis


Summary of results : Summary of results Pushover analysis shows that frames possess high ductility and post-yield stiffness Knee elements begin to yield at just 0.08g but remain stable up to 0.56g EC8 approach appears highly conservative for this type of structure, ATC40 approach unsafe


Conclusions : Conclusions Stable dissipative behaviour can be achieved using standard sections, appropriately reinforced Large increases in knee element load occur after initial yield Yielding and energy dissipation in experiments can be tracked using thermal imaging Accurate FE modelling of all aspects of knee element behaviour did not prove possible – web buckling was particularly problematic Design methods based on pushover analysis may be suitable for frames incorporating dissipative elements, but some further development of these approaches is desirable


Current/future work : Current/future work Testing of other dissipators, e.g. Jarret, Hyde devices Real-time substructure testing Further design and analysis studies using ten-storey frames, different dissipators