The efficient and safe performance of nuclear power plants of the future requires remote monitoring, control, and condition-based maintenance in order to maximize their capacity factor. Small and medium reactors, in the 50–500 MWe power range, may become commonplace for certain applications, with a design features for remote deployment. Such a reactor may be part of a smaller electrical grid, and deployed in areas with limited infrastructure. Typical applications include power generation, process heat for water desalination, and co-generation. There are other considerations in the deployment of these reactors: development of effective I&C to support nuclear fuel security monitoring, longer than normal fuel cycle length, and increased autonomy in plant operation and maintenance. A Model Predictive Controller (MPC) for the IRIS (International Reactor Innovative and Secure) system has been developed as a multivariate control strategy for reactor power regulation and the control of the helical coil steam generator (HCSG) used in IRIS. A MATLAB-SIMULINK model of the integral reactor was developed and used to demonstrate the design of the MPC. The two major control actions are the control rod reactivity perturbation and the steam control valve setting. The latter is used to regulate the set point value of the superheated steam. The MPC technique minimizes the necessity of on-line controller tuning, and is highly effective for remote and autonomous control actions. As an important part of the instrumentation & control (I&C) strategy, sensor placement in next generation reactors needs to be addressed for both control design and fault diagnosis. This approach is being applied to the IRIS system to enhance the efficiency of reactor monitoring that would assist in a quick and accurate identification of faults. This is achieved by solving the problem from the fault diagnosis perspective, rather than treating the sensor placement as a pure optimization problem. The solution to the problem of sensor placement may be broadly divided into two tasks: (1) fault modeling or prediction of cause-effect behavior of the system, generating a set of variables that are affected whenever a fault occurs, and (2) use of the generated sets to identify sensor locations based on various design criteria, such as observability, resolution, reliability, etc. The proposed algorithm is applied to the design of a sensor network for the IRIS system using multiple design criteria. This enables the designer to obtain a good preliminary design without extensive quantitative information about the process. The control technique will be demonstrated by application to a real process with actuators and associated device time delays. A multivariate flow control loop has been developed with the objective of demonstrating digital control implementation using proportional-integral controllers for water level regulation in coupled tanks. The controller implementation includes self-tuning, control mode selection under device or instrument fault, automated learning, on-line fault monitoring and failure anticipation, and supervisory control. The paper describes the integration of control strategies, fault-tolerant control, and sensor placement for the IRIS system, and demonstration of the technology using an experimental control loop.
