Development Of Coal Blending Optimization Model And Expert System In Boiler And Simulation And Optimization Research On Coal Combustion In A Boiler

The reasonable collocation of blended coal is very important for the safe, economic and stable operation of the boiler. The research is from a major industrial project entitled "research on the adaptability of multi-coal" which focuses on the coal blending technology in power plants in HuNan Province. A series of site investigation and experiments were conducted. A nonlinear prediction model for the properties of the blended coal and coal blending optimization model were build based on neural network, genetic algorithm, enumerative algorithm and simulated-annealing algorithm. An expert system for coal blending was developed. The temperature in the boiler was simulated and the simulation results were compared with the experimental data, and the parameters Of the boiler were optimized numerically. The main contents and conclusions of this research include:(1)15kinds of typical coals were chosen which are popularly used in YiYang Power Generation Limited Company and these coals were blent into45kinds of coals by certain methods. Net calorific value, elementary and industrial analysis, flammability index and ignition and stable combustion index were tested by experiments. Compared the results of the blended coal properties from experiments with the results calculated by linear weighted average of the component coals, most of the properties from the two methods show nonlinear.(2) A nonlinear prediction model of blended coal properties was established based on generalized regression neural network. Net calorific value, elementary and industrial analysis, ignition temperature and ignition and stable combustion index were predicted by this model. Compared prediction results, linear weighted average results with experimental results, prediction errors are under5%, which are smaller than that calculated by linear weighted average method.(3) Three coal blending models based on enumerative algorithm, simulated-annealing algorithm and genetic algorithm were developed, in which, the lowest price of the blended coal was chosen as the objective function and coal properties were calculated by generalized regression neural network. An Expert System for coal blending based on the coal varieties of HuNan power plants was developed. It has the following main functions:updating the database of the single coal; predicting the coal properties and elementary analysis Of blending coal according to the database; finding the best coal blending scheme by enumerative algorithm, simulated-annealing algorithm and genetic algorithm according to the constraints presumed by the users.(4) Based on RNG k-ε turbulent models, Lagrangian stochastic trajectory model, P1radiative heat transfer model, the temperature field in the boiler was simulated by Fluent6.3. The predicted temperature results were found to be in good agreement with the measured data. The predicted temperature were analyzed to pcompare the applicability of various blending schemes. It shows that the price can be set to be objective function, and the enumerative algorithm shows good prediction accuracy.(5) The influence of the main operating parameters on the combustion in the boiler furnace was studied numerically. The optimized operating parameters were determined to be excess air coefficient of1.2, the primary air fraction of20%, the primary air temperature of433K, the secondary air temperature of620K and uniform powder casting.The research results of this paper have been applied to YiYang Power Generation Limited Company. The expert system can be used as the assistant decision-making for boiler operators to optimize coal blending, operation and combustion. The research results can also be applied to other power plants.