Model Reuse: How to Avoid Hidden Costs and Truly Benefit from Previous Work
Brett Collins, Jason K. Howlett, Thomas R. McCarthy, Christopher A. Schultze · AIAA Scitech 2021 Forum · 2021
View Video Presentation: https://doi.org/10.2514/6.2021-0788.vid In the simulation realm, opportunities for reuse are often plentiful, and it has grown to be an expectation that this is always the best path. Quite often it is, but the answer is not always as straightforward as some assume. There is a lot of allure to leveraging prior successful models as a means of jump starting a new effort. Often the ideas of the credibility or similarity of functionality are especially appealing. This tends to lead to optimistic estimates of implementation and compatibility, but these need to be examined in an honest, reflective way. There are also some key, less considered, facets of simulation models that should be investigated to truly assess a model’s reuse potential and the costs associated with it. The architecture and sustainability of a model relative to the new environment it will be integrated into warrants consideration as these are often significant cost drivers that are overlooked in reuse scenarios. When all of these components are carefully reviewed, it becomes evident that in some cases, it is more beneficial and specifically less expense to begin from a fresh start. On the surface, this seems counter intuitive, because isn’t starting with something always better than starting with nothing? This misconception is subtly propagated by metrics like Technical Readiness Level (TRL) which are monotonically increasing. Often today, engineers feel obliged to reuse without fully considering the topography of the problem, as if they must surely be leaving value on the table or somehow unnecessarily reinventing the wheel. This paper seeks to walk through the definitions, value, and cost considerations of credibility, functionality, architecture, and sustainability to provide a means to generate a situationally relevant assessment of the actual potential value of a reuse opportunity. Reuse is usually the right answer, but sometimes it is not, and that possibility should be fairly left in the decision space. A case study is provided to illustrate the application of intentionality in the choice of whether or not to reuse models.