Considerations for the Design of Deployable SOC Systems with Multi-Stack Modules: Applying Experience from Kw-Scale Test Tystems

Abstract

Introduction : Commercial-scalesolid oxide cell (SOC) systems offer a unique and attractive solution to the challenge of supplying high-efficiency and fuel-flexible power at competitive system costs using low-carbon fuels [1, 2]. These systems require repeating SOC stack assemblies to be packaged into multi-stack modules (MSMs) to bring system capacity to the MW scale. During the system design phase, relatively simple models (e.g., single-cell representations) of the SOFC power block are often used to develop system-level models for performing trade studies, developing system requirements, and/or estimating performance. Such simplified approaches typically assume that thousands of cells operate equivalently, and neglect consideration of the coupled thermal-fluidic, electrochemical, and power electronic aspects associated with multi-stack assemblies. These simplifications may be computationally efficient but can also lead to overestimation of system efficiency performance, under prediction of both higher thermal gradients that may be present within stacks and considerable inter-stack heat transfer, and larger variance than expected in other stack operating parameters, such as fuel utilization. While careful design of multi-stack modules can help to mitigate the negative impacts of scaling SOC systems, there are interacting effects to balance of plant (BOP) components that should be addressed to ensure that the system meets both performance targets, such as power density, efficiency, and operability within the stack operating envelope. This work draws on kW-scale testing of MSM assemblies and explores how impactful deviations in stack-to-stack operation can result in reduced system performance, especially in cases where significant gas maldistribution is present between stacks within MSMs or between multiple MSM arrays. The presentation provides an overview of computational modeling and experimental testing to assess how single-cell representations versus more detailed (3-D stack and MSM) models for the SOFC power block impact stack and system performance estimation. In particular, the effect of gas maldistribution on system efficiency, SOFC operational limits, and BOP sizing is quantified through experimentally validated model-based simulation. Methodology: The system concept modeled in this work is an 80 kW rated hybrid SOFC-internal combustion engine system [3, 4]. The engine is used as a bottoming cycle to increase system efficiency by utilizing the remaining reactant gases in the SOFC exhaust stream to generate power. Fresh reactant gases are conditioned and compressed to the system operating pressure of 3 bar absolute. The fuel stream is blended with anode exhaust recycle gas and preheated and partially reformed in a packed bed reactor before entering the SOFC modules which are made up of twelve 5 kW stack assemblies packages into 3 MSMs. Positive displacement compressor hardware brings fresh air up to the operating pressure and is preheated to supply the SOFC modules. Preheating is done through gas-to-gas recuperative heat exchangers that utilize the hot exhaust gas streams to preheat the inlet gas streams. Unreacted exhaust gas that is not recycled is sent to a spark-ignited internal combustion engine (SI-ICE) to produce additional power to increase system efficiency and overcome parasitic loads in the system. A flow diagram of the system is shown below. Two SOFC modeling approaches are compared for the system concept described above. The first uses a simplified approach that only models a single repeating unit (RU) for the SOFC modules. This approach assumed that all cells would act identically to one another and that no effects arise due to stack and module construction. The SOC cells are modeled using a one-dimensional “down the channel” framework to capture the temperature, species composition, and current distributions over the active area of the cells. The simplified single-cell representation of the SOC body simply multiplies the expected outputs of the cell by the cell count. By contrast, the second framework uses an MSM model in which each SOFC stack is modeled independently with individually tunable parameters for their thermal, flow, and electrochemical operation. These models are tuned to manufacturer data for specific stacks as well as experimental characterization of 1 and 5 kW stacks done at Colorado School of Mines. Each stack is manifolded such that gases are delivered to each stack via the common port. Within a module, the flow of reactant gases to each stack is not equal and is quantified by its maldistribution; represented as a percentage difference from the ideal fraction of flow delivered to each stack as seen in equation 1 for stack in position i as follows: The module external manifolds are modeled as a discrete pipe network and are benchmarked against a high-fidelity CFD model of the manifolds in COMSOL Multiphysics and experimental characterization of manifolds and module operation [5-7] . The experimental test bed used to inform the multi-stack and system levels models is a grid connected, 36kW capacity, pressurized rig capable of testing both multi-stack modules and single-stack test modules (STMs) within a pressure vessel equipped with numerous feedthroughs for routing of gas plumbing, diagnostics, and electric power lines. This vessel maintains stack operating pressures up to 10 bar a . The test bed is capable of running on reformed natural gas as well as simulated gas mixtures. The MSM is outfitted with thermocouples and pressure taps at the inlet and exhaust ports of each stack for diagnostic measurements. In-line heaters placed within the vessel bring the gases to operating temperature [5-8]. The data collected from experimental characterization of MSMs helps to inform modeling efforts. First, pressure drop characteristics for each stack within a module are quantified to accurately predict the gas maldistribution to each stack for a given flow condition. Secondly, a set of cell material tuning parameters found to fit model polarization curves to the experimental data. Finally, effective insulation parameters for each stack are quantified using temperature data of the outer module case taken at multiple points under load. Results: The hybrid system configuration defined above is simulated using both the single-cell and MSM representations for the SOFC assembly in three cases. Case A uses the single-cell representation where all cells are assumed to operate equally within the SOFC assembly. Cases B and C model the full MSM assemblies with a maximum gas maldistribution in the stack external manifolds of 10% and 20%, respectively. This gas maldistribution is imposed such that the first stack on the manifold rail received the maximum flow (10% or 20%, respectively) and reduced linearly to the fourth stack which received the least flow (-10% or -20% respectively). Table 1 shows the system operating conditions, efficiency, and net system power for the cases tested. The average fuel utilization for all stacks is held constant at 65% and the cathode outlet temperature among all stacks is constrained to maximum of 630°C. The average stack current for all cases was 32 amps. Comparing Case A in which a single cell model was used to cases B and C which use MSM models shows that the introduction of MSMs decreases the system efficiency by up to 9% in the case of ±20% gas maldistribution. Because the maximum cathode outlet temperature is constrained at 630°C to ensure safe operation, cases B and C must flow more air through the cathode to account for the stack which received the lowest airflow due to maldistribution even though the module outlet temperature falls well below the 630°C limit. Therefore, the parasitic load of the air compressor hardware increases significantly. This increase in load results in a drop of efficiency of up to 9% for case C. Additionally. The system power is de-rated to take up the additional parasitic load by up to 10 kW. Conclusions : This study explores the limitations of simplified modeling approaches for commercial-scale solid oxide cell (SOC) systems. Conventional single-cell representations fail to capture critical thermal-fluidic, electrochemical, and power electronic interactions within multi-stack modules (MSMs). In the effects studied. Including experimentally validated MSM models captures phenomena that may require de-rating system power by up to 10%, to mitigate potential losses not shown with simplified models. These results highlight the necessity of more comprehensive modeling approaches during system design to accurately predict performance parameters and ensure MW-scale SOC systems can meet both efficiency targets and operational stability requirements while maintaining competitive costs in the low-carbon energy landscape. References: [1] D. E. Tew, R. A. Cox-Galhotra, V. R. Lecoustre, M. Lyubovsky, and G. L. Soloveic

Identifiers

DOI

10.1149/ma2025-031226mtgabs

Journal

ECS Meeting Abstracts

Year

2025