Design and Control Co-optimization of a Mixed Hybrid Electric Powertrain Architecture

Electrification of vehicles can improve energy efficiency and reduce emissions in vehicle operations. Studies have focused on designing powertrains of several topologies, including serial, parallel, and power-split hybrid powertrains. In this study, a design and control co-optimization is demonstrated for a novel mixed-powertrain architecture. The component sizes of the mixed powertrain are optimized to minimize the fuel and component costs. Nested optimization is applied with a surrogate integrated operation and control model that evaluates powertrain performance in the inner loop. The surrogate model is trained to capture the powertrain performance under a near-optimal power management approach, namely, the equivalent consumption minimization strategy (ECMS). Using sequential quadratic programming, optimal results are obtained and verified using a high-fidelity powertrain model which is equipped with ECMS control. Multiple optimal designs can be selected for this multi-objective optimization problem. All designs can either improve the fuel economy or reduce the component sizes, or both, depending on the weighing of fuel economy and component cost. Compared to the original design, one of the optimal solutions can improve the fuel economy by 24.5% while keeping component sizes similar. Another optimal design can achieve similar fuel economy while reducing the rated power of the electric machines by 49.2% and 30.1%, the battery voltage by 51.3%, and battery capacity by 72.9%. Four selected commercially available powertrain architectures are investigated to benchmark the mixed powertrain. The fuel economies of the benchmark architectures are up to 15% different compared to the fuel economy of the mixed powertrain. The range in both component and fuel cost of the Pareto front of the mixed powertrain is widest among the powertrains in this study.