基于液化石油气的裂解过程建模与优化

Modeling and optimization of cracking process for LPG feedstock

随着乙烯裂解原料日趋多样化,液化石油气(LPG)的使用比重逐渐增大,而对于以LPG为原料的裂解过程建模与优化研究却较少,因此本文对以LPG为裂解原料的裂解过程进行建模与优化研究。利用裂解机理模拟LPG裂解过程,该过程同时考虑裂解原料的组分变化和操作条件变化对关键裂解产品收率的影响,然后根据实验设计原理产生适量仿真数据,建立了以原料组成与操作条件为输入、关键裂解产品收率为输出的PSOBP神经网络收率模型。通过模型验证,得到工业现场丙烯乙烯比值与本文模型预测得到的丙烯乙烯比的平均相对误差仅为2.142%。基于该模型利用PSO算法对LPG原料组成与操作条件进行优化,得到裂解产品总收益最大的LPG原料组成分布和操作条件,优化后的产品总效益为23.5524万元h^(-1)。

With ethylene cracking raw materials becoming increasingly diversified, the use of liquefied petroleum gas （LPG） is increasing gradually, but the research of modeling and optimization of cracking process for LPG feedstock is less. In this paper, the LPG was used as the cracking feedstock and the cracking progress was used for modeling and optimization. The LPG cracking progress was simulated by cracking mechanism, the progress took the changes of the cracking feedstock composition and operating conditions into aecout, and generated amount of simulation data. The PSOBP neural network modeling method was used to establish a model with the input of raw material composition and operating conditions, by the model validation, average absolute relative deviation of industrial PE ratio and the model predicted PE ratio is only 2.142%. Finally, LPG feedstock composition and operating conditions are optimized by PSO optimization algorithm and the maximum yield of decomposition products and the optimized operation conditions are obtained, the total optimized products value is 235.524 thousand yuan per hour.