Date of Award


Document Type


Degree Name

Doctor of Philosophy (PhD)


Electrical Engineering


This research endeavor began with the design and construction of a new hydrogen test facility at the National Renewable Energy Laboratory (NREL). To improve the electrical link of wind-based electrolysis the characterization of a proton exchange membrane (PEM) elect) /zer under varying input power was performed at NRELs new test facility. The commercially available electrolyzer from Proton Energy Systems (PES) was characterized using constant direct current (DC), sinusoidally varying DC, photovoltaics and variable magnitude and frequency energy from a 10 kW wind turbine.

At rated stack current and ~ 40°C the system efficiency of the commercial electrolyzer was measured to be 55%. At lower stack current it was shown that commercial electrolyzer system efficiency falls because of the continuous hydrogen purge (-0.1 Nm3 hr'1) used to maintain the hydrogen desiccant drying system. A novel thermoelectric-based dew point controller is designed and modeled to reduce the penalty to renewable sources because they do not always operate at 100% of rated stack current. It is predicted that the thermoelectric design when operated 100% of the time at full current to the thermoelectric modules would consume 3.1 kWh kg'1 of hydrogen. Using the higher heating value of hydrogen and a stack efficiency of 60% to produce the hydrogen that is continuously vented, the desiccant system consumes about 5.7 kWh kg'1.

Design of the UND electrolyzer sub-systems responsible for all aspects of water, power to the stack, and hydrogen conditioning enables more flexible and precise experimental data to be obtained than from an off-the-shelf system. Current-voltage (IV) characteristic curves were obtained on the UND system at temperatures between 7 ~ 70°C. The anode and cathode exchange current densities are fitted to 2.0 E-06 e0'04jT and 0.12 e0026T A cm"2 respectively. Stack conductivity was fitted to 0.001T + 0.03 S cm"1. The three coefficients represent physical stack parameters and are extracted fiorn the temperature dependent IV curves and the Butler-Volmer equation as the model using the Levenberg-Marquardt nonlinear curve fitting algorithm.

The characterization, dew point controller and modeling work will enable improved efficiency, cost reductions and better modeling of the wind to electrolysis system.