Traditional method for design involves considerable human interaction and therefore is time consuming. It is also heavily dependent on the intellectual level and understanding of the user. With the availability of high performance computing systems it is possible to automate many of these processes to get automatic result. This not only reduces the time for experimentation of newer designs but also leads to new concepts as these are algorithm based.
ZN has developed modules for multi-disciplinary optimization of various systems using genetic algorithm approach. The software requires constraints in the design to be defined as a mathematical function (cost function) to be given as input. The software then generates and tests newer designs for the performance using ZNs proprietary analysis software and produces offsprings that are healthier. The simulations run on HPC system and produce offsprings (new designs) ranging from few hundreds to ten thousand. Designs in the acceptable band are then given to the user for further analysis.
In one such instance ZN designed a winglet with fixed wing geometry. The design was to be validated for constraints of weight, aerodynamics and structures. First newer designs were tried for aerodynamics and a few were selected to be further subjected to structural analysis. The resulting design estimated a 14% reduction in drag with reasonable weight penalty. In some cases simpler methods of optimization are also used based on domain expertise and repeated analysis.
See a related case study: Aero-Structural Optimization of Winglet for Regional Transport Aircraft