Making Use of Process Tomography Data for Multivariate Statistical Process Control

Bundit Boonkhaoa, Rui F. Lia, Lande Liua, Xue Z. Wanga*, Richard J. Tweedieb, Ken Primrosec, Jason Corbettc and Fraser K. McNeil-Watsonc

aInstitute of Particle Science and Engineering, School of Process, Environmental and Material Engineering, University of Leeds, Leeds LS2 9JT, UK

bIndustrial Tomography System Plc, Speakers House, 39 Deansgate, Manchester M3 2BA, UK cMalvern Instruments Ltd, Enigma Business Park, Grovewood Road, Malvern, Worcestershire WR14 1XZ, UK

ABSTRACT

This paper describes a novel strategy for making effective use of on-line process tomography measurements for process monitoring. The electrical resistance tomography (ERT) sensing system equipped with sixteen electrodes provides 104 conductivity measurements every 25 milliseconds. The data has traditionally been used for construction of images for display purpose. In this study, ERT data was used for multivariate statistical process control (MSPC). Data at pre-defined normal operational conditions was processed using principal component analysis. The compressed data was used to derive two statistics; Hotelling’s T2 and squared prediction error (SPE). The Hotelling’s T2 and SPE charts predict the probability that the process being monitored has undergone statistically significant changes from previous state or the so-called normal operational state, in terms of mixing quality. The methodology is illustrated by reference to a case study of a sunflower oil/water emulsion process.

Keywords electrical resistance tomography (ERT); multivariate statistical process control (MSPC); principal component analysis (PCA); emulsion.