A lot has been written about the merits of different simulation software tools. Less has been written about simulation as a problem-solving strategy – a technology in its own right, independent of the software tool used.
In a simulation project, the focus is on understanding. For example, to reduce inventory in your supply chain, seek first to understand why inventory is as high as it is. Simulation creates understanding by tracing what happens in your system over time, which accounts for the state it is in.
As a simulation model gets developed, you learn about the factors that determine how your system behaves. Moreover, you identify the rules you use to make routine decisions. Change the rules and you will change how the system performs, often dramatically.
The most important rules are put into the simulation model so that the model more closely approximates your operation. The mere development of a model makes you think in depth about rules, and enhances your understanding.
You are shipping products by truck. Full trucks are cheaper and you don’t want to overpay for freight. So you might use a rule like “wait until we have a full truck before letting shipments go.”
But, you also want to get products to customers quickly. If you wait for full trucks, products take longer to get to customers. Are you better to wait for full trucks, or let them go earlier to reduce delivery time, even though it costs more?
The answer: It depends. It depends on the demand level, differences between full and partial truck costs, how well customers tolerate delays, and other factors. Simulation lets you explore and quantify the tradeoffs among these factors.
Once a model has been completed, we validate it to make sure the results make sense. Then, we test out different scenarios to see how improvements can be made. The insights you gain during model construction and validation are often what provide ideas for improvement.
A well-designed model contains all the right “levers” to pull to see how changes affect the system. These levers (called parameters) can be systematically varied in a simulation experiment so you can plot how the system responds to changes.
SDI puts a high priority on doing simulation experiments; we like to say that a simulation model is only as good as your ability to experiment with it. So when we apply simulation technology, we emphasize creating models with all the right parameter “levers”.
