A genetic algorithm-based framework for the optimal seismic retrofitting of reinforced concrete buildings by steel-jacketing

A large amount of buildings and infrastructures in the world are reinforced concrete (RC) frame structures designed prior to the entry into force of seismic guidelines and seismic detailing rules. Seismic risk associated with these structures is significant due to their low lateral load-carrying capacity and insufficient ductility. In particular, RC columns play a critical role to the seismic performance, being the location of most of the structural deficiencies (poor concrete, inadequate transverse reinforcement, lack of seismic details). One of the extensively used retrofitting technique for columns is the steel-jacketing. It consists of the installation of a cage made of steel angles and battens providing additional confinement and transverse reinforcement to the RC elements,and compensating their ductility lack. The main issues that structural engineers face in the design of this kind of interventions regard the determination of the position and the amount of the retrofitting to exploit the maximum effect, reducing costs and invasiveness of the intervention. Currently, the design of these retrofitting interventions is mainly based on engineer’s intuition and experience and, hence, this could lead to an over-estimated design, associated with an increase of economical and downtime costs. The master’s degree thesis addressed the use of genetic algorithms (GA), proposing a rational method to optimise the seismic retrofitting of the existing RC structures with steel-jacketing. The optimisation is performed both for the position of the retrofitting system(topological optimisation) and for the amount of steel used for the jacketing, by varying battens interaxis. The research space consists of all the combination of retrofitted columns with all the different battens spacings. The metaheuristic procedure allows obtaining the optimal solution without the need of evaluating all the possible solutions that could involve huge computational effort. The main GA operators (selection, crossover, and mutation) concur to explore the research space roughly and evolve the suitable results toward better solutions. The GA analysis aims to select the cheapest retrofitting solution among the feasible ones. The cost of each candidate solution is evaluated by the objective function, which takes into account material and workmanship related costs. The feasibility of each solution is verified by the results of static pushover analyses in the framework of the N2 method from the results carried out a 3D fibre-section model, developed in the OpenSees software platform. The proposed approach has been tested for different case study structures subject to different structural deficiencies (plan and height irregularities, local shear failures, influence of masonry infills), highly representative of the class of RC existing structures built in the middle of 1900. From the obtained results, it can be concluded that the proposed optimisation framework can effectively reduce RC building retrofitting and downtime costs controlling safety levels. Cost minimisation is not directly connected to a reduction of safety levels, but on the contrary, the optimisation allows discarding ineffective retrofitting solutions for which higher costs are associated with lower safety.This method could be an efficient support to the designer for choosing the cost-effective configuration of the intervention who eventually will have the final decision based on his engineering judgment.