Reasons to Simulate
by Van Norman
How should we make decisions in the face of uncertainty and risk?
Computer-based simulation can help. Simulation is a computer model of a real-world process or system. It involves generating a history of the system and analyzing that history to draw inferences concerning the system's operating characteristics. Why should we develop models to simulate systems? There are six reasons.
1. Models help us make better decisions. Because a computer-based model reflects exactly what we want to do, we can make decisions in the model and determine what their effects are going to be. We can test every aspect of a proposed change before committing resources.
The simulation tools we're talking about visualize dynamically how the system actually operates. That visualization gives us insight into how to make better decisions. Our actual plants are so big and complex, there's no place to stand to look at everything going on. And because these systems are so complex and interrelated, if we change one thing, often it perturbs or interacts with another area.
A model allows us to experiment with alternatives. It's very, very difficult to make physical changes in your real facility, but it's easy to make changes in a computer-based model. So it allows you to investigate many more alternatives as you look at what you might do to make improvements in your system.
2. Models develop our understanding. Many physical models, like the blueprint of a system, aren't dynamic--but the real system is. A simulation model is the only model that's going to give you insight into how the system will perform under various operating conditions. Simulation results will tell you exactly how the system is going to work. Good simulation models can be within a few percent of how systems will actually perform.
3. Models can help us diagnose problems. Because a simulation model is computer-based, we can control time. We can run the model very fast to look at days or weeks worth of production performance in a few minutes, or we can slow the model down and spend minutes looking at a few seconds of operations.
Operations are complex, and it's difficult to always understand the interactions of the various elements. A model captures those interactions and shows us how the system actually operates.
4. Models can identify constraints. We can discover the problems of operational bottlenecks. We can also exercise a system beyond its design constraints. We design systems to operate under various scenarios, but we know the world in the future is going to be different. A model allows us to plan for the changes we need to make in the future by driving the system--the model of the system--beyond its design constraints.
5. Models are a powerful way of building consensus. If we bring all of the people affected by a decision together and show them in the model how the decision is going to solve a problem or make an improvement, we can get everyone to buy into the change and understand what they need to do to support the change.
6. Models are a way of training the overall team. Training models allow for learning and experimentation. If you're looking into a new system that has lots of flexibility and can be operated in various modes, bring in the senior managers before the system is built and allow them to play with the model and experiment with it. Then they can come up to speed quicker about how the system is going to work.
We don't allow pilots to fly 747s right off the bat--we train them via physical simulation models. That's because it's expensive to learn with the real system.
Van Norman is CEO of AutoSimulations Inc. E-mail, vnorman@autosim.com; phone, (801) 298-1398, ext. 120; fax, (801) 298-8186.