(587d) Model Predictive Control for a Large-Scale Chemical Looping Combustion System in a Packed Bed Reactor

As the severity of global warming is more widely recognized, there is increasing focus on mitigating climate change by achieving net-zero carbon dioxide emissions, which is driving a need for effective carbon capture technologies such as chemical looping combustion (CLC). In CLC, an oxygen carrier (OC) is added to the combustion process in order to generate energy while avoiding contact between the air and the fuel. This is done by alternatingly exposing the OC to air and fuel, respectively oxidizing the OC to produce energy and regenerating the OC. By avoiding contact between the air and fuel, the CO₂ generated as a combustion byproduct will be produced in a stream which primarily consists of carbon dioxide and steam. Water can be easily condensed out, resulting in a concentrated CO₂ product stream without the need for additional energy-intensive separation processes. CLC in packed bed reactors (PBRs) is being increasingly investigated because it does not require a separation process to recover the OC and because the oxygen carrier can be used more effectively (i.e., more of the OC can be oxidized or reduced during each stage) [1].

Recent modelling studies have investigated the feasibility of packed bed CLC in large-scale systems for energy generation in 350-400 MW_e power plants [2]. The heat produced during the oxidation stage can be sent to a downstream turbine to produce energy. The system is dynamic, as the oxygen carrier gradually approaches full conversion, and the outlet states – such as the temperature – will vary throughout each stage. Gas turbines can be operated over a certain temperature range (i.e., 1100-1900 K) and are typically required to operate at steady-state since large fluctuations in temperature can damage the turbines [3]. As such, an adequate control system for the oxidation stage is needed to maintain an outlet air stream which remains constant near the desired temperature. Previous studies have focused on the optimal inlet mass flowrate of air and fuel for energy generation [2] [3] whereas other studies have investigated dynamic control strategies [4] [5]. Nevertheless, most CLC literature considers a fixed mass inlet flowrate for each stage. Model predictive control (MPC) is a widely used model-based control strategy that is particularly suitable for cases with constraints and multiple inputs and outputs. To the authors’ knowledge, the application of MPC for packed bed CLC has not been addressed in the literature.

In this work, a dynamic mechanistic model for CLC in a PBR is presented. The proposed model considers the microscale particle model as well as the macroscale bulk reactor, making it a multiscale model. The model consists of coupled partial differential equations and is implemented using method of lines, discretizing the particles in the radial direction using centered finite differences and the reactor in the axial direction with orthogonal collocation on finite elements. The system of differential equations is then solved using the Radau integration method. The multiscale nature of this system results in a relatively large number of states – 920 states in the oxidation stage, and 2520 states in the reduction stage. While using nickel oxide as the OC and methane as the fuel, this model was compared to experimental data at various operating conditions for validation. The model showed good agreement with literature data for both the oxidation and reduction stages, suggesting that it provides reasonable predictions for the behaviour during CLC over a range of different conditions. Hence, the proposed dynamic model could be extended to a large-scale system suitable for industrial energy generation, using a reactor length of 11 m and diameter of 5.5 m, as proposed by Spallina et al. [2].

The dynamic mechanistic model developed in this work was linearized around a nominal operating condition and was used as the internal model in an MPC strategy. For the oxidation stage, the inlet mass flowrate and temperature of air will be manipulated online to control the reactor’s outlet temperature. For the reduction stage, the inlet mass flowrate and mass fraction of fuel are adjusted to control the outlet gas concentrations to maintain a concentrated CO₂ stream. In both cases, closed-loop linear MPC is used to control the variables of interest in the outlet stream while accounting for process operation and the constraints within the system (e.g. maintaining an inlet flowrate within reasonable bounds and an outlet temperature within the turbine’s operational range). The closed-loop MPC results provide a control strategy that can maintain a sufficiently high outlet temperature during the oxidation stage and a concentrated CO₂ and H₂O outlet stream during the reduction stage. Additional cases involving changes in the load during operation were also investigated and provided satisfactory results. The present study has revealed strategies to effectively generate energy without damaging the downstream turbine and while isolating the carbon dioxide into a pure CO₂ stream. Hence, the proposed study can be used to determine a feasible approach for industrial implementation of CLC technology in a PBR for energy generation.

[1] S. Noorman, M. van Sint Annaland and H. Kuipers, "Packed Bed Reactor Technology for Chemical-Looping Combustion," Industrial & Engineering Chemistry Research, vol. 46, pp. 4212-4220, 2007.

[2] V. Spallina, P. Chiesa, E. Martelli, F. Gallucci, M. Romano, G. Lozza and M. van Sint Annaland, "Reactor design and operation strategies for a large-scale packed-bed CLC power plant with coal syngas," International Journal of Greenhouse Gas Control, vol. 36, pp. 34-50, 2015.

[3] L. Han and G. M. Bollas, "Dynamic optimization of fixed bed chemical-looping combustion processes," Energy, vol. 112, pp. 1107-1119, 2016.

[4] L. Han and G. M. Bollas, "Dynamic optimization of fixed bed chemical-looping combustion processes," Energy, vol. 112, pp. 1107-1119, 2016.

[5] K. Toffolo and L. Ricardez-Sandoval, "Optimal Design and Control of a Multiscale Model for a Packed Bed Chemical-Looping Combustion Reactor," IFAC PapersOnLine, vol. 54, no. 3, pp. 615-620, 2021.