by Jon M Quigley
Simulation activities can help evoke the requirements for a product without actually having to build the product first to learn something. Simulation need not be highly technical, though it can be. I have seen simulation of screens for an instrument cluster human machine interface (HMI) that made use of excel links to illustrate the appearance and show how the navigation worked. This was a cheap and quick way of learning how the product should work.
There are more complicated ways, however, that typically employ sophisticated software and hardware. We can develop complex systems of components to simulate our product. We can use these tools to develop our requirements by testing our concepts before we build the prototype parts. We can refine our algorithms by exercising the subassembly within the context of a simulated system. We will need to confirm our models for the simulation, in fact represent the real world stimuli expected. To do that we will need to understand the use of the product by the customer and environmental exposures. Ultimately, we need to establish a correlation between the simulation and the actual world use and exposure. This provides us with a level of certainty of the simulation results.
Not only can we use simulation at the system and subsystem interaction level, but we can use simulation on the subassemblies of the system. We can use circuit simulation tools to understand the design capabilities well in advance of building a prototype part. Experience suggest this as an under used tool with the belief that modeling the individual components difficult. Many electronic component suppliers provide these models. I recall using a simulation program in the 90’s to evaluate my power supply design for an automotive application. Once the virtual design appears to handle the defined transients, I knew it was time to get a real prototype part.
We can go far with virtual designs and simulation and we discuss this topic in a number of our books (especially in Testing of Complex and Embedded Systems). However, at some point, we need to confirm prototype parts will indeed react the way our simulation and models indicate. To do this, we will still require actual prototype parts. We reduce the number of prototypes required, and we reduce the risk of the capability and quality of those first parts through judicious use of simulation and modeling.