(208e) Robust Model Predictive Control for a Nanofluid-Based Solar Thermal Power Plant

A recent increase in energy demand, along with global warming and depletion of fossil fuels, have propelled the development of alternative sustainable energy technologies. Among the new technologies, parabolic trough collectors have stood as a promising alternative due to their performance regarding energy recovery, which can be enhanced by the employment of nanofluids as the heat transfer fluid. Moreover, while traditional controllers are commonly found in these concentrators, more advanced techniques such as Model Predictive Control would be preferred since in addition to their capability of dealing with the intermittent nature of solar radiation, this optimal-control based method can be tuned to minimize operating costs or energy usage. Nevertheless, since the equations that accurately describe the behavior of nanofluids properties are still in development, and these controllers are model-based, they should be robust enough to reject disturbances and track set-points in spite of the uncertainty in the aforementioned properties.

In this sense, the work presented seeks an approach for the optimal control of a complete Al₂O₃-water nanofluid-based solar thermal power plant, steering it from initial conditions to target conditions, aiming to assess the performance and robustness of a proposed MPC controller applied to the such-mentioned power plant, under uncertain conditions at the nanofluid parameters and variations on the intensity of solar radiation. In order to implement proper model predictive control actions, all the components of the system (a parabolic trough collector, energy storage tanks and an Organic Rankine Cycle) were described in terms of energy and mass balances. The mathematical model was formulated as a Nonlinear Programming model and solved by well-known optimization algorithms embedded in the GAMS optimization environment, followed by a simulation of the optimal control actions, repeating this procedure according to the control intervals set at the controller. Furthermore, the robustness of the controller was tested by studying three different cases: one considering that the equations selected perfectly describe the system, and two more cases considering +20% and -20% modeling error in the nanofluid thermal conductivity, respectively. Note that a modeling error in the thermal conductivity triggers a cascade-effect over several parameters depending on it, including the heat transfer coefficient.

Results show how under wide-range uncertain scenarios, the implementation of MPC controller was found to be suitable and robust in these novel nanofluid-based solar thermal power plants. The system displayed favorable dynamic behavior and was able to reject strong radiation disturbances properly by the action of the proposed MPC controller, leading to generation of energy close to the one expected from off-line nonlinear programming optimizations. Therefore, these results indicate that MPC controllers are appropriate for their implementation on Concentrated Solar Power, and can be valuable in the transition towards renewable energies.