Preface

This book provides a timely reference text for academics, undergraduate and graduate students, and practitioners alike in the area of process monitoring and safety, as well as product quality assurance using multivariate statistics. The rapid evolution of this research area over the past 20 years is mainly driven by significant advances in computer horsepower and the ever growing demand from industry to effectively and efficiently monitor production processes. As an example, Nimmo (1995) outlined that the US-based petrochemical industry could save an estimated $10 bn annually if abnormal conditions could be detected, diagnosed and appropriately dealt with. Moreover, the demand from the oil and gas industry, other chemical engineering and general manufacturing industries is also a result of ever tighter government legislation on emissions and increased safety standards of their products.

The wide range of applications of multivariate statistics for process monitoring, safety and product quality is of considerable interest to the readership in chemical, mechanical, manufacturing, electrical and electronic, industrial and other related engineering and science disciplines. This research text serves as a reference for introductory and advanced courses on process safety, process monitoring and product quality assurance, total quality management of complex technical systems and is a supplementary text for courses on applied statistics and process systems engineering. As a textbook and reference, this book pays particular attention to a balanced presentation between the required theory and the industrial exploitation of statistical-based process monitoring, safety and quality assurance.

To cater for the different audiences with their partially conflicting demands, the scope of the book is twofold. The main thrust lies on outlining the relevant and important fundamental concept of multivariate statistical process control or, in short, MSPC and to demonstrate the working of this technology using recorded data from complex process systems. This addresses the needs for the more how-does-it-work and what-does-it-do oriented readership of this book, which includes undergraduate students, industrial practitioners and industrially oriented researchers. The second pillar is the theoretical analysis of the underlying MSPC component technology, which is important for the more research-oriented audience including graduate students and academicians.

The twofold coverage of the material results from the research background of both authors, which is centered on academic research in process monitoring, safety, product quality assurance and general process systems engineering, and their participation in numerous industrial R&D projects, including consultancy concerning the application of MSPC and the development of commercial software packages. As this book carefully outlines and discusses, the main advantage of the MSPC technology is its simplicity and reliance on recorded data and some a priori knowledge regarding the operation of the process system. On the other hand, this simplicity comes at the expense of stringent assumptions, including that the process is stationary and time-invariant, and that the process variables follow a Gaussian distribution.

With this in mind and based on academic and industrial R&D experience, the authors are convinced that MSPC technology has the potential to play an important role in commercial applications of process monitoring, safety and product quality assurance. This view is also supported by the arrival of software that entered the value-added market for commercially available packages, which includes AspenMultivariate™, Wonderware, SIMCA-P (to name but a few), consultancy companies, such as Perceptive Engineering Ltd., Eigenvector Research Inc. and statistical data analysis software, e.g. STATISTICA, SAS®.

The first thrust of MSPC work for monitoring complex process systems emerged in the late 1980 and the early 1990s and lays out a statistically sound concept under these basic assumptions. It is important to note, however, that if a process ‘unfortunately forgets’ to meet the above assumptions, the corresponding monitoring charts may produce false alarms or the sensitivity in detecting minor upsets is compromised. From the end of the 1990s until now, research work that has enhanced the core MSPC methodology has removed some of these stringent assumptions. This, in turn, allows the enhanced MSPC technology to be applicable in a more practically relevant environment.

Besides the required theoretical foundation of the MSPC methodology, this book also includes a detailed discussion of these advances, including (i) the monitoring of time-variant process systems, where the mean and variance of the recorded variables, and the relationship between and among these sets, change over time, (ii) the development and application of more practically relevant data structures for the underlying MSPC monitoring models and (iii) the development of a different construction of monitoring statistics and charts which significantly improves their sensitivity in detecting incipient fault conditions.

This book ideally supplements the good number of research texts available on multivariate statistics, statistical process control, process safety and product quality assurance. In particular, the research text brings together the theory of MSPC with industrial applications to demonstrate its usefulness. In particular, the mix of theory and practice in this area is rare; (exceptions include Mason and Young (2001)). Moreover, good and solid reference that address the theory as well as the application of component technology are rarely written for the industrial practitioner whose experience is pivotal in any process monitoring, safety and product quality assurance application.

To comprehend the content of this book, the readership is expected to possess basic knowledge of calculus including differentiation, integration and matrix computation. For the application study, a basic understanding of principles in physics and chemistry is helpful in following the analysis of the application studies and particularly the diagnosis of the recorded fault conditions. To enhance the understanding of the presented material and to improve the learning experience, each chapter presenting theoretical material, except the last two, includes a tutorial session which contains questions and homework-style projects. The questions assist with the familiarization of the covered material and the projects help the reader to understand the underlying principles through experimenting and discovering the facts and findings presented in this book either through self-study reports or team-based project reports. The calculations can be carried out using standard computational software, for example Matlab®.