Optimal design of feeding system in steel casting by constrained optimization algorithms based on InteCAST

Abstract

The traditional foundry industry has developed rapidly in recently years due to advancements in computer technology. Modifying and designing the feeding system has become more convenient with the help of the casting software, InteCAST. A common method of designing a feeding system is to first design the initial systems, run simulations with casting software, analyze the feedback, and then redesign. In this work, genetic, fruit fly, and interior point optimizer (IPOPT) algorithms were introduced to guide the optimal riser design for the feeding system. The results calculated by the three optimal algorithms indicate that the riser volume has a weak relationship with the modulus constraint; while it has a close relationship with the volume constraint. Based on the convergence rate, the fruit fly algorithm was obviously faster than the genetic algorithm. The optimized riser was also applied during casting, and was simulated using InteCAST. The numerical simulation results reveal that with the same riser volume, the riser optimized by the genetic and fruit fly algorithms has a similar improvement on casting shrinkage. The IPOPT algorithm has the advantage of causing the smallest shrinkage porosities, compared to those of the genetic and fruit fly algorithms, which were almost the same.