Shiro Matsumura

Improvement of de-NOx device control performance using a software sensor

Abstract This paper presents a mathematical estimator, named a ‘software sensor’ for predicting the NO x in the fluegas of power plants. The NO x prediction by the software sensor is used in the control loop of NH 3 gas injection to the de-NO x device to reduce the NO x emission from the stack. By the use of the prediction in place of the actual measurement of NO x by the analyzer, the delay time in the control system is eliminated, and as a result the control performance can be improved, thus resulting in a reduction in operating cost.

Improvement of De-NOx Device Control Performance Using Software Sensor

Abstract This paper presents a mathematical estimator, named a “software sensor᾿ by the authors, which was developed to predict the NOx concentration in the fluegas of fossil fuel power plants. The NOx prediction by the software sensor is used in the control loop of NH gas injection to the de-NOx device which is installed between the boiler-outlet and the stack-inlet in the fluegas path in order to reduce the NOx emission from the stack. By the use of the prediction in place of the actual measurement of NOx by the NOx analyzer, the delay-time in the feedback control system is eliminated, and as a result the control performance of the NH gas injection can be remarkably improved, thus resulting in the operating cost reduction. In the paper, the software sensor based on an ARX model, model parameter estimation, consideration on the software sensor implementation are described together with some results of simulation studies, and field tests which demonstrate the advantages obtained by introducing the proposed software sensor.

Automatic realization of human experience for controlling variable-pressure boilers

Abstract Control methodology of thermal power plants has been established on a long time accumulation of human experience: it produces sufficient control performance. Tuning of parameters in the controllers, however, is a complicated process. In this paper, the authors propose a method to determine the parameters theoretically based on a nonlinear separation control method. By using the method, the complicated and time-consuming parameter tuning process can be replaced mostly by a theoretical approach using computer programs. The usefulness of the proposed method was verified by means of computer simulation studies in which nonlinear power plant models were used.

Adaptive Control for Steam Temperature of Thermal Power Plant

Abstract A discrete time adaptive control system was applied to the steam temperature control of a boiler for electric power generation. In the system. the adaptive controllers are applied in parallel with the conventional PID controllers for supcrheated and reheated steam temperatures. The plant dynamics was described with polynomials whose pararneters are adaptively estimated from the data of the plant under routine operation. The adaptive control signals are synthesized on the basis of the estimated parameters so as to achieve the specified control objective. The system was applied to a 375MW plant and found to realize far better control performance than the conventional PID control system.

Modeling and Denox Control System for Fossil-Fuel-Fired Power Plants

Abstract In this paper, a de-NOx system, that complies with the envirorunental regulation and at the same time saves the operation cost, is introduced. The paper consists of two parts, i.e., the modeling of the NOx producing process and the design of the de-NOx control system. At first, a multi-input single-output ARX, model describing the relationship between the NOx content in the flue gas and process input variables relevant to it is identified through the Recursive Least Square Method. The model is used to predict the current NOx gas content which is otherwise unknown because of the remarkable measurement delay. The validi ty of the model is confirmed by comparing the prediction of NOx with its measurement at the actual plant. The controller design is performed on the basis of this model, assuming that the model represents the behavior of the actual plant. A control system consisting of an adaptive feedforward controller and a PID feedback controller is designed. The effectiveness of the controller was examined through simulation studies, which showed the proposed de-NOx system had an excellent performance to achieve the objectives.