“MathWorks tools enabled us to simulate various powertrains, develop accurate plant models, test control strategies, and validate the overall design,” says UWAFT captain, Matthew Stevens.
The MathWorks provided training in MATLAB®, Simulink®, Stateflow®, and PSAT, a modeling program based on Simulink. “Having a product with a quick learning curve or that students were already trained in was critical to the team’s success,” Stevens comments. “MathWorks tools could be used for multiple stages in the design process, minimizing the number of software programs that students needed to learn.”
UWAFT developed more than 400 PSAT simulations to compare fuels, technologies, and powertrain sizing. Optimization Toolbox™ and a sophisticated Design of Experiments enabled them to understand the relationship between component size and vehicle performance and then determine the optimal powertrain.
They used Simulink to develop a plant model of the fuel cell power system, which included the engine, the battery, the fuel cell, and a DC/DC converter.
MATLAB, Simulink, Stateflow, and Control System Toolbox™ enabled them to develop the hybrid control strategy (HCS), which determines the amount of power coming from the fuel cell. MATLAB helped them to find the optimal appropriation of power between the fuel cell and battery over a specific drive cycle.
The DC/DC converter boosts the fuel cell voltage and controls the power from the fuel cell. Simscape Electrical™ was used to model the circuit, which was controlled by a PI controller. The team investigated the frequency response and stability of the circuit using Bode plots and pole-zero maps in MATLAB. Simulation enabled them to verify correct operation, determine circuit efficiency, and calculate values and ratings of inductors and other components.
Because the fuel cell turns on and off, it creates a discontinuous function that is difficult to optimize using traditional methods. To find the optimal control benchmark, UWAFT therefore used Global Optimization Toolbox, which does not require the function to be continuous. They also used Deep Learning Toolbox™ to model the hydration of the membranes within the fuel cell stack.
The team used Embedded Coder® to target UWAFT satellite controllers throughout the vehicle for on-target rapid prototyping using an embedded controller.
They are currently testing powertrain components, refining the vehicle control strategy, integrating the advanced fuel cell into the vehicle, and researching weight-reduction possibilities.
“We are definitely interested in using other MathWorks products throughout the competition, and hopefully in our future careers in many other projects,” Stevens says.