High Fidelity Transient Simulations of the Multi-module HTGR Special Purpose Reactor

Abstract

In this report we present the methodology and sensitivities of a novel simulation capability that couples state-of-the-art control algorithms with high-fidelity reactor physics simulation tools. The summary of the contributions of this report are: • Development of a novel simulation capability that combines state-of-the-art control algorithms with high-fidelity reactor physics simulations • 3D multiphysics transient load follow simulations of an High-Temperature Gas-Cooled Reactor (HTGR) microreactor • Detailed sensitivity analysis of parameters affecting controller performance This report provides a review our representative HTGR-like microreactor, and previously developed Model Predictive Control (MPC) based autonomous control algorithm for the reactivity control system. We then give an overview of the high-fidelity simulation techniques and the developments necessary to perform load-follow transient simulations. A detailed description of the reduced order reactor model and corresponding state-space model for the controller then follows. The MPC formulation and closed loop controller structure is described where adaptive MPC is used to capture the nonlinear effects inherent in the point reactor model, and observer is developed to enable the coupling between the controller and high-fidelity simulation tool. An “off the shelf” optimization library for Operator Splitting solver for Quadratic Program (OSQP) is used to solve the control optimization problem.

Through several numerical experiments, we observe that the MPC and point reactor state-space model provide excellent control of the reactivity control system where tracking errors between the core power and set point are within 0.234%, with control inputs remaining within constraints. Transient load-follow results from a 3D high-fidelity multiphysics reactor simulation coupled with the MPC controller to calculate rod position.

Extensive sensitivity tests are also performed to gain insights into the performance of the controller in terms of some of its tuning parameters, and accuracy of key reactor physics quantities needed in the state-space model. Overall, the sensitivities of the parameters evaluated revealed that the controller is quite robust. Most parameters would need to differ by more than an order of magnitude to significantly degrade the controller performance. The two largest sensitivities identified are the accuracy of the constraints applied to the controller, and accuracy of the control drum worth curves. A reduction of the constraints to 10% of their nominal value increased the maximum tracking error to 8%, while having control drum reactivity worths off by ±60% demonstrated the controller can become saturated and provide highly oscillatory control inputs (although it was still able to provide an accurate control solution). Typical technical specifications for zero power physics testing and measurement of reactivity worth require significantly tighter agreement than 60% between the calculated design values and measured values. Further, a detailed understanding of reactivity control system drive constraints is necessary for the final safety analysis report to satisfy licensing requirements. Therefore, we conclude that current regulatory procedures for the design and operation of reactors should ensure sufficient performance of an MPC controller (excluding risks due to failure modes of the controller which is beyond the scope of this report). We also include as an appendix a draft of a journal article we plan to submit based on the work documented in this report.