Today’s systems are often large and complex with many interacting parts and elements. To deal with such complexity, system designers and operators turn to models of their systems to gain insights. They key word is “insights”. Any modeling that does not result in some insight into the system is just modeling for modeling sake, or, at best, improves the understanding of the system by merely trying to model it. We take a different view by looking at the spectrum of modeling (see figure). Modeling and simulations range from the rudimentary to the very detailed, and their performance ranges from the inaccurate to the highly accurate. Furthermore, models typically, but not always, become more accurate with more detail. Orbit models follow this trajectory where one can use two-body ideal models and proceed to to very accurate n-body models with perturbations. These models increase the accuracy, but often require much more detail and processing resources. Neither model along this trajectory is necessarily good or bad; their utility is determined by the questions that are asked of the model in the context of the available information.
Across the spectrum of models, there are some very fundamental models that exhibit a high degree of accuracy without much need for detail. These can be beautifully elegant models that reveal fundamental insights into the behavior of nature or systems. These models appear, but not often in engineering work. However, they are the models that any modeler strives for. At the other end of the spectrum are models that have a high degree of detail. These are the models that may or may not be accurate, but include models for just about everything in the system regardless of its impact. Often these models are so complex and have so much in them, that users cannot really determine what determined the model’s results. Consequently, these models do not provide much insight. It’s actually quite easy to develop these models, since the modeling really never has to make a decision about what’s important. At Questiny Group, we strive to be in the upper left region of the figure. This is what we call a “balanced model.” It is where user’s can derive the maximum amount of insight from the model or simulation. The level of detail in the model is balanced with the accuracy of the model. In this way, users can gain an understanding of their system because they can connect the results to fundamental characteristics of their system.
We, in fact, advocate a modeling “flight” plan that begins with relatively crude models of the system (perhaps with low accuracy, although the accuracy is often unknown at this stage), then proceed to add details to the model. Often modeling stops when the users feel they have added sufficient detail, but we recommend a further step that proceeds to remove or simplify the models. By simplifying the model and observing the results, users can begin to determine the fundamental characteristics of their system that result in particular model behaviors or outputs. In this manner, users gain a fundamental understanding of their system. At some point, as more detail is removed or the model becomes overly simplified, the accuracy suffers. The models are then backed off from this to the point where balanced detail and accuracy are achieved.
This philosophy of modeling is great, but does not mean anything unless it can be expressed in a modeling environment. This is exactly what the Questiny’s WiNS program does. It provides a modeling environment where users can explore the full model spectrum for their system-adding and removing detail at will, and derive fundamental insights into their system.
We welcome your feedback on this thought, please use the following form to send us your thoughts: