Optimization Techniques for Hybrid Electric Vehicle Analysis Using ADVISOR

System optimization is an extremely important step in the vehicle design process. Recently the National Renewable Energy Laboratory (NREL), both internally and through its subcontractors, has been actively developing and applying optimization tools to vehicle systems analysis problems associated with hybrid electric vehicles. We are applying both locally- and globally-focused optimization tools to these analysis problems. This paper describes the current status of the tools under development and the application of these tools to a specific problem. The optimization tools evaluated include FMINCON (gradient-based constrained optimization routine included in the MATLAB Optimization Toolbox), DIRECT (non-gradient based optimization routine), Response Surface Approximations and Direct Gradient-based Optimization (commercial routines available in VisualDOC 2.0 from Vanderplaats R&D). The four optimization algorithms were applied to a simple two-dimensional problem so that the solution methods could be visualized and the relative performance quantified. The tools were then each applied to the optimization of a fuel cell-powered hybrid electric vehicle modeled in ADVISOR 3.1. Both the component sizes and the energy management strategy where included as design variables. The fuel economy was maximized under the constraint that vehicle performance must not be less than its conventional vehicle counterpart. Vehicle and/or component costs were not included in this study but could easily be included if suitable models were identified. Based on this work, the following conclusions can be drawn: The impact of multiple local minimums and a small amount of objective function 'noise' (due to necessary state of charge balancing for hybrid vehicles) caused gradient based optimizers to stop before they were able to find design points as good as those found by the non-gradient based DIRECT routine. For the fuel cell-powered hybrid electric SUV design problem, several local optimums and one likely global optimum design have been determined. The global optimum is based on 2461 function evaluations. Based on the assumptions in this study, the optimal fuel cell-powered hybrid electric SUV would consist of a 66 kW (net) fuel cell, 107 kW and 16.8 kWh battery pack (twenty-eight 49 Ah modules, 372 volts) and a 126 kW traction motor. The optimal control for this system turned out to be mainly a series thermostat design with some power following capability under peak loads. This vehicle is expected to achieve a City/Highway composite fuel economy of 56.5 mpgge (miles per gallon gasoline equivalent).