An Evolutionary Approach to Rapid Development of Simulation Models

Funder: Engineering and Physical Sciences Research Council. grant number: EP/E037631/1
Value: £104,025

The effective design of innovative manufacturing systems is a difficult task. This is not only because manufacturers usually do not have any way of getting early visibility of production costs and reliable process times, but also capturing, storing and reusing best practice manufacturing processes are not easy. Levels of financial investment can be immense and the consequences of poor system design can be catastrophic to the whole business. Discrete-event simulation, or simulation as it is more popularly known, is one of the most useful and worthwhile tools that industrialists can apply during manufacturing design and redesign. Such simulation models typically include representations of machines, part flows, operation rules, schedules, conveyors and people, and then predict the performance of manufacturing facilities in terms of measures such as production volume, manufacturing lead-time, etc. Although simulation techniques have been around for more than 30 years in the UK, evidence suggests that their usage within the manufacturing industry is relatively low. This is mainly because a) compared to desktop applications, simulation tools are much more difficult to use and thus require specific skills to operate them, b) simulation tools are often overly-capable for the problems being addressed and are expensive to acquire and c) building, verifying and validating the models are time-consuming and resource-intensive. The challenge this research attempts to address is how to improve the uptake of simulation techniques within the UK-based manufacturing industry by positively disrupting the barriers of simulation usage through the reduction of model development time. The novelty of the proposed research pushes the boundary of current research in the area of rapid simulation model development.

The proposed research will provide a substantial advancement of knowledge from previous work by:1. Exploring the use of cladistics and evolutionary analysis as a basis of the classification of both manufacturing systems and the problems being addressed using simulation.2. Investigating a new way of rapidly generating simulation model templates based on both typical manufacturing systems to be modelled and the problems to be addressed.3. Shifting the concept of 'model building' towards 'model assembling', where model elements can be retrieved from a ready-to-use component library and the complete model will be generated automatically in order to speed up the model development.