Study And Application Of Whole Process Optimization Of Coal Blending For Thermal Power Plants

With the rapid social and economic development of China, the demand of coal for power generation is also increasing sharply. Thus, for coal-fired power plants, coal blending combustion is becoming more commonly as more and more off-specification coals are received under the present coal market in China. Basic research upon coal blending combustion characteristics has been studied for a long time with many useful results. In recent years, the utilization of coal blending technologies have been widely implemented in the power industry.PCDM (pulverizing different coals via different mills and blending inside the boiler) is one of coal blending technologies which is widely used for most of coal-fired power plants in China. In this study, with consideration of global optimization in the process of coal blending combustion, coal blending optimization, coaling decision, combustion optimization and also the manegement in whole process have investigated. Relative decision-making models were proposed and software for the whole process optimization of PCDM has been developed and utilized in power plants.Firstly, a coal blending optimization model was established. Blended coal properties prediction model has been studied based on the data from experiments. The results indicate that, moisture, sulfur content and calorific value of blended coal can be considered as linear, ash content can be calculated via formula, while volatile and ash fusion temperature could not be predicted via the former ways but could obtain better predictions by using neural network. Based on this coal properties prediction model, pollution emission characteristics prediction model and a fuzzy identification approach of coal spontaneous ignition characteristic have been established. These models were applied in the coal blending decision process to guarantee the decision scientifically efficient. Then, physical programming has been introduced for the coal blending objectives description, which overcomes the defects of linear calculated way and traditional fuzzy decision model need a human determined weights. Physical programming can makes the objective function values much more practically meaningful. Based on the objective function, exhaustive method, improved genetic algorithms and multi-objective genetic algorithm have been compared and the results suggest that intelligent algorithm is not always necessarily optimal, but the algorithm should be chosen according to the size of optimizing space and calculation time requirements. In addition, to achieve high-precision ratio of PCDM in the process of coal blending, a combinational optimization model of mills has been established which include coal-mills combination and mill outputs combination. Under a certain blending program, by blending precision could be improved by adjusting the output of the mills on-line, which was called as " two-class coal blending optimization" in this paper. At last, based on the prediction model of blended coal properties, mathematical descriptions of blending requirements, the results of optimization algorithm and combinational optimization model of mills, an intelligential model for the power plant coal blending was proposed, which contains both mathematical calculation models and practical situations of power plants and modifies the blending decisions by blending knowledge base and rule base.One of the problems for the optimization in the whole process of PCDM is the real-time identification of coal types in boiler. In this study, a dynamic monitoring technique of bunker was developed to solve this difficulty which caused by the delay from coaling to burning. It was worked by monitoring the layered interface of different coals. Furthermore, a new method by the heat balance calculating inside the mill was also inventied to validate the results by maching the moisture content in coal. By combining the two approaches, the accurate online coal identification could be achieved.Based on coal indetification, this paper also investigated the operation optimization methods of coal blends combustion for thermal power plant using a case-based reasoning(CRB) approach and a rule-based reasoning(RBR) approach. As a prerequisite, considering safety, economy and environment protection in the process of PCDM, an online evaluation for coal blends combustion performance based on coal identification has been proposed,. Then, case description, case evaluation, case retain, case retrieval and case-base reasoning which are the main issues of CBR method were studied. A case-generating algorithm employing an evolutionary strategy was proposed in which the case base evolves while retaining new cases. The evolutionary strategy can significantly guarantee the case base in an appropriate size. Then, data-mining algorithms were employed to find the association rules of high boiler efficiency and low NOx.Based on these rules, a rule base was established. Integrated CBR and RBR respective advantages and disadvantages, a system knowledge was present for a expert system which used for multiple coals combustion optimization.For the whole process management of coal in the process of PCDM, in this paper, the idea of intelligent management of coal yard was introduced. It includes auto stack/fetching decision and real time status monitoring of coal yard. This paper proposed a power plant dynamic programming of coal stock(DPCS) with the idea of Preservation Coal Price (PCP). PCP indicates that power plant cannot blend low-price coal blindly, but should keep a balance between coal quality and price when using coal blending. At last, as a major innovation of this work, a whole process optimization and decision-making system for coal blending was developed based on the above all research. It can supervise every process of coal and make synthesized optimization, and combine the process of stack, blending, fetch, burning, purchase to make multiple objective coordinated optimizations, which are intended to reach a safe, economic and environmental coal blending. It could be used as a global fuel management and dispatch platform. Also, visualization techniques are developed to model coal yard and dynamic bunker. The system was firstly applied in a600MW power plant, the boiler efficiency was increased by0.7%while the SO2emissions was decreased by20%. So far, the system has been promotion applied in more than10power plants and achieved good results.