GEP-based Modeling for Predicting Sponge Iron Metallization in Persian Direct Reduction (PERED) Method

Subject Areas : Extractive metallurgy

Mehdi Firouzi ¹ , Mojtaba Sadeghi Gogheri ² , Mojtaba Firouzi ³ , Masoud Kasiri-Asgarani ⁴ , Hamid Reza Bakhsheshi-Rad ⁵

1 - Research & Development, Sirjan Jahan Steel Complex (SJSCO), Sirjan, Iran; Baft Steel Complex, Baft, Iran
2 - Research & Development, Sirjan Jahan Steel Complex (SJSCO), Sirjan, Iran; Advanced Materials Research Center, Department of Materials Engineering, Najafabad Branch, Islamic Azad University, Najafabad, Iran
3 - Department of Computer Engineering, Qom university of Technology, Qom, Iran
4 - Advanced Materials Research Center, Department of Materials Engineering, Najafabad Branch, Islamic Azad University, Najafabad, Iran
5 - Advanced Materials Research Center, Department of Materials Engineering, Najafabad Branch, Islamic Azad University, Najafabad, Iran

Received: 2021-07-25 Accepted : 2022-09-02 Published : 2021-11-01

Keywords: Mathematical Modeling, PERED, Direct Reduced Iron (DRI), Gene Expression Programming (GEP),

Abstract :

The Persian direct reduction method (PERED) is a suitable method for producing sponge iron on an industrial scale. The challenge of all sponge iron production plants is to supply sponge iron with suitable metallization to steel factories. Accordingly, determining and adjusting the various parameters affecting metallization in each plant is necessary to produce the appropriate amount and quality of sponge iron. In this study, first, the effects of output rate, process flow, water- steam flow rate, bustle temperature, bustle CH4 level, CO2 reform, average pellet size (PIDa), pellet strength (CCS), process gas water temperature, and furnace bed average temperature on spongy iron metallization were investigated. Then, an attempt was made to model the sponge iron grade produced by the PERED method using the Gene Expression Programming (GEP) software. To carry out modeling, data on the affecting variables of metallization were collected for 58 days. The best R2 values for the training and testing sets were 0.974 and 0.27 with a low error rate for both (0.047 and 0.376 in RMSE and 0.001 and 0.141 in MSE, respectively. The results of the sensitivity test indicated that CO2 reform gas, bustle CH4 level, and average pellet size had the most significant effect on metallization.

References:

 

 

References

[1] S. Baig, B. Murray: Duke Univ., 2016.

[2] F.M. Mohsenzadeh, H. Payab, M.A. Abdoli, Z. Abedi, Ekoloji, 27(2018) 959–967.

[3] M. Kazemi, M. Saffari pour, D. Sichen: Metallurgical and Materials Transactions B., 2017, vol. 48B, pp. 1114–1122.

[4] G. KIM, P. C. Pistorius: Metallurgical and Materials Transactions B., 2009, vol. 40B, pp. 17–24.

[5] E. A. Mousa, A. Babich, D. Senk: Metallurgical and Materials Transactions B., 2013, vol. 45, pp. 617–628.

[6] I. Sohn, R. J. Fruehan: Metallurgical and Materials Transactions B., 2013, vol. 36B, pp. 605–612.

[7]  F. M. Mohsenzadeh, H. Payab, M. A. Abdoli, Z. Abedi: Ekoloji., 2017, Vol. 27, pp. 959–967.

[8]  P. Cavaliere: Ironmaking and Steelmaking Pro., 2019, pp. 419–484.

[9]  B. Voelker, H. Michishita, T. Wright: Google Patents., 2020.

[10] B. Anameric, S. K. Kawatra: Miner. Process. Extr. Metall. Rev., 2007, vol. 207, pp. 59–116.

[11]  F. O. Boechat, R. M. de Carvalho, L. M. Tavares: KONA Powder Part., 2018, Vol. 35, pp. 216–225.

[12] M. Khosravi, M. Zeraati: Particle Science and Tech., 2019, Vol. 5, pp. 145-159

[13]  A. D. Mehr: Hydrol., 2018, Vol. 563, pp. 669–678.

[14]  H. Ebrahimzade, G. R. Khayati, M. Schaffie: Environ. Chem. Eng., 2018, Vol. 6, pp. 3999–4007.

[15]  L. ZHANG, T. Qiu-hua, C. A. Floudas, D. Ming-xing,: Mach. Des. Manuf., 2017, Vol. 2, pp. 51-60.

[16]  A. K. Sista Kameshwar, W. Qin: Mycology., 2020, Vol. 11, pp. 22-37.