Skip to main content

Agile System Design

Design of Agile Production Facilities

In order to support an agile manufacturing system the engineers running it must continually reconfigure it to handle new models and ever changing variant volumes. This involves monitoring the manufacturing cells and the production schedules and ensuring that:

  • The system has sufficient capacity to deliver the schedule
  • As few parts of the system as possible are under utilised
  • Alternative routings exist in case of machine breakdowns

If cells or processes within cells are under utilised then they can be replaced by cells or processes that can reduce the loading on other cells provided that there are alternative routes available for variants that do use them. If cells are over utilised then it is essential that duplicate cells are produced in order that volumes can be achieved. Performing this continual improvement to the facility design is a complex problem though essential to economic running of such a facility. Within this project an expert system known as SIMAID (simulation aid) has been developed (11). This allows the engineer to input data about the structure of the different vehicles to be manufactured and the processes required to join them. Information about model volumes, area available for the system, working patterns etc. are also input and this data is used to develop a number of alternative layouts for the system. These are then judged against input criteria such as cost, machine utilisation etc. and are displayed as options. The user can then either modify some of the input constraints and run the package again or take the system design into a discrete event simulation package to perform a more in depth analysis of the situation.

The system does not produce optimal designs as this is seen as both impossible and not required as the input data in terms of volumes will either be predicted or historical and will therefore not be correct. Instead the systems are designed to operate over a range of different production volumes with as little sensitivity to volume and model variation as is possible.

It is perhaps no surprise that the production schedule and system design are intrinsically linked. It is rare however that the two are considered together in the early stages of a project. Often a simple change to the manufacturing system such as the addition of a process into one cell can remove the need for complex rescheduling. Similarly rescheduling variants can remove the need to make large changes to the manufacturing system. The fact that the systems developed using this approach are decoupled allows cells or processes to be inserted wherever needed so that specific low volume variants can be handled without affecting the efficiency of the entire manufacturing system.

Results from SIMAID have been compared with real case studies from the automotive industry and have been almost identical in content to designs produced manually, though the structure of the facility is often different (i.e. two parallel lines running at half speed rather than one single line.) Trials have shown that using SIMAID layouts can be produced in less than 10% of the time taken by experienced system designers. It’s use has been recognised as important during the early stages of system concept design and quotation as it accurately predicts equipment and space requirements.

Outcomes:

Publications

Young, K.W., Muehlhaeusser, R., Piggin, R.S.H. and Rachitrangsan, P. (2001) Agile Control Systems, Automobile Engineering, pp 189-195, Vol 215 No. D2, Proceedings of the Institution of Mechanical Engineers, ISSN 0954-4070