Robustness of steel frames further to a column loss: development of analytical methods for practitioners

Abstract

In order to fill the lack of practical guidelines in the European recommendations regarding the structures robustness, research are conducted in Liège University on this topic. This master thesis focuses on the "alternative load path method" which analyses a structure further to the loss of a structural element due to an exceptional event. More specifically, the topic of this master thesis is the analytical model developed in Liège University designed to characterize the response of a steel frame further to the loss of one of its columns. A particular plastic mechanism is assumed to form in the "directly affected part" above the lost element while the rest of the structure is assumed to behave elastically.

The last version of the model and its implementation, after years of incremental modifications, have not yet been properly stated and verified. Therefore, the first part consists to detail and verify extensively the analytical model and its implementation developed so far. A particular emphasis was brought to the field of applications and the assumptions made during each development step.

Afterwards, a first improvement of the initial model is implemented to enhance its accuracy and to widen its field of application. This improvement includes the combination of Matlab and BeamZ. After clearly stated the methodology and its implementation, new developments are validated through the comparison of different scenarios with Finelg simulations. Besides, during the validation of the new implementation, the plastic mechanism assumed in the initial model have been adjusted due to inconsistencies with results obtained in Finelg.

Finally, a second implemented improvement aims to introduce the yielding of the "indirectly affected part" of the structure, that was not taken into account in the initial analytical model. Once the methodology and its implementation are clearly detailed, the implementation is verified and validated through comparison with Finelg simulations.