Superimposing direct search methods for parameter optimization onto dynamic simulation models

An integrated modular software package has been developed by the Programme Group of Systems Analysis and Technological Development (STE) of the Nuclear Research Centre at Jülich (KFA) to provide automatic optimization of a set of user defined decision variables. This optimization module containing different procedures for direct search algorithms can be added to our FORTRAN based Data-Model-Interface (DMI) for dynamic simulation (1). The latter is formulated independently from an actual simulation model as well as from the actual parameters, the objective, and the constraints which are chosen. Once initiated it controls all user specified input values and the linkage between the optimization algorithm and the simulation model. Taking the KFA-energy model (2) as an example one application for finding a specific energy policy is demonstrated.