Skip to main content
European Commission logo
English English
CORDIS - EU research results
Content archived on 2024-06-18

Design oriented flow distribution optimization of the solid oxide fuel cell stack operating under electric load

Final Report Summary - HYDROSOFC (Design oriented flow distribution optimization of the solid oxide fuel cell stack operating under electric load.)

Solid oxide fuel cells (SOFCs) are attractive electric power generators because of their high energy conversion efficiency, environmental benefits and multi-fuel capabilities. However, flow maldistribution of reactant gases in the stack results in performance loss and decreased electric efficiency.

In the HYDROSOFC project several alternative approaches were used to predict reactant flow distribution in the SOFC stack with the goals of the:
1. selection of the computational method suitable for the engineering quality prediction of the reactant flow distribution in the stack,
2. evaluation of the most important factors affecting flow distribution in the SOFC stack operating under electric load, and
3. experimental verification of the calculated flow field pressure drop.

Firstly, a semi-three-dimensional computational fluid dynamic (semi-3D CFD) flow distribution model was developed to simulate flow and pressure distribution in the SOFC stack operating under electric load conditions. In the following step a two-dimensional analytical model, based on the system of mass and momentum equations (analytical model), was implemented for the stack operating under open cell voltage conditions. Finally, a two-dimensional stack model based on the hydraulic network approach (hydraulic model), was developed for the stack operating under electric load. The fuel cell flow field and manifold geometry details were included in all models, for both fuel and oxidant, in order to enable a systematic models' comparison.

Flow distribution of oxidant and fuel in the stack depended primarily on dimensional factors such as the fuel cell flow field and gas manifold geometry. These factors defined hydraulic resistance of fuel cell flow field and manifolds. The ratio of pressure drop in the inlet and outlet manifolds to the pressure drop in the fuel cell flow field defined flow maldistribution in the stack. In addition to fuel cell flow field and manifold dimensions, the manifold configuration, i.e. Z-flow or U-flow, had the most pronounced effect on the flow distribution. The Z-flow configuration yielded better flow distribution when compared to U-flow configuration. Other factors, such as chemical reactions on the anode side including water-shift reaction and methane steam-reforming, as well as stack operation under electric load, affected flow distribution to a lesser extent. They influenced pressure drop and flow distribution as a result of changes in the reactant flow rate velocity and gas mixture viscosity. For example, fuel dynamic viscosity of the 50 % hydrogen (H2) and 50 % nitrogen (N2) fuel mixture changed by only 3 % between stack inlet and outlet at 80 % fuel utilisation. However, when natural gas and steam were supplied to stack for internal reforming in the stack, inlet fuel gas viscosity was almost 20 % lower. As a result, pressure drop in the cell flow field was lower and the resulting flow maldistribution was higher for the stack fuelled with the natural gas and steam mixture. In both cases, flow maldistribution increased with the higher electric load of the stack, even though the changes were not significant. In most cases, much less than 1 % increase in flow maldistribution was predicted.

The results of computational models were verified using experimental setup for the pressure drop measurements. The oxidant flow rates corresponding to electric load of 5000 A/m2 and the oxidant utilisations in the stack ranging from 15 % to 50 % were used. Several flow characteristics were measured, including inlet and outlet flow rates and differential pressure across the cathode side. All three models provided reliable estimates of pressure drop for all flow conditions, with discrepancies between data and model being lower than 10 % for all the cases that were tested. There were no adjustable parameters used by any of the computational models. Finally, the hydraulic model of the stack seemed to provide a feasible option to calculate flow distribution in the stack with sufficient accuracy for engineering estimates.