Problems in manufacturing and logistics can be very complex. Handling them in a quantitative way is therefore a complicated process. Just as people use tools for the calculation of static what-if models (i.e. spreadsheets), there is a need for tools that can solve complex dynamic problems. Such tools do exist - in the form of decision support software for the simulation and analysis of manufacturing and logistic systems.
Some problems - such as those with a large number of interdependent variables - do not lend themselves to ready translation into a system of mathematical comparisons, subsequently to be optimised with OR techniques such as linear programming. It is pre-eminently these kinds of problem for which simulation is suitable.
Simulation and Reality
Simulation means imitating reality. A simulation model can be helpful for obtaining insight into the functioning of systems and, subsequently, for the step by step improvement of these systems. Manufacturing and logistic systems normally have waiting time in them - waiting times of resources (under-occupation), of products (stocks) and customers (delivery reliability). Simulation software can help you to make the right trade off between each.
With simulation you can test your decisions before you take them. Weaknesses in the design of manufacturing systems are discovered before the system is actually built. Simulation on a computer used to be characterised by labour-intensive model building, i.e. programming, as a result of which the use of simulation has been restricted to projects of particular types where the programming skills have been developed over time. A disadvantage of this historic approach is that because of the required knowledge in programming, the one who builds the model is usually not the one who formulates the problem.
New forms of simulation software make simulation more accessible - perhaps not quite as a general-purpose tool, but certainly as a tool for the power-user rather than the programmer. This allows models to be built by the engineer who understands the problem, and means that models can be built and analysed quickly and easily.
In this world of graphical interfaces, such packages also have an emphasis on visuality, so that the user can maintain an overall 'view' of the model.
A typical 'production process modelling package' allows the user to build up a model by positioning resources like machines, buffers, conveyors, reservoirs, warehouses, transporters, roads and labour on the screen. Then, one can define routings (production stages in their respective orders) in these layouts.
The user has full freedom in defining the flow of products through a model, from straightforward routings to very complex, product-related and condition-dependent routings.
To detail the model, the user can use a large range of pre-defined parameters (like cycle times, arrival rates, batch sizes) and of course can add his/her own parameters and decision rules.
The simulation can then be started and often animated without further alterations. The simulation may be run for a fixed period of (simulated) time or until some pre-determined condition is met.
Some packages allow the user to stop the simulation and then to 'single-step' through to examine what is happening in great detail.
After the simulation has run, the user can call up appropriate analytical information (like waiting hours, numbers produced, throughput time, utilisation rates, costs of the system). Rapid alterations can then be made to the model to evaluate alternative approaches and configurations.
Modelling is really fast. Not only because the software is menu driven, but also because the way of modelling concurs with the user's way of thinking (what you see is what you simulate) and because the model is interactively built and run; each change made to the model can immediately be verified by simulating and animating the model without delay.
Modelling is flexible, because of the structure of the software and because of the possibilities for the user to add his/her own rules, conditions, parameters and attributes, all within the menu structure!
Many of the packages have typical standards/values for a range of machines and services, but the user can override these.
Animation serves two purposes. The first is the verification of a model. Does the model behave as intended? The second is a convincing presentation of results. Good, recognisable animation is more convincing to an audience than a simple numeric model.
see Simulation Dynamics at http://www.simulationdynamics.com/
see Logistics Simulation at http://www.logsim.co.uk/
Send E-mail to
TSN@The-Saudi.Net with questions or
comments about The Saudi Network.