National Institute of Clean-and-Low-Carbon Energy

建立煤气化模型是研究气化过程的重要手段，为解决当前气化机理模型在变工况条件下存在残差较大，模型单次运行只能模拟一个工况，无法连续运行来表征气化炉长周期运行状态的问题，首先利用流程模拟软件Unisim，以国家能源投资集团榆林化工有限公司3000 t/d水煤浆气化炉为对象，建立了基于反应动力学的气化炉单工况机理模型，并基于ActiveX技术开发了气化炉的长周期机理模型，然后建立了机理模型并联BP神经网络模型的混合模型，对混合模型的模拟精度进行了验证与分析，最后对混合模型在气化指标预测和气化性能分析方面的应用进行了研究。结果表明：单工况机理模型对气化温度和合成气组成的模拟误差小于5%；长周期机理模型可以预测气化炉温的变化趋势，模拟值与运行值间的相关系数为0.822，但是模拟值的波动较大，气化温度模拟值的相对标准偏差(RSD)是运行值的3.8倍；气化混合模型的模拟精度比机理模型有着显著提高，其模拟值与运行值的变化趋势吻合良好，气化温度及合成气组成的误差小于2%；在更换工况后混合模型仍然有较高的准确性，模拟误差由之前的2%增大至4%以内。随着氧煤比从480增加至498 Nm3/m3，混合模型计算的气化温度由1160 ℃升高至1306 ℃，合成气中CO2摩尔分数由18.6%升高至19.9%，CO摩尔分数由42.6%降低至42%，H2摩尔分数由37.7%降低至36.8%，该变化趋势与机理模型保持一致，但是混合模型对机理模型有着明显的修正作用，其中对气化温度修正约38 ℃，对合成气中CO、CO2、H2的摩尔分数分别修正了−1%、−0.9%、1.5%。

The current gasification mechanism model has large residuals under variable working conditions and the model can only simulate one working condition in a single run and cannot be run continuously to characterize the long-cycle operation of gasifier. This paper firstly used the process simulation software Unisim to establish the long-cycle mechanism model of gasifier based on the reaction kinetics of 3000 t/d coal-water slurry gasifier in Yulin Chemical Co. of National Energy Investment Group. Then, a hybrid model of the mechanism model in parallel with the BP neural network model was developed, and the simulation accuracy of the hybrid model was verified and analyzed. Finally, the application of the hybrid model in gasification index prediction and gasification performance analysis was investigated. The application of the hybrid model for gasification index prediction and gasification performance analysis was investigated. The results show that the simulation error of the single-case mechanistic model for gasification temperature and syngas composition is less than 5%. The long-period mechanistic model can predict the variation trend of gasification furnace temperature, and the correlation coefficient between the simulated and operational values is 0.822, but the simulated values fluctuate greatly, and the relative standard deviation (RSD) of the simulated gasification temperature values is 3.8 times of the operational values. The simulation accuracy of the hybrid model for gasification is significantly better than that of the mechanistic model. The simulation accuracy of the gasification mixing model is significantly better than that of the mechanistic model, with the simulated values matching well with the trend of the operating values, and the errors of gasification temperature and syngas composition are less than 2%. The mixing model still has a high accuracy after the change of working conditions, and the simulation error increases from 2% to less than 4%. As the oxygen to coal ratio increases from 480 Nm3/m3 to 498 Nm3/m3, the gasification temperature calculated by the mixing model increases from 1160 ℃ to 1306 ℃, the CO2 content in the syngas increases from 18.6% to 19.9%, the CO content decreases from 42.6% to 42%, and the H2 content decreases from 37.7% to 36.8%, which are consistent with the mechanistic model. The hybrid model has a significant correction effect on the mechanistic model, in which the temperature is corrected by about 38 ℃, and X(CO), X(CO2) and X(H2) are corrected by −1, −0.9 and 1.5 percentage points respectively.

国家能源投资集团科技创新资助项目(GJNY-20-119,GJNY-22-28)

方薪晖，安海泉，刘臻，等. 基于混合模型的煤气化过程模拟研究[J]. 煤炭学报，2023，48(9)：3554−3561.

FANG Xinhui，AN Haiquan，LIU Zhen，et al. Simulation of coal gasification process based on hybrid model[J]. Journal of China Coal Society，2023，48(9)：3554−3561.