Authors: Hsin-Li Chan; Byung Rae Cho
Addresses: Department of Industrial Engineering, Clemson University, Clemson, SC 29634, USA ' Department of Industrial Engineering, Clemson University, Clemson, SC 29634, USA
Abstract: Chan and Cho (2013) studied the specifications-based robust design optimisation for a nominal-the-best type (N-type) quality characteristic under the normality assumption as an underlying distribution of the characteristic. In the real life engineering environment, however, various quality characteristics are of smaller-the-better (S-type) or larger-the-better (L-type) and their distributions are heavily skewed. As an extension of Chan and Cho (2013), this paper proposes specifications-based robust design (RD) models for S- and L-type quality characteristics, which are constrained by pre-specified upper and lower specification limits (USL and LSL), respectively. This paper demonstrates that the Weibull distribution is effective in modelling the S- and L-type characteristics, derives the truncated mean and variance from the Weibull parameters, and incorporates those truncated statistical moments into response surface-based RD models. Finally, numerical examples illustrate how the optimal factor settings are obtained from the proposed models.
Keywords: quality characteristics; smaller-the-better; S-type; larger-the-better; L-type; truncated Weibull distribution; response surface methodology; RSM; optimisation; robust design; modelling.
International Journal of Experimental Design and Process Optimisation, 2013 Vol.3 No.4, pp.364 - 383
Available online: 05 Mar 2014 *Full-text access for editors Access for subscribers Purchase this article Comment on this article