1
Introduction

Sally Brailsford,1 Leonid Churilov2 and Brian Dangerfield3

1Southampton Business School, University of Southampton, UK

2Florey Institute of Neuroscience and Mental Health, Melbourne, Australia; RMIT University, Melbourne, Victoria, Australia

3Salford Business School, University of Salford, UK

1.1 How this Book Came About

To begin at the end … the final chapter in this book, by Michael Pidd, contains both a backwards and a forwards look at system dynamics and discrete-event simulation. Historically, both modelling approaches originate from around the same time, the late 1950s and early 1960s. However, over the intervening decades they developed into separate scientific and practitioner communities, each with its own learned societies, academic journals and conferences. Discrete-event simulation (DES) has been a core subject on MSc programmes in operational research (OR) or management science (MS) from the 1970s onwards, and is a standard technique in the ‘OR/MS toolkit’. For many operational researchers, ‘simulation’ is synonymous with DES, and indeed the aim of the UK OR Society's own Journal of Simulation (quoting directly from the journal's web site) is to provide ‘a single source of accessible research and practice in the fast developing field of discrete-event simulation’ (http://www.palgrave-journals.com/jos/index.html). However, this is not true of system dynamics (SD): the SD community was, and still is to some extent, distinct from the OR community. While there is obviously some overlap in membership, there are many members of the international SD Society who are not members of their national OR Society (and vice versa). SD was certainly not taught on the MSc in OR at the University of Southampton in the 1980s and 1990s.

In 2000, the Simulation Special Interest Group of the UK OR Society held a joint meeting with the UK Chapter of the SD Society, entitled ‘Never the Twain Shall Meet’. At this meeting David Lane presented a paper (Lane, 2000) in which he discussed the differences between SD and DES and posed the question about whether they were ‘chalk and cheese’ or were actually two sides of the same coin. This meeting led to the foundation of a new OR Society Special Interest Group called ‘SD+’ whose aim was to bring the SD and OR communities together. The ‘+’ in SD+ was broader than just DES: it included many other OR techniques and approaches with which SD could interface. The ‘Never the Twain’ meeting also led to a number of academic papers exploring the similarities and differences between DES and SD, including the well-known study by Robinson and Morecroft which forms Chapter 9 of this book. Indirectly, it led to this book itself!

In some application areas the use of SD has expanded rapidly since 2000, and healthcare – the specialist application field of all three editors of this book – is one such area. However, despite initiatives like SD+, it is still true to say even today that SD is less well known in the mainstream OR community than DES. The number of DES papers at the annual Winter Simulation Conference, the major US conference on simulation, always greatly exceeds the number of SD papers. The main aim of this book is to begin to address this disparity. The book provides an integrated overview of SD and DES, a detailed comparison of the two approaches from a variety of perspectives, and a practical guide to how both may be used, either separately or together.

1.2 The Editors

As editors, we should declare our own personal interests. Having started out in the early 1990s as a dedicated user of DES, Sally Brailsford became interested in SD as a result of the ‘Never the Twain’ meeting, and joined David Lane in co-founding the SD+ group. Subsequently she became a zealous convert, using SD for several modelling studies. Like many other researchers, she was fascinated by the relationship between DES and SD, and in particular in the domain of healthcare (Brailsford and Hilton, 2001), but first used SD in practice in a project to model demand for emergency healthcare in Nottingham, England (Brailsford et al., 2004) which is described in detail in Chapter 6.

Leonid Churilov has a firm belief that real management problems do not come cleanly separated by disciplinary lines and, as a result, can rarely be comprehensively addressed using a single given modelling method. This basic premise is the source of continuous motivation for his keen interest in combining and contrasting different OR/MS techniques for management decision support. His research, in particular, included combining DES and clustering/classification techniques for decision support in hospital emergency departments, the use of both DES and SD for process systems modelling, and the original value-focused process engineering methodology that integrates the approaches from both the decision sciences and business process modelling domains. His work on philosophical underpinnings of both DES and SD worldviews from the critical realist perspective is featured in Chapter 5.

Brian Dangerfield discovered SD (industrial dynamics as it then was) over 40 years ago after a period in an OR unit in industry. Working at the (then) University of Liverpool School of Business Studies, he was a researcher on a project looking at the role of stocks in the UK economy. Rather than take the obvious econometric route he realised that an understanding of the macro role that stocks played in the workings of the economy would be much enhanced if, instead, macro-economic models were SD simulation models. Variables representing stocks had to be divorced from the flows which changed them. He was impressed with the way (i) SD models were forged at the policy level, (ii) separated out resource flows from the information flows driving changes in those resources and (iii) were also able to embrace relevant ‘soft’ variables which, using another methodology, might be excluded altogether. In sum he concluded that, given his overall knowledge and real-world experience with OR (including traditional simulation techniques), the SD methodology was just about the most promising in the landscape of OR, and his subsequent research career has concentrated on its use and development.

1.3 Navigating the Book

To our knowledge this book is unique – there is no similar coverage in one single volume. Books which cover both methodologies (e.g. Pidd's Computer Simulation in Management Science, 2009) merely split the page coverage – there is no attempt at integration, or a detailed comparison and description of how both approaches may be used in the same project. This book provides a seamless treatment of a variety of topics: theory, philosophy, detailed mechanics and practical implementation, all written by experts in the field. While some chapters are aimed at beginners, others are more advanced; the book also includes three software chapters which are very practical in nature.

The book is structured in seven unequal sections, three of which only contain one chapter. This Introduction forms the first section, and Pidd's concluding chapter the seventh. The second section, ‘Primers’, contains two chapters which provide a basic introduction to DES and SD respectively. These chapters are both written by experts with many years' experience of teaching and using each technique: Chapter 2 on DES (Stewart Robinson) and Chapter 3 on SD (Brian Dangerfield). The authors assume no prior knowledge of either technique, or an academic background in mathematics, statistics or OR. They are aimed at students and practitioners alike. The aim of both primers is to provide sufficient understanding to enable the average reader to get a basic grasp of the topic and be able to appreciate the subsequent chapters. The primers do not attempt to provide the breadth and depth of material offered in a typical MSc course in SD or DES, and they do not contain a great deal of technical detail. References are provided for anyone who does wish to delve deeper into the technicalities!

The third section also consists of a single chapter. By way of contrast with Chapter 2 and 3, the authors of Chapter 4 (Kathy Kotiadis and John Mingers) take a strongly academic stance. This chapter provides a theoretical background for much of the discussion in later chapters about combining different modelling paradigms. This chapter, which was originally published as a research article in the Journal of the Operational Research Society, discusses the combination of problem structuring methods with hard OR methodologies. Kotiadis and Mingers reflect on the barriers to such combinations that can be seen at the philosophical level – paradigm incommensurability – and cognitive level – type of personality and difficulty of switching paradigm. They then examine the combination of soft systems methodology and DES within a healthcare case study. They argue, by way of the practical application, that these problems are not insurmountable and that the result can be seen as the interplay of the soft and hard paradigms. The idea of yin and yang is proposed as a metaphor for this process.

The fourth section, ‘Comparisons’, is by far the largest and represents the heart of the book – its raison d'être. It contains five chapters, all of which...