During the study, global and local

During the study, global and local optimization algorithms were compared. For the multi-objective problem, similar results are obtained using both methods. This allows us to conclude that this optimization problem has an objective function that is bothconvex and unimodal. However, the addition of design variables and constraints may change the nature of the problem, and the use of local methods may not be possible. In this particular case, the use of a local optimization technique is preferred since itsapplication results in the same solution as the global methods using less computational time. For the single-objective optimization problem, the differing results of the two methods suggest that the objective function is in this case non-unimodal with atleast one local maxima in addition to the global maximum value. Furthermore, the benefit of using both method types is demonstrated by the large variance in operating conditions revealed by the solutions: in practice, one maximum, although at a slightly lower value, may actually be the superior choice. This interesting result suggests that it is prudent to continue to employ both types of algorithms in tandem during any optimization, if possible.This work reveals a large number of possible avenues of future research. The existing fuel cell stack and system models would benefit from fewer assumptions and simplifications; for example, the assumed constant pressure drop in the humidifier and in the cooling channels and fully humidified membrane. A new optimization could be performed that takes into accountnot only the operating conditions as design variables but also the geometric parameters of the fuel cell. Indeed, a combination of operating parameters with physico-chemical parameters as design variables would result in a comprehensive optimization of a fuel cell system. Also, the objective function can be modified to reflect the operating conditions of an actual fuel cell application: the weighting factors, set to 1 for this study, could be matched to the particular power cycle of the application, increasing the accuracy of the model. Subsequent papers will reflect this ongoing attempt to increase the sophisticationand accuracy of the stack and system models.