Dynamic modeling of reversible solid oxide cell stack and control strategy development

[Display omitted] •Temperature gradients and reactant utilization are controlled with two PI controllers.•PI controllers can ensure safe ReSOC mode transition under various current ramp rates.•Maximum cell inlet-outlet temperature difference is 62 °C in SOFC mode at 3211 A/m2.•Controllers fix reactant utilization in seconds and temperature gradients in minutes. Solid oxide cell systems (SOCs) are increasingly being considered for electrical energy storage and as a means to boost the use of renewable energy and improve the grid flexibility by power-to-gas electrochemical conversion. The control of several variables (e.g., local temperature gradients and reactant utilization) is crucial when the stacks are used in dynamic operation with intermittent electrical power sources. In the present work, two 1D models of SOC stacks are established and used to investigate their dynamic behavior and to select and tune a suitable control strategy. Subsequently, safe operating ranges were determined to meet the thermal constraints of the stack by analysing not only the fuel cell (SOFC) and electrolyzer (SOEC) individual modes but also the switching between the two modes when the stack operates reversibly. The dynamic analysis shows that the control loops of our multi-input (reactant molar flow rates), multi-output (reactant utilization and maximum local temperature gradients) control system are strongly decoupled. Therefore, a proportional integral control strategy can be used to prevent dangerous stack operating conditions in dynamic operation. Finally, the controllers were tuned, and their transfer functions were reported. Convective heat transfer via air flow allows controlling the temperature of the solid structure of the cell/stack component, thus avoiding issues related to temperature variation during transient operation. Moreover, the reactant utilization controllers can avoid component fracture or degradation owing to fuel starvation under dynamic operation. The process can be approximated by two first order transfer functions. It can help in the design of more complex control systems in the future if necessary, with embedded process models, such as model predictive control. Results in the simulation environment are preparatory to the programming phase of an actual controller in real-world applications.