Optimisation Framework AutoOpti
At the Institute of Propulsion Technology, we are developing an integrated numerical design system for the aeromechanical design of turbomachinery components.
The AutoOpti software package for automated multidisciplinary optimisation of turbomachinery components is used for everything from two-dimensional profile aerodynamics to three-dimensional multidisciplinary optimisation of multi-stage configurations. The core of the optimisation algorithms for aerodynamics, mechanics, aeroelastics and acoustics is based on the evolution strategy, which allows the simultaneous improvement of several objective functions and the fulfilment of many constraints. Robustness and the ability to overcome local minima are maintained. The moderate convergence speed is raised to a level necessary for industrial and research practice by hardware-related acceleration methods and in particular by response surfaces (metamodels). The response surfaces (neural networks and kriging) can now also take into account gradient information of the target functionals, a prerequisite for the effective use of adjointTRACE in the process chain.
Neural networks and/or kriging models approximate the relationships between target functions and free design parameters based on the results available in the database. A preliminary optimisation is then performed on these response surfaces and only the most promising data sets are transferred to the numerically complex process chain.
For usability reasons, the surrogate models have been integrated into the root process of the core optimiser. However, a uniform optimality criterion is now required for all optimisations on the surrogate models, which has been achieved through the development of the "Expected Volume Gain" (EVG). A new member is sought which, together with the members still under evaluation, maximises the expected volume gain in the target function space at the current Pareto front. This criterion can be used to identify the member with the maximum expected optimisation progress among data sets with the same Pareto rank. Another major advantage of the TOE criterion is the use of additional information. For example, the uncertainties for the objective function values of a data set provided by the response surfaces, as well as information about which data sets are currently being evaluated, are included in the TOE criterion.
In all adjustments, care is taken to ensure that the online controllability of AutoOpti, such as the modification of objective functions, constraints and limits of free parameters during an ongoing optimisation, is maintained and further improved. AutoOpti has now reached a high level of maturity and is being used successfully in several industrial companies.