Optimization of pressure vessel design using pyopt

• Main Authors: Woldemichael, D.E., Woldeyohannes, A.D.
• Format: Article
• Published: Asian Research Publishing Network 2016
• Online Access: https://www.scopus.com/inward/record.uri?eid=2-s2.0-85009183740&partnerID=40&md5=cfa543737626069c0601b08aaf96a189; http://eprints.utp.edu.my/25737/
• Institution: Universiti Teknologi Petronas

pyOpt is an open source python based object oriented framework for nonlinear constrained optimization problems. In this study, we used pyOpt to solve pressure vessel design problem. Among the available optimizers in pyOpt, SLSQP (Sequential least squares programming), COBYLA (Constrained Optimization by Linear Approximation), ALPSO (Augmented Lagrangian Particle Swarm Optimizer), NSGAII (Non Sorting Genetic Algorithm II), MIDACO (Mixed Integer Distributed Ant Colony Optimization), and ALGENCAN (Augmented Lagrangian with GENCAN) were used. The effect of initial design variables on convergence was investigated for six different regions. The initial design variables for MIDACO and SLSQP should be within the design variable bound while COBYLA and ALPSO provide good result when the initial point is greater than the upper bound. On the other hand, NSGAII and ALGENCAN converge to the optimum value regardless of the initial value. The optimum results from all optimizers were compared with published literatures. Except for ALPSO with mixed discrete variables, the results are in good agreement with maximum percentage error of less than 5. Therefore, pyOpt can be considered as an alternative option to solve engineering design optimization problems. © 2006-2016 Asian Research Publishing Network (ARPN). All rights reserved.