Mathematical modeling is widely used throughout any scientific industry when trying to predict the behavior of dynamic systems. Oftentimes it is desirable that these models be simple and efficient, while still delivering accurate data. This paper builds and examines an energy-based (bond graph) model of an automotive fuel delivery system and suggests which elements are required to produce a Proper Model. The Model Order Reduction Algorithm (MORA) provides a mechanism to quantitatively rank each element in the model and determine its contribution to the system dynamics. Utilizing this approach, a model is developed that retains 98% of the complete system energy from 12 of the most active of 25 elements. This model requires 46.6% less simulation time while continuing to provide an adequate prediction of the system response.