高级检索

    梁朝林, 沈本贤, 刘纪昌, 陈晓龙. 用延迟焦化逐步回归法模型预测焦化产物的分布[J]. 华东理工大学学报(自然科学版), 2009, (2): 185-191.
    引用本文: 梁朝林, 沈本贤, 刘纪昌, 陈晓龙. 用延迟焦化逐步回归法模型预测焦化产物的分布[J]. 华东理工大学学报(自然科学版), 2009, (2): 185-191.
    Prediction for Product Distribution of Delayed-Coking Process by Stepwise Regression Model[J]. Journal of East China University of Science and Technology, 2009, (2): 185-191.
    Citation: Prediction for Product Distribution of Delayed-Coking Process by Stepwise Regression Model[J]. Journal of East China University of Science and Technology, 2009, (2): 185-191.

    用延迟焦化逐步回归法模型预测焦化产物的分布

    Prediction for Product Distribution of Delayed-Coking Process by Stepwise Regression Model

    • 摘要: 因原油产地、加工方法等不同,减压渣油延迟焦化反应产物的分布及组成差异较大。对17种减压渣油的物理化学性质,如密度、残炭、灰分、金属含量、组成(饱和分、芳香分、胶质、沥青质)等进行了全面分析,并在管式焦化反应装置上分别进行焦化反应实验。采用逐步回归分析法对11组数据进行分析,首先确定对焦化反应产物分布的各项指标(包括焦炭收率、≤200 ℃轻油收率、>200 ℃重油收率、气体收率、气体中氢气浓度及甲烷浓度)影响显著的减压渣油理化性质,然后用多元回归拟合得到各理化性质的回归系数,建立延迟焦化装置产品分布预测模型,并根据6种减压渣油的性质对焦化产物分布进行预测,结果表明该预测模型的相对误差均小于5%。与已有模型相比,该模型不但可以同时预测焦化气体、轻油、重油和焦炭产率,还可以预测焦化气体中氢气和甲烷含量,从而为预测渣油的焦化产物分布、优化调配焦化原料的合理加工提供较为全面的参考信息。

       

      Abstract: The coking product distribution and composition of the various vacuum residues are very different because these vacuum residues come from various oil fields and refined ways. Properties of 17 vacuum residues, such as density, residue carbon (electric furnace method), ash, metallic content, SARA (saturate, aromatics, resins, asphaltene) fractions, are analyzed. Coking reactive tests of the vacuum residues are respectively appraised by using the tubular coking reactor. According to the stepwise regression analysis on 11 series of data, the physical and chemical properties of the vacuum residues, which significantly affect coking product distribution(such as the yield of coke, ≤200 ℃ light oil, >200 ℃ heavy oil, and gas, and the content of hydrogen and methane in gas), are determined and their regression indexes are fitted. A model, which can predict the product distribution of the delayed coking units, is established. The model is tested by predicting the coking product distribution of six vacuum residues with different properties, and all the relative errors are within the permitted limit of 5%. Compared with other models existed, this model can predict not only the yield of the coking gas, light oil, heavy oil, and coke, but also the content of hydrogen and methane in coking gas. This model can provide overall information on coking products distribution and optimized coking feedstock process.

       

    /

    返回文章
    返回