|Institution:||KTH Royal Institute of Technology|
|Keywords:||MATHEMATICS; Applied mathematics; Optimization, systems theory; MATEMATIK; Tillämpad matematik; Optimeringslära, systemteori; Physics, Chemistry, Mathematics; fysik/kemi/matematik; Master of Science - Mathematics; Teknologie masterexamen - Matematik|
|Full text PDF:||http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-102764|
An electromagnetic stirrer (EMS) is a device widely used in metal melting processes. An EMS is normally placed under the bottom of a metal melting furnace. The presence of an EMS establishes a movement of the melts inside the furnace through the interaction between the electromagnetic force generated by the stirrer and the metallic melts. The design of an EMS requires the use of finite elements method (FEM)-based approach. As a common knowledge, when FEM is involved, the design analysis of a stirrer is a time-demanding process. Optimal design of an EM Sposes additional challenge concerning total optimization time, since a large number of design analyses is necessary. The state-of-the-art optimal design of an EMS involving about 500 design analyses requires typically 72 hours total optimization time. This obviously convinces for a more efficient approach for optimal design of an EMS. In this work, meta models of the system performance of an EMS are developed based on FE simulation results involving 40 hours of computation. Two optimization problem formulations are considered. The design optimization is able to be conducted based on the developed meta models with drastically reduced total optimization time (from 72 hours to two minutes),with acceptable accuracy of the optimization results. It is evident that the meta model-based design optimization will play an important role of future EMS development at ABB.