In April 2018, a graduation project in the field of architecture was concluded. The project entailed the design of a temporary community centre in the earthquake-damaged historic town centre of L’Aquila, Italy. The underlying social concept was focused on participation of the local population in the construction of the community centre. The design includes a grid-like timber shell structure, built-up out of a repetition of only two main structural and light-weight elements: the members and the joints.
The load-bearing structure of this architectural design formed the basis for this graduation thesis, which is aimed at its optimisation. The optimisation is carried out focused on many objectives, e.g. its weight, its environmental impact, its structural validity. The goal of this thesis can be summarised by the following research question. Given the architectural design, how can the main load-bearing structure be optimised for seismic contexts using parametric modelling, considering the following objectives: maximal demountability and structural simplicity, minimal material use, structural weight and environmental impact?
An optimisation method has been chosen combining automated and manual processes, as this gives a high sense of control to the designer. Using computational and parametric modelling techniques (Rhino, Grasshopper and Karamba), a database comprising 10 836 different structural configurations has been generated with different geometries and materials. The database contains the generated outcomes to structural and environmental analyses for each of the structural configurations. The database entries have been ranked manually based on structural, architectural and social demands, resulting in a top 28 models. Having defined a ranked set of structural configurations, manual checks and iterations can take place. The first step is checking the global and local stability using proposed joint designs as input. The next step is checking the safety of the structure during seismic events using a response spectrum analysis. As the Top-1 design passed all analyses, no iterations were deemed necessary.
In conclusion, with regard to the design, further structural optimisation could take place by reconsidering the shape of the structure. However, the process of automated database generation followed by manual exclusion, ranking and iterations give a great sense of control compared to automated optimisation alternatives. This optimisation process has proven to be very effective in finding a design that balances all objectives well.