Industrial Innovations

Industrial Innovations

Development of a Data Mining-Based Predictive Model for Petrochemical Compressor Failures: A Case Study of Mahabad Petrochemical Complex

Document Type : Original Article

Authors
Farhad Chavoshini 1
Mahdi Yousefinejad Attari 1
Fahime Lotfian Delouyi 2
1 Department of Industrial Engineering, Bon.C., Islamic Azad University, Bonab, Iran.
2 Department of Mechanical Engineering, Facuity of Engineering, Zabol University, Zabol, Iran.
10.66224/jii.3.3.39
Abstract
A majority of production equipment worldwide today experiences failures in its working life cycle and must be repaired and maintained. Since production downtime incurs enormous financial implications, it is extremely crucial to find solutions for taking preventive action. Moreover, the use of sophisticated equipment—whose spare parts are expensive and sometimes unavailable due to political or technical limitations—has elevated the status of maintenance systems within companies, with it becoming a key determinant in productivity. A maintenance system not only reduces the final cost of production but also impacts the overall organizational value chain. Elements such as production speed and delivery time, product quality, production line reliability, and organizational flexibility all come under the influence of maintenance decisions. Therefore, in industries such as petrochemicals, where high-level machinery is utilized in the production systems, equipment reliability and maintenance gain greater significance. In this study, to analyze and predict equipment failures by using data mining methods, a 1304640 record dataset measured between December 20, 2021, and May 20, 2022, from 17 measurement sensors on one of the 3000 -series compressors of the Mahabad Petrochemical Complex were used. These data were examined, analyzed, and mined with Weka software. The statistical summaries and correlation relationships among the data were studied initially. Then, the reasons for equipment failure and the critical path leading to compressor shutdown were evaluated using the application of K-Means clustering, decision trees, and feature selection methods. Finally, based on the findings, recommendations were provided to focus maintenance efforts on critical sensors.
Keywords
Data Mining
Petrochemical Industry
Clustering
Decision Tree
Critical Root Cause
[1]    Shams H, Khorasgani GRH, Esfadan GA, Farsijani H, Shahmansouri A. Presentation of a Mathematical Model to Examine the Economic Advantages of Maintenance Strategies. System Engineering and Productivity. 2025;5(2). doi: https://doi.org/10.22034/sep.2025.20 51634.1264. [in Persian]
[2]     Gorji MA, Jamali MB. Comprehensive Model for Evaluating Maintenance and Repair Policies Based on Interval fuzzy Numbers. System Engineering and Productivity. 2022;2(1). doi: https://doi.org/10.22034/sep.2022.243404. [in Persian]
[3]     Izadikhah M, Garshasbi D. Using Data Mining and Three Decision Tree Algorithms to Optimize the Repair and Maintenance Process. Journal of New Researches In Mathematics. 2019;5(17):167–78. doi: https://sid.ir/paper/370289. [in Persian]
[4]    Wang L. Data Mining Technology Integrates Analysis in Petrochemical Testing Data. Procedia Computer Science. 2025;261:946–53. doi: https://doi.org/10.1016/j.procs.2025.04.482.
[5]    Giwa SO, Taziwa RT, Sharifpur M. Dependence of composition-based approaches on hybrid biodiesel fuel properties prediction using artificial neural network and random tree algorithms. Renewable Energy. 2023;128. doi: https://doi.org/10.1016/j.renene.2023.119324.
[6]    Mallia J, Francalanza E, Xuereb P, Borg M, Refalo P. Implementation of an intelligence-based framework for anomaly Implementation of an intelligence-based framework for anomaly detection on the demand-side of sustainable compressed air systems. Procedia Computer Science. 2024;232:1554–63. doi: https://doi.org/10.1016/j.procs.2024.01.153.
[7]    Namuduri S, Narayanan BN, Davuluru VSP, Burton L, Bhansali S. Deep learning methods for sensor based predictive maintenance and future perspectives for electrochemical sensors Journal of The Electrochemical Society. 2020;167(3). doi: https://doi.org/10.1149/1945-7111/ab67a8.
[8]    Salehian Z, Jahan A. Establishing a Reliability-based Maintenance methodology in a Gas Pressure Reduction System. System Engineering and Productivity. 2022;1(1). doi: 10.22034/sep.2022.243401. [in Persian]
[9]     Filz M-A, Langner JEB, Herrmann C, Thiede S. Data-driven failure mode and effect analysis (FMEA) to enhance maintenance planning. Computers in Industry. 2021;129:103451. doi: https://doi.org/10.1016/j.compind.2021.103451.
[10]     Peng Y, Lin J-R, Zhang J-P, Hu Z-Z. A hybrid data mining approach on BIM-based building operation and maintenance. Building and Environment. 2017;126:483–95. doi: https://doi.org/10.1016/j.buildenv.2017.09.030.
[11]     Atashhgar K, Saravany A. Failure prediction and diagnostic analysis for lathes using a hybrid approach including artificial neural netswork and decision making block. Internnational Journal of Indusrial Engineering & Prodction Management. 2016;27(2):143–56. [in Persian]
[12]     Karimi M, Afshr Kazemi  mohammad ali. Predicting failures and planning ATM maintenance by using data mining technique. Modern Research in Decision Making. 2016;1(3):113–30. [in Persian]
[13]     Bagherighadikolaeia SP, Ghousi R, Haeri A. A Data Mining Approach for Forecasting Failure Root Causes: A Case Study in an Automated Teller Machine (ATM) Manufacturing Company. Journal of Optimization in Industrial Engineering. 2020;13(2). doi: 10.22094/JOIE.2020.1863364.1630.
[14]     Rodriguez PC, Marti-Puig P, Caiafa CF, Serra-Serra M, Cusidó J, Solé-Casals J. Exploratory Analysis of SCADA Data from Wind Turbines Using the K-Means Clustering Algorithm for Predictive Maintenance Purposes. Machines Article. 2023;11(270). doi: https://doi.org/10.3390/machines11020270.
[15]     Cheng C-W, Yao H-Q, Wu T-C. Applying data mining techniques to analyze the causes of major occupational accidents in the petrochemical industry. Journal of Loss Prevention in the Process Industries. 2013;26(6):1269–78. doi: https://doi.org/10.1016/j.jlp.2013.07.002.
[16]     Arena S, Florian E, Zennaro I, Orrù PF, Sgarbossa F. A novel decision support system for managing predictive maintenance strategies based on machine learning approaches. Safety Science. 2022;146. doi: https://doi.org/10.1016/j.ssci.2021.105529.
[17]     Santos SP, Costa JAF. Application of multiple decision trees for condition monitoring in induction motors. In: 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence). 2008. p. 3736–41. doi: 10.1109/IJCNN.2008.4634334. 
[18]     Tran VT, Yang B-S, Oh M-S, Tan ACC. Fault diagnosis of induction motor based on decision trees and adaptive neuro-fuzzy inference. Expert Systems with Applications. 2009;36(2, Part 1):1840–9. doi: https://doi.org/10.1016/j.eswa.2007.12.010.
[19]     Allah Bukhsh Z, Saeed A, Stipanovic I, Doree AG. Predictive maintenance using tree-based classification techniques: A case of railway switches. Transportation Research Part C: Emerging Technologies. 2019;101:35–54. doi: https://doi.org/10.1016/j.trc.2019.02.001.
[20]     Burmeister N, Frederiksen RD, Hog E, Nielsen P. Exploration of Production Data for Predictive Maintenance of Industrial Equipment: A Case Study. IEEE Access. 2023;11:102025–37. doi: https://doi.org/10.1109/ACCESS.2023.3315842.
[21]     Kamel H. Artificial intelligence for predictive maintenance. In: Journal of Physics: Conference Series. 2022; 2299(1). doi: https://doi.org/10.1088/1742-6596/2299/1/012001
[22]     Falzone S, Kolodziej JR. Condition monitoring of a reciprocating compressor using wavelet transformation and support vector machines. Proceedings of the Annual Conference of the Prognostics and Health Management Society, PHM. 2017;39–45. doi: https://doi.org/10.36001/phmconf.2017.v9i1.2191.
[23]     Golmoradi M, Ebrahimi E, Javidan M. Compressor fault diagnosis based on SVM and GA. Vibroengineering Procedia. 2017;12:49–53. doi: https://doi.org/10.21595/vp.2017.18392.
[24]     QUINLAN J.R. Induction of Decision Trees. Machine Learning. 1986;1:81–106. doi: https://doi.org/10.1007/BF00116251. 
[25]     Golmoradi M, Ebrahimi E, Javidan M. Fault diagnosis of compressor based on decision tree and fuzzy inference system. Vibroengineering Procedia. 2017;12:54–60. doi: https://doi.org/10.21595/vp.2017.18398.
[26]     Elangovan M, Kumar SS, Bharathi Ganesh HB. Condition monitoring of a valve in a reciprocating compressor using machine learning approach. International Journal of Applied Engineering Research. 2015;10(13):33078–81. 
[27]     Aravinth S, Sugumaran V. Prediction of air compressor condition using vibration signals and machine learning algorithms. JVC/Journal of Vibration and Control. 2023;29(5–6):1342–51. doi: https://doi.org/10.1177/10775463211062330.
[28]     Nessaib K, Lakehal A. Multi Sources Information Fusion Based on Bayesian Network Method to Improve the Fault Prediction of Centrifugal Compressor. Strojnicky Casopis. 2022;72(1):109–24. doi: https://doi.org/10.2478/scjme-2022-0011. 
[29]     Salehi S, Sadedel M. Condition Monitoring of Reciprocating Compressors using Probabilistic Neural Network and Optimization with Genetic Algorithm. Iranian Journal of Mechanical Engineering Transactions of ISME. 2023;25(3). [in Persian]
[30]     Jamali M, Fargi Z, Rabiee M. Data mining of depressed patients to improve and evaluate its relationship with music. System Engineering and Productivity. 2021;1(4):49–73. doi: https://doi.org/10.22034/sep.2022.243416.
[31]      Kalmegh SR. Comparative Analysis of WEKA Data Mining Algorithm RandomForest, RandomTree and LADTree for Classification of Indigenous News Data. International Journal of Emerging Technology and Advanced Engineering. 2008;9001(1):507–17. 

  • Receive Date 19 November 2025
  • Revise Date 27 January 2026
  • Accept Date 27 January 2026