Plenary Talk with Prof. Mehrdad Zangeneh
Prof. Mehrdad Zangeneh will be giving a Plenary Talk - Multi-Objective, Multi-Disciplinary Inverse Design Based Automatic Optimization of Turbomachinery - on Thursday, November 26th at 9 am GMT. Turbomachinery manufacturers face many challenges such as legislative pressures to improve efficiency (e.g ECOdesign directive in EU) as well as competitive pressure to reduce cost and developments times. Many of these challenges require innovative design solutions that can solve multi-point/multi-objective and multi-disciplinary problems.
The 3D inverse design method computes the turbomachinery blade geometry for a specified distribution of blade loading and pressure field. The method enables designers to optimize turbomachinery vanes and blades by exploring a large design space without the trial and error of traditional design methodologies. There are also computational advantages in using inverse design as an optimization strategy. In inverse design, the optimization is parametrized through the blade loading and not the blade geometry, which can significantly reduce the number of design parameters to cover the same design space. This feature improves the speed and accuracy of automatic optimization. In particular, by using the inverse design approach it is possible to achieve accurate surrogate model based optimization. This approach can then be used to solve difficult multi-point, multi-objective and multi-disciplinary problems under industrial time scales.
ADT's Workshop on Inverse Design Based Multi-Point Optimization
On the same day, Prof. Mehrdad Zangeneh, Dr. Jiangnan Zhang and Dr. H. Watanabe of Ebara will also run a workshop at 2:25pm GMT on Inverse Design Based Multi-Point Optimization.
In this live demonstration, the set up of a multi-point optimization of a mixed flow pump impeller by using a 3D inverse design method, which generates the blade shape for a specified distribution of blade loading, will be presented. The workshop starts with the key targets for the optimization and briefly presents how the inverse design method can be used to parametrize the blade geometry by using blade loading distribution. One key aspect for successful optimization is the way the range of design parameters are selected. The way the inverse design method facilitates the choice of range of design parameters is presented. In total 5 design parameters are used to parametrize the blade shape and the Design of Experiment method Optimal Latin Hypercube is used to create design matrix consisting of 22 geometries. These geometries are then run at two operating points of 100% design of design flow and 40% of design flow.
Three cases could not be converged in CFD and hence a surrogate model was formed by using the 19 converged cases. The surrogate model used was Kriging. The process of using the surrogate model to obtain a trade off between the two efficiencies or Pmin ( as a measure of cavitation) will be shown live during the workshop. The resulting blade shapes and flow field as predicted by the inverse design method will be shown. Also shown will be the results of validation of the surrogate model by CFD. Audience participation and question and answer opportunity to be provide at each stage of the set up of the optimization.
Interested in Multi-point Optimization of Pumps using 3D Inverse Design Method?
You might be interested in a technical paper we released about the use of 3D Inverse Design method to perform multi-point design optimization on pumps. It's titled: How a 3D Inverse Design Method Supports Leading Pump Manufacturers to Meet Emissions Legislation.
|