Numerical Simulation And Operation Monitoring Of Direct Air Cooling System For Thermal Power Units

Direct air-cooled units have been widely used in northern China due to their excellent water-saving capabilities.However,to fully utilize environmental wind,the air-cooled condenser was arranged in the open air and the fan array was laid out in clusters,which resulted in the dual effects of environmental factors and clustering on the fan equipment.Moreover,the large number of fans and harsh operating environment made it difficult to monitor the operation of air-cooled fans.In this paper,a combined approach of CFD numerical simulation and intelligent algorithms was applied to monitor the operation of axial fans in direct air-cooled units.Firstly,the equipment layout and operating parameters of the target unit were introduced.Considering the geographical location and climate environment of the unit,nine different environmental conditions were selected for numerical simulation.Then,the numerical simulation model of the air-cooled island and main equipment of the target unit was established.The calculation method,control equations,boundary conditions,and corresponding method for calculating unit back pressure were described.The model was verified for grid independence and accuracy,and it could accurately reflect the actual operation of the air-cooled island in direct air-cooled units.Secondly,based on the numerical simulation model of the air-cooled island,the operating status of the air-cooled island under different speeds,directions,and environmental temperatures was compared and analyzed.The airflow and temperature distribution of each air-cooled unit were greatly affected by the environmental wind direction,and the effect of the backflow on the windward-side air-cooled unit gradually expanded as the wind speed increased.Then,by calculating the temperature field,heat exchange,and other parameters of the unit under different environmental conditions,the performance difference of the two units under the influence of environmental wind was obtained.Furthermore,the K-means algorithm was applied to reasonably classify the aircooled fans of units#1 and#2 based on the numerical simulation results and actual historical operating data.Finally,based on the reasonable classification of air-cooled fan clusters,a convolutional LSTM neural network was used to establish a prediction model for the operating status parameters of each cluster fan.The optimal model parameters were confirmed through parameter tuning,and the final model achieved good prediction performance in different months.Moreover,a fan operating status monitoring strategy was proposed based on the prediction model,which could efficiently and reliably alarm the fault status of each fan cluster.