Prestressed concrete slab decks are a normally typology used in Spain to resolve structures overpasses. Optimization is of great interest to get more affordable designs, enabling better utilization of the resources they need. However, such structures optimization has not been the subject of attention from researchers in structural optimization. Moreover, few contributions in this area have suffered from a highly theoretical approximation, which has hindered its implementation in engineering design. 
The aim of this work has been to apply optimization techniques to this structural typology. Metaheuristic techniques have been used, allowing the problem to be raised in a more complex way, which has been used for a complete definition of the deck and all its components, while allowed to impose any checks that codes require for this type structures. 
After an exhaustive literature review, which revealed the state of the art in optimization of concrete structures, characteristics of the problem have been defined. It’s been distinguished voided slabs from solid slabs, as it has proved impossible to consider one a particular case of the other. Then an informatics program has been implemented, that includes the following features: random generation of a deck, automatic check of a deck, cost evaluation from measurements of all its components, and three heuristic optimization algorithms. 
The optimization algorithms implemented are based on three metaheuristics, belonging to the so-called local search algorithms: simulated annealing (SA), threshold acceptance (TA), and old bachelor algorithm (OBA), which have proved effective for optimization of other structural concrete types. Calibration of the algorithms has been carried out by mean of tests with different parameters. Comparison of the results led to discard the OBA algorithm to show a lower efficacy for the tested parameterizations. SA and TA algorithms, by contrast, show very similar results, so statistical inference tests have been made involving different test scenarios. The results have not been able to determine the most effective heuristic. Moreover, applying the Extreme Value Theory, it has been estimated that the best results achieved with both algorithms differ by less than 0.15% over the cost of the global optimum of the problem. 
Finally, we have studied the number of restarts necessary to achieve quality solutions with a sufficiently high probability, using a stabilization of averages and standard deviations criterion. 
This has developed an optimization algorithm that combines thirty-five reboots with SA and other thirty-five reboots with TA. This has been applied to four types of board, representative of those used in overpasses frequently in practice. The results show how the optimal characteristics of decks are: voided slabs are always cheaper than solids, even for the smaller spans considered (17 m); optimized decks saving are up to 13% in the estimated average cost of real structures; and finally, optimized decks show low concrete ratios, moderate slender and active steel ratios, and relatively high passive steel ratios.