In this paper, a reliability allocation process is performed for parallel-serial systems, which components' uptime are assumed to follow that Weibull law. Generally, the uptime of components of a mechanical system follows Weibull law. The problem is formulated as two optimisations problems in which the decision variables are the parameters of the components up time law. The first problem consists of maximising the system's reliability under cost constraints, and the second one consists of minimising the cost of the system under reliability constraints. The optimisation problems are solved using Genetic Algorithms based on the real number and binary encoding on the one hand, and Simulated Annealing based on the previous encoding to create neighbour of feasible solutions on the other hand. A numerical example is given to compare the results obtained by each algorithm.