Title: Development of multivariate process monitoring strategy for a typical process industry

Authors: Anupam Das; Swarnambuj Suman; Amresh Kumar Sinha

Addresses: Department of Mechanical Engineering, National Institute of Technology Patna, Patna, Bihar 800005, India ' Department of Mechanical Engineering, National Institute of Technology Patna, Patna, Bihar 800005, India ' Department of Mechanical Engineering, Sikkim Manipal Institute of Technology, Gangtok, India

Abstract: The study demonstrated the application of partial least square regression technique in the development of a monitoring strategy for a process industry. The process industry under consideration is an integrated steel plant engaged in the production of steel billets. The steel making shop which has been the focus of the study is a complex process replete with numerous and inter-related process, feedstock and quality characteristics. The main challenge addressed in this paper is the development of a monitoring strategy for the concerned steel making shop taking into account all the characteristics (process, feedstock and quality) simultaneously. The strategy thus devised seems to bode well, as it was correctly able to ascertain the status (in state of statistical control or out-of-control) of the process. Further capability studies via the employment of a multivariate process capability index were carried out to determine the efficacy of the process in producing end products. The capability study highlighted the fact that the process needs readjustment as a substantial amount of end products produced were out of specifications.

Keywords: partial least square regression; PLSR; process industry; Hotelling T² chart; capability index; multivariate statistical process control; MSPC; steel making shop.

DOI: 10.1504/IJPQM.2017.085844

International Journal of Productivity and Quality Management, 2017 Vol.22 No.1, pp.1 - 21

Received: 16 Feb 2016
Accepted: 17 May 2016
Published online: 16 Aug 2017 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article