Abs. 670, 205th Meeting, 2004 The Electrochemical Society, Inc.

Diesel Reforming with SOFC Anode Recycle

Rodney L. Borup, Michael A. Inbody, Jose I. Tafoya, Dennis R. Guidry and W. Jerry Parkinson Los Alamos National Lab MST-11, MS J579 P.O. Box 1663 Los Alamos National Laboratory Los Alamos, NM 87545 e-mail: Borup@lanl.gov, Phone (505) 667 2823 Fax (505) 665 - 9507 ABSTRACT One application being developed for Solid-OxideFuel-Cells is for auxiliary power on-board vehicles. For this application, the fuel will likely be the same fuel as is used to operated the vehicle, thus for heavy duty vehicles should be diesel fuel. SOFCs have shown the ability to directly oxidize hydrocarbon fuels, however the power densities are low. Thus SOFC-APU applications currently require fuel reforming to operate at their highest power density and lowest system cost. Our goal is to develop fundamental knowledge supporting the on-board reforming of diesel fuel for SOFC/APU (Solid Oxide Fuel Cell Auxiliary Power Unit) applications. This includes understanding of the parameters that affect fuel processor performance, lifetime and durability. This work examines potential methods for system integration of a diesel reformer for transportation applications with a SOFC. In particular, we examine the diesel reforming operating conditions suitable for the SOFC, and include modeling and experimental measurements examining recycling of the SOFC anode exhaust into the diesel reformer. Diesel fuel has a strong tendency to pyrolyze upon vaporization, and has a high propensity for carbon formation during reforming for hydrogen production. Carbon formation can be suppressed by the addition of high levels of water/steam to the reformer, however this requires a source of water. The system integration for water recovery requires substantial additional components, such as heat exchange, condensers, and water recirculation tanks and pumps. The possibly preferred method for water recovery for SOFC APU systems is to recycle the SOFC anode exhaust, thus recovering the water produced in the fuel cell. Various methods to inject and mix air, steam, fuel with the addition of simulated anode recycle have been used to make reforming measurements. The diesel reforming operation has been modeled and measured during for various anode recycle ratios, to examine relative hydrogen and carbon monoxide production, axial catalyst temperature profiles and carbon formation. The incorporation of a fuel nozzle into a diesel reformer has shown the nozzle temperature to be important to prevent the nozzle from clogging from carbon formed due to the pyrolysis of diesel fuel components at the reformer injection point. Use of simulated SOFC anode exhaust recycle has also been used to examine the change in performance on the diesel reformer. This complicates reactor design and operation, especially considering high concentrations of hydrogen in the anode recycle will tend to auto-ignite. In addition, the amount of recycle has a large tendency to change the operating characteristic of the fuel reformer. Large anode recycle ratios require higher O/C ratios for similar adiabatic temperature rises, and the light-off

of the reforming catalyst is decreased. This is demonstrated in figure 1, where the auto-thermal reforming of diesel fuel is shown with simulated SOFC anode recycle of 20 and 30 %. The 30 % recycle shows a much slower temperature rise in the auto-thermal reformer catalyst bed than does the 20% recycle, and requires higher O/C for the same temperature rise. Figure 1. Catalyst Temperature Profiles During AutoThermal reforming of diesel fuel with 20 % and 30 % SOFC anode recycle. Analysis of carbon formed during the partial oxidation and stream reforming of liquid hydrocarbons show different carbonaceous deposits depending upon operating and formation conditions. Modeling of air/steam/diesel fuel mixtures indicate that carbon formation is favored until relatively high steam content or high operating temperature. There seem to be at least three amorphous forms that are all slightly different and all have slightly different thermodynamic functions. Differences in pressure show only a slight change in the results.