This document summarizes a methodology for sizing mechatronic systems using bond graphs and genetic algorithms. It proposes using bond graphs to develop structural models that satisfy control criteria like controllability and observability. Behavioral models are derived from the bond graph models. Genetic algorithms are then used to optimize the bond graph elements, with controllability and observability gramians as the fitness function. The method is demonstrated by applying it to size resistance (R) and capacitance (C) elements in a bond graph model of an automatic scooter transmission. Optimization yields values of 0.4227 for R and 0.3132 for C. Validation involves dynamic simulation to compare the sized model against the initial model.