4.3 State-of-the-art in Optimization

For systems where partially analytical descriptions are available, the determination of local optima often easy to obtain. However, the problem is still hard and may require a huge computational effort and the computational costs to find a global extrema is still NP-hard. Yet, for industrially relevant applications the complexity increases and the number of suitable optimizers, which can be applied to these types of problems decreases rapidly. Because the optimization setup requires knowledge of the optimization process itself as well as an expert knowledge of the methods used in the simulation tools in the optimization process.

Many discussed optimization strategies have prerequisites to guarantee an effective operation of the optimization algorithm. However, most of these requirements cannot be fulfilled from optimization problems of industrial interest. Therefore, the user has to decide for some approximations to optimize a certain aspect of this particular problem. Nevertheless, this method often yields reasonable intermediate results which can be used for further improvement using different optimization strategies. Hence, the whole optimization task often becomes a stepwise optimization strategy with different applied optimization techniques.