Journal Home Online First Current Issue Archive For Authors Journal Information 中文版

Engineering >> 2019, Volume 5, Issue 6 doi: 10.1016/j.eng.2019.02.013

Artificial Intelligence in Steam Cracking Modeling: A Deep Learning Algorithm for Detailed Effluent Prediction

a Laboratory for Chemical Technology, Department of Materials, Textiles and Chemical Engineering, Ghent University, Gent 9052, Belgium
b SynBioC Research Group, Department of Green Chemistry and Technology, Faculty of Bioscience Engineering, Ghent University, Gent 9000,

Received:2018-09-14 Revised:2019-01-07 Accepted: 2019-02-11 Available online:2019-10-09

Next Previous


Chemical processes can benefit tremendously from fast and accurate effluent composition prediction for plant design, control, and optimization. The Industry 4.0 revolution claims that by introducing machine learning into these fields, substantial economic and environmental gains can be achieved. The bottleneck for high-frequency optimization and process control is often the time necessary to perform the required detailed analyses of, for example, feed and product. To resolve these issues, a framework of four deep learning artificial neural networks (DL ANNs) has been developed for the largest chemicals production process—steam cracking. The proposed methodology allows both a detailed characterization of a naphtha feedstock and a detailed composition of the steam cracker effluent to be determined, based on a limited number of commercial naphtha indices and rapidly accessible process characteristics. The detailed characterization of a naphtha is predicted from three points on the boiling curve and PIONA (paraffin, isoparaffin, olefin, naphthene, and aromatics) characterization. If unavailable, the boiling points are also estimated. Even with estimated boiling points, the developed DL ANN outperforms several established methods such as maximization of Shannon entropy and traditional ANNs. For feedstock reconstruction, a mean absolute error (MAE) of 0.3 wt% is achieved on the test set, while the MAE of the effluent prediction is 0.1 wt%. When combining all networks—using the output of the previous as input to the next—the effluent MAE increases to 0.19 wt%. In addition to the high accuracy of the networks, a major benefit is the negligible computational cost required to obtain the predictions. On a standard Intel i7 processor, predictions are made in the order of milliseconds. Commercial software such as COILSIM1D performs slightly better in terms of accuracy, but the required central processing unit time per reaction is in the order of seconds. This tremendous speed-up and minimal accuracy loss make the presented framework highly suitable for the continuous monitoring of difficult-to-access process parameters and for the envisioned, high-frequency real-time optimization (RTO) strategy or process control. Nevertheless, the lack of a fundamental basis implies that fundamental understanding is almost completely lost, which is not always well-accepted by the engineering community. In addition, the performance of the developed networks drops significantly for naphthas that are highly dissimilar to those in the training set.



Fig. 1

Fig. 2

Fig. 3

Fig. 4

Fig. 5

Fig. 6

Fig. 7

Fig. 8

Fig. 9

Fig. 10

Fig. 11

Fig. 12

Fig. 13

Fig. 14


[1]  Amghizar I, Vandewalle LA, Van Geem KM, Marin GB. New trends in olefin production. Engineering 2017;3(2):171–8. link1

[2]  Campbell M, Hoane AJ Jr, Hsu F. Deep blue. Artif Intell 2002;134(1–2):57–83. link1

[3]  Gibney E. Google AI algorithm masters ancient game of Go. Nature 2016;529 (7587):445–6. link1

[4]  Chowdhury GG. Natural language processing. Annu Rev Inf Sci Technol 2003;37(1):51–89.

[5]  Yin W, Kann K, Yu M, Schütze H. Comparative study of CNN and RNN for natural language processing. 2017. arXiv:1702.01923. link1

[6]  Bojarski M, Del Testa D, Dworakowski D, Firner B, Flepp B, Goyal P, et al. End to end learning for self-driving cars. 2016. arXiv:1604.07316. link1

[7]  Li D, Gao H. A hardware platform framework for an intelligent vehicle based on a driving brain. Engineering 2018;4(4):464–70. link1

[8]  Maltarollo VG, Honório KM, Ferreira da Silva AB. Applications of artificial neural networks in chemical problems. In: Suzuki K, editor. Artificial neural networks—architectures and applications. Rijeka: InTech; 2013. p. 203–23. link1

[9]  Day CP. Robotics in industry—their role in intelligent manufacturing. Engineering 2018;4(4):440–5. link1

[10]  Brettel M, Friederichsen N, Keller M, Rosenberg M. How virtualization, decentralization and network building change the manufacturing landscape: an Industry 4.0 perspective. Int J Inf Commun Eng 2014;8(1):37–44. link1

[11]  Lasi H, Fettke P, Kemper HG, Feld T, Hoffmann M. Industry 4.0. Bus Inf Syst Eng 2014;6(4):239–42. link1

[12]  Zhong RY, Xun X, Klotz E, Newman ST. Intelligent manufacturing in the context of Industry 4.0: a review. Engineering 2017;3(5):616–30. link1

[13]  Zhou K, Liu T, Zhou L. Industry 4.0: towards future industrial opportunities and challenges. In: Proceeding of the 12th International Conference on Fuzzy Systems and Knowledge Discovery (FSKD); 2015 Aug 15–17; Zhejiang, China. New York: IEEE; 2015. p. 2147–52. link1

[14]  Yuan Z, Qin W, Zhao J. Smart manufacturing for the oil refining and petrochemical industry. Engineering 2017;3(2):179–82. link1

[15]  Zhang L, Mao H, Liu L, Du J, Gani R. A machine learning based computer-aided molecular design/screening methodology for fragrance molecules. Comput Chem Eng 2018;115:295–308. link1

[16]  Bajorath J. Computer-aided drug discovery. F1000Res 2015;4:630. link1

[17]  Peplow M. Organic synthesis: the robo-chemist. Nature 2014;512(7512):20–2. link1

[18]  Coley CW, Rogers L, Green WH, Jensen KF. SCScore: synthetic complexity learned from a reaction corpus. J Chem Inf Model 2018;58(2):252–61. link1

[19]  Goh GB, Hodas NO, Vishnu A. Deep learning for computational chemistry. J Comput Chem 2017;38(16):1291–307.

[20]  Sedghi S, Huang B. Real-time assessment and diagnosis of process operating performance. Engineering 2017;3(2):214–9. link1

[21]  Bogle IDL. A perspective on smart process manufacturing research challenges for process systems engineers. Engineering 2017;3(2):161–5. link1

[22]  Castillo PAC, Castro PM, Mahalec V. Global optimization of nonlinear blendscheduling problems. Engineering 2017;3(2):188–201. link1

[23]  Van Geem KM, Pyl SP, Reyniers MF, Vercammen J, Beens J, Marin GB. On-line analysis of complex hydrocarbon mixtures using comprehensive twodimensional gas chromatography. J Chromatogr A 2010;1217(43):6623–33. link1

[24]  Van Geem KM, Marin G, Muñoz Gandarillas A, Zhang Y, Du W, Qian F. Plant wide optimization for high value added products: a steam cracking case study [presentation]. In: The 30th Ethylene Producers’ Conference; 2018 Apr 22–26; Orlando, FL, USA; 2018. link1

[25]  Hudebine D, Verstraete JJ. Molecular reconstruction of LCO gasoils from overall petroleum analyses. Chem Eng Sci 2004;59(22–23):4755–63. link1

[26]  Verstraete JJ, Revellin N, Dulot H, Hudebine D. Molecular reconstruction of vacuum gasoils. ACS Div Fuel Chem 2004;49(1):20–1. link1

[27]  Van Geem KM, Hudebine D, Reyniers MF, Wahl F, Verstraete JJ, Marin GB. Molecular reconstruction of naphtha steam cracking feedstocks based on commercial indices. Comput Chem Eng 2007;31(9):1020–34. link1

[28]  Ranzi E, Dente M, Goldaniga A, Bozzano G, Faravelli T. Lumping procedures in detailed kinetic modeling of gasification, pyrolysis, partial oxidation and combustion of hydrocarbon mixtures. Prog Energy Combust Sci 2001;27 (1):99–139. link1

[29]  Sadrameli SM. Thermal/catalytic cracking of hydrocarbons for the production of olefins: a state-of-the-art review I: thermal cracking review. Fuel 2015;140:102–15.

[30]  Van Geem KM, Reyniers MF, Marin GB. Challenges of modeling steam cracking of heavy feedstocks. Oil Gas Sci Technol 2008;63(1):79–94. link1

[31]  Van Geem KM, Reyniers MF, Marin GB. Two severity indices for scale-up of steam cracking coils. Ind Eng Chem Res 2005;44(10):3402–11. link1

[32]  Van Geem KM, Zˇajdlík R, Reyniers MF, Marin GB. Dimensional analysis for scaling up and down steam cracking coils. Chem Eng J 2007;134(1–3):3–10. link1

[33]  Van Geem KM, Reyniers MF, Pyl S, Marin GB, Zhou Z. Effect of operating conditions and feedstock composition on run lengths of steam cracking coils [presentation]. In: AIChE Spring National Meeting; 2009 Apr 26–30; Tampa, FL, USA; 2009. link1

[34]  Green WH Jr. Predictive kinetics: a new approach for the 21st century. Adv Chem Eng 2007;32:1–50. link1

[35]  Van de Vijver R, Vandewiele NM, Bhoorasingh PL, Slakman BL, Seyedzadeh Khanshan F, Carstensen HH, et al. Automatic mechanism and kinetic model generation for gas- and solution-phase processes: a perspective on best practices, recent advances, and future challenges. Int J Chem Kinet 2015;47 (4):199–231.

[36]  Hopfield JJ. Artificial neural networks. IEEE Circuits Device 1988;4(5):3–10. link1

[37]  Mahanta J. Introduction to neural networks, advantages and applications [Internet]. Deeplearningtrack; [updated 2017 Jul 9; cited 2018 Aug 3]. Available form: Introduction-to-NEURAL-NETWORKS-Advantages-and-Applications. link1

[38]  Pyl SP, Van Geem KM, Reyniers MF, Marin GB. Molecular reconstruction of complex hydrocarbon mixtures: an application of principal component analysis. AIChE J 2010;56(12):3174–88. link1

[39]  Niaei A, Towfighi J, Khataee AR, Rostamizadeh K. The use of ANN and the mathematical model for prediction of the main product yields in the thermal cracking of naphtha. Pet Sci Technol 2007;25(8):967–82. link1

[40]  Sedighi M, Keyvanloo K, Towfighi J. Modeling of thermal cracking of heavy liquid hydrocarbon: application of kinetic modeling, artificial neural network, and neuro-fuzzy models. Ind Eng Chem Res 2011;50(3):1536–47. link1

[41]  Ghadrdan M, Mehdizadeh H, Boozarjomehry RB, Darian JT. On the introduction of a qualitative variable to the neural network for reactor modeling: feed type. Ind Eng Chem Res 2009;48(8):3820–4. link1

[42]  Szegedy C, Toshev A, Erhan D. Deep neural networks for object detection. In: Proceedings of the 27th Annual Conference on Neural Information Processing Systems; 2013 Dec 5–8; Lake Tahoe, NV, USA. San Diego: NIPS; 2013. p. 2553–61.

[43]  Seif G. I’ll tell you why Deep Learning is popular in demand [Internet]. Medium; [cited 2018 Aug 3]. Available from: link1

[44]  Shamsuddin SM, Ibrahim AO, Ramadhena C. Weight changes for learning mechanisms in two-term back-propagation network. In: Suzuki K, editor. Artificial neural networks—architectures and applications. Rijeka: InTech; 2013. p. 53–82. link1

[45]  Rumelhart DE, Hinton GE, Williams RJ. Learning representations by backpropagating errors. Nature 1986;323(6088):533–6. link1

[46]  Tetko IV, Livingstone DJ, Luik AI. Neural network studies. 1. Comparison of overfitting and overtraining. J Chem Inf Comput Sci 1995;35(5):826–33. link1

[47]  Hornik K, Stinchcombe M, White H. Multilayer feedforward networks are universal approximators. Neural Netw 1989;2(5):359–66. link1

[48]  Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks. In: Proceedings of the 27th Annual Conference on Neural Information Processing Systems; 2012 Dec 3–6; Lake Tahoe, NV, USA. San Diego: NIPS; 2012. p. 1097–105. link1

[49]  Srivastava N, Hinton G, Krizhevsky A, Sutskever I, Salakhutdinov R. Dropout: a simple way to prevent neural networks from overfitting. J Mach Learn Res 2014;15(1):1929–58. link1

[50]  Ng AY. Feature selection, L1 vs. L2 regularization, and rotational invariance. In: Proceedings of the 21th International Conference on Machine Learning; 2004 Jul 4–8; Banff, AB, Canada. New York: ACM; 2004. p. 78. link1

[51]  Chollet F. Keras: the Python deep learning library [Internet]. [cited 2018 Aug 3]. Available from: link1

[52]  Abadi M, Barham P, Chen J, Chen Z, Davis A, Dean J, et al. Tensorflow: a system for large-scale machine learning. In: Proceedings of the 12th USENIX conference on Operating Systems Design and Implementation; 2016 Nov 2–4; Savannah, GA, USA. Berkeley: USENIX Association; 2016. p. 265–83.

[53]  Jolliffe IT. Principal component analysis. New York: Springer; 2011.

[54]  De Maesschalck R, Jouan-Rimbaud D, Massart DL. The Mahalanobis distance. Chemom Intell Lab Syst 2000;50(1):1–18. link1

[55]  Mahalanobis PC. On the generalized distance in statistics. Proc Natl Inst Sci India 1936;2:49–55. link1

[56]  Van Geem KM, Reyniers MF, Marin G. Taking optimal advantage of feedstock flexibility with COILSIM1D. In: Proceedings of 2008 AIChE Spring Meeting and Global Congress on Process Safety; 2008 Apr 6–10; New Orleans, LA, USA. New York: American Institute of Chemical Engineers; 2008. p. 391–404. link1

[57]  Vervust A, Amghizar I, Munoz A, Van Geem KM, Marin G. Full furnace simulations and optimization with COILSIM1D. In: Proceedings of 2016 Spring Meeting and 12th Global Congress on Process Safety; 2016 Apr 10–14; Houston, TX, USA. New York: American Institute of Chemical Engineers; 2016. p. 21. link1

[58]  Paraskevas PD, Sabbe MK, Reyniers MF, Marin GB, Papayannakos NG. Group additive kinetic modeling for carbon-centered radical addition to oxygenates and b-scission of oxygenates. AIChE J 2016;62(3):802–14. link1

[59]  Saeys M, Reyniers MF, Marin GB, Van Speybroeck V, Waroquier M. Ab initio group contribution method for activation energies for radical additions. AIChE J 2004;50(2):426–44. link1

[60]  Van de Vijver R, Sabbe MK, Reyniers MF, Van Geem KM, Marin GB. Ab initio derived group additivity model for intramolecular hydrogen abstraction reactions. Phys Chem Chem Phys 2018;20(16):10877–94.

[61]  Davis AC, Francisco JS. Ab initio study of hydrogen migration across n-alkyl radicals. J Phys Chem A 2011;115(14):2966–77. link1

[62]  Gao CW, Allen JW, Green WH, West RH. Reaction mechanism generator: automatic construction of chemical kinetic mechanisms. Comput Phys Commun 2016;203:212–25. link1

[63]  Merchant SS. Molecules to engines: combustion chemistry of alcohols and their applications to advanced engines [dissertation]. Cambridge: Massachusetts Institute of Technology; 2015. link1

[64]  Fannin G. Distillation process analyser with ASTM 86 compliance. Petro Industry News 2013 Aug/Sep;14(4):40.

[65]  Ferris AM, Rothamer DA. Methodology for the experimental measurement of vapor–liquid equilibrium distillation curves using a modified ASTM D86 setup. Fuel 2016;182:467–79. link1

Related Research