Since the 1960s engineerings and scientists have developednumerical models to stimulate complex physical phenomena, such as the flow of a fluid in or around an object (such as an airplane ora human artery), as well as biological or chemical phenomena (suchas distillation). The sophistication and robustness of these models, combined with major advances in large scale optimization, will make it possible optimize many complex systems in the near future. The talk will focus on the types of models that seem well suited for optimization, and on the obstacles that must be overcome.