Modeling And Operation Optimization For Nuclear Power Plant With Multiple Modular High Temperature Gas-Cooled Reactors

Developing advanced technologies of nuclear energy is regareded as an important measure to deal with the energy crisis and environmental pollution problems.As one of the promising technologies in building the Generation ? nuclear energy systems,the modular high temperature gas-cooled reactor is characterized by its inherent safety,economic competitiveness,standardization and modularization,and potential broad applicability.The world's first multi-modular demonstration plant for high temperature gas-cooled reactor pebble bed module(HTR-PM)adopts the mode of "dual-modules with one-turbine".Due to different reactor type.complex mechanism,unique power generation structure as well as the inability to directly learn from the operation experience of existing single-reactor nuclear power plants,the operation of HTR-PM faces enormous challenges.To achieve safe,reliable,economical and efficient operation for HTR-PM is a severe test.Model-based operation optimization has important guiding significance for the operation of HTR-PM.Based on the industrial background of HTR-PM and the professional background of process system optimization,the thesis focuses on reducing model plant mismatch and promoting model-based optimization to converge to the true optimum especially when wide-range operation is required and model-plant mismatch is inevitable.The main contents and contributions of this work can be summarized as follows:(1)Nonliear rigorous modeling for HTR-PM.The steady-state nonlinear rigorous model of HTR-PM is established,and this model is verified in typical operating conditions.Due to simplifications,assumptions,and empirical formulations in describing this complex system,there inevitably exists mismatch between the model and the plant.A systematic parameter estimation method is designed to improve the prediction accuracy of the model itself.Integrating parameter estimability analysis,simultaneous estimation and reliability evaluation,this method is able to avoid the ill-conditioning caused by complex model structure,over-parameterization and sparse measurement.The parameter estimation results of the HTR-PM rigorous model verifies the effectiveness of the systematic method.(2)Integrated parameter mapping and real-time optimization for operation optimization.Being valid only in a certain operating range,process model will lose accuracy if blindly applied to a wide range.The resulting model-based operation may not only lose the optimality but also lose the feasibility which jeopardizes the safe operation of the process.With the increasing demand for operating over wide conditions,the model-based optimization faces more severe challenges.As the existing operation optimization strategies are unable to quantitatively determine the application range of the model,integrated parameter mapping and real-time optimization algorithm in the framework of trust region is proposed,which adaptively updates the application range of the model based on model evaluation.The integrated algorithm converges to the true optimum despite the structural mismatch.Under reasonable assumptions,the convergence of the integrated algorithm to the true optimum is proved.Significant load changes in HTR-PM verify the effectiveness of the the integrated algorithm.(3)Dual adaptation strategy for operation optimization.The integrated algorithm in the framework of trust region ensures the model accuracy by updating parameter mapping at each step.When updating the sensitivities of parameters with respect to manipulated variables involved in parameter mapping,it is necessary to perturb the process multiple times to obtain the relevant measurements at high cost.For large-scale complex systems with many manipulated variables,updating parameter mapping is even more expensive.To reduce the cost,an intuitive idea is to reuse the model as much as possible until it is necessary to update the model.Based on the above idea,the adaptive model update mechanism compatible to trust region extension or based on gradient related steps is designed,which updates the model only when necessary.Dual adaptation algorithms with adaptive update mechanism of the model and its application range are therefore designed,which avoid frequent perturbation to the process.Dual adaptation algorithms converge to the true optimum despite obsolete or inaccurate model.Under reasonable assumptions,the convergence of dual adaptation algorithms to the true optimum is proved.Williams-Otto process optimization,significant load changes and load allocation optimization in HTR-PM verify the effectiveness of dual adaptation algorithms.