| Přispěvatelé: |
Frangi, Andrea, Steiger, René, Kaufmann, Walter, Winter, Stefan, Köhler, Jochen |
| Popis: |
When designing any building, is it easy to think that damage scenarios such as column loss are too rare to explicitly design for. This, along with several other biases of our brains, make us unable to ensure structural safety by explicitly designing for accidental scenarios, thus making structural robustness – the ability to withstand an unforeseen damage without being disproportionately affected – the desired structural property. Nature has many examples of redundancy, proofreading, and error correction mechanisms from which we can get inspiration and realise that robustness is not a feature to add to a structure, but rather an intrinsic property that comes with a careful design process. In the last 50 years, structural robustness has received the most attention for concrete and steel buildings, whose generally ductile behaviour usually implies the existence of alternative load paths (i.e., redundancy). This is not necessarily the case with tall timber buildings, a structural typology of unprecedented size where ductility is not implicit and where the use of the naturally complex wood poses an increased risk of unknown global behaviour when damaged. A holistic framework to design robust buildings is the main proposal in this thesis. The framework has a qualitative part whereby a conceptual design spanning multiple levels of the building scale (from material to whole building) ensures safety against both localised and systematic damage. Quantitatively, the calculation of a robustness index – i.e., the (dis)proportionality between initial and subsequent damage risks, is insufficient: it is more complete to compare the indices between a starting design and various improvements. The robustness index calculation must be averaged over many damage scenarios. The robustness quantification according to the proposed framework is demonstrated via the design and analysis of a 15-storey tall timber building. A dynamic, nonlinear Finite Element Model is set up and run multiple times in a probabilistic input space to investigate the response of the building to three accidental damage scenarios. The results are used to train a Random Forest classifier in order to calculate the expected range of collapse – if any – for all scenarios, and thus the average robustness index. This is repeated for two improved design options which have a truss “strong floor”, and two other options which have increased connection strength and/or ductility. The “strong floor” improvements perform considerably better than the connection-only improvements, primarily because “just adding ductility” does not guarantee that the alternative load paths find their way to the ground in all damage scenarios. A sound conceptual design is of utmost importance. A more detailed design of the floor slabs and the addition of appropriate structural damping in the numerical model are the next steps to improve the accuracy of the model. The full robustness quantification procedure is lengthy and complicated and only makes sense in high importance buildings. A qualitative-only (e.g., for low importance buildings), or hybrid qualitative- quantitative procedure (e.g., for medium importance buildings) are valid possibilities for the practicing engineer. |
