In 1987, I was appointed as a Research Associate at Newcastle University on an SERC/ACME project entitled "The application of computer aided production management (CAPM) systems in engineer to order companies". This work, in collaboration with the NEI Group, involved identifying the characteristics of engineer-to-order/ make-to-order capital goods companies; the status of their CAPM systems; common CAPM problems and finally to conduct research to identify possible solutions to the CAPM problems identified. The first stage of the work involved conducting a survey of nine companies and used systems and data modelling methods (IDEF and Entity-Relationship Diagrams) to describe the systems. This was the first time that a systematic approach had been applied to investigating CAPM in the capital goods sector. This work was published in a number of papers in the period 1989-1995.
 

The general limitation of these survey and modelling methods is that they provide static views, which provide only a snapshot in time. They do not provide a mechanism for dynamically testing alternative scenarios. A novel large-scale discrete event simulation of a manufacturing facility interacting with a CAPM system was therefore developed and used to investigate the issues identified by the surveys. This work made a number of contributions. Firstly, it was the first simulation model that included the necessary features to model the complex environments encountered in MTO/ETO capital goods companies. Secondly, it was the first simulation that modelled the interaction of a CAPM system with a manufacturing facility. Thirdly, data structures capable of representing complex products and processes were devised, together with an efficient simulation algorithm that enables large problems to be evaluated in a reasonable amount of time; Fourthly, it provided a framework for a series of designed experiments that provided significant insight into CAPM issues in ETO/MTO companies. The case studies were based upon manufacturing data obtained from NEI Parsons. This work formed the basis of my PhD thesis, papers 4,18,19,20,21,22,23 and 25.

Simulation provides a mechanism for testing and evaluating scenarios. It does not enable 'optimal' solutions to be identified. Indeed many of the heuristics used by CAPM systems are known to generate sub optimal results. Pupong Pongcharoen, in his PhD has developed a planning tool based upon Genetic Algorithms that enables 'optimum' schedules to be generated that minimise penalty costs due to earliness and lateness. The GA procedure proposed by Goldberg has been modified to include four repair procedures that 'correct' schedules that are infeasible. New genetic operators have also been developed and tested. This work is based upon the work of Shawn Hines who developed a graphical Tk-Tcl interface for the simulation model. Data obtained from Parsons has been used for the case studies. An important focus of this work has been the identification of appropriate GA parameters and operators that produce good solutions quickly. Efficient experimental designs are necessary to investigate the large number of factors concerned. We have collaborated with Dave Stewardson (ISRU) on statistical matters. Papers 6,12,14,37,44,45,46,56 and 58 are based upon this work.
 

The original simulation modelling was deterministic. The simulation model was developed by Adrian Wall (PhD) to include stochastic effects based upon a wide range of theoretical or empirical distributions. This was an important development as ETO/MTO are subject to a wide range of uncertainties. A further PhD student Dong Ping Song has developed planning methods that take into account uncertainties in process times and developed stochastic search optimisation techniques (papers 8,9,11,35,42,48 and 55). Two papers have recently been submitted that explore the selection of dispatching rules in capital goods companys that produce complex products in stochastic environments.

 

The simulation model contains a detailed description of manufacturing facilities. Methods of performing cluster analysis were developed as a mechanism for identifying potential manufacturing cells. Hinrichs (MEng) implemented Rank Order Clustering and Boe and Cheng’s Close Neighbour Algorithm (paper 26). Adrian Wall (MPhil) implemented hierarchical clustering methods using similarity coefficients and dendograms. Chris Lee (MBA) and Roger Aubrey (MSc) applied these techniques at Vickers Defence Systems. The problem with these methods is that the results are often hard to interpret and it is common for there to be few clusters identified. This is due to the high component variety and the nature of processes involved in the manufacture of capital goods. Hinrichs (MEng) therefore applied a Simulated Annealing algorithm, which was used to generate 'optimal' layouts for Parsons’ Heavy Machine Shop. Wee (MEng) developed a Genetic Algorithm approach to the same problem (paper 47 and also the basis for papers 15,54 and 59).
 

Tony Wells (PhD) was the Facility Manager at Siemens Semiconductors and was interested in the layout and configuration of semiconductor plants. His work is based upon four case-studies undertaken at plants in the UK, Germany, USA and Taiwan. He is investigating the relationships between plant configuration performance benchmarks. Tony is now Managing Director of MW-Zander UK and derives much of his business from this research.
 

With the deregulation of the Power Industry there has been considerable restructuring within the capital goods industry. The reasons and effects of these changes are described in paper 52. There has been a general shift from in-house manufacturing towards a concentration on design and assembly, with manufacturing being outsourced. I have been involved in interdisciplinary work in supply chain management with Dr McGovern (Business School) in collaboration with AMEC Offshore, Clarke Chapman, Siemens Power Generation Systems, VA Tech and Wellman Booth. Papers 2,3,5,7,31,39 and 52 are a result of this work. This research identified the crucial role that specifications play in supply chain relationships. This is the subject of further research supported by the Small Grants Committee. It is hoped that this work will be of interest to the EDC companies. Some preliminary research on the simulation of supply chains in the red meat industry was conducted with Professor Harvey and Dr McLeay in AEFM (paper 32).
 

The survey methods used in the early CAPM work have been developed to use business process analysis (used in papers 3,16,24,31,33,34,39 and 53). This has been the basis of our work in knowledge management (papers 38, 39 and 51). I was an invited speaker at an EPSRC/ESRC workshop (paper 40). My experience of CAPM systems, systems analysis, business processes and computing has provided the background to the three Teaching Company Programmes. Papers 24,33 and 34 are based upon this work. A further paper has been submitted that explores the shaping of technology by political processes within the firm. The revenue from the Teaching Company Programmes was used to support two research associates (Phil Medlam and Shawn Hines) and one PhD student (Dong Ping Song).
 

The work on the development of optimised scheduling methods will continue as it is necessary to further develop the method to deal with larger more complex problems. This has several aspects. Firstly, the program code needs to optimised to reduce the required computation time. Secondly, it is necessary to develop models to predict the size of the search space for different problems. The relationships between the complexity of schedules and appropriate GA configuration will also be considered. The relative performance of Genetic Algorithm approaches with other methods such as Taboo Search and Simulated Annealing will be considered. It is also hoped to simultaneously apply these methods to manufacturing layout problems. This would represent a significant advance in manufacturing system design. This type of work would be ideally suited to overseas PhD students. The second aspect of the proposed work relates to the application and development of these methods for use in industry.