Predicted vs. Proven: A Validated Machine Learning and Desirability Based Framework for Multi Response Optimization of EN31 Bearing Steel Turning Parameters
Authors
Department of Mechanical Engineering, College of Technology and Engineering (CTAE), Udaipur, Rajasthan, India (India)
Department of Mechanical Engineering, College of Technology and Engineering (CTAE), Udaipur, Rajasthan, India (India)
Department of Mechanical Engineering, College of Technology and Engineering (CTAE), Udaipur, Rajasthan, India (India)
Department of Mechanical Engineering, College of Technology and Engineering (CTAE), Udaipur, Rajasthan, India (India)
Article Information
DOI: 10.51584/IJRIAS.2026.11090122
Subject Category: manufacturing engineering
Volume/Issue: 11/9 | Page No: 1799-1818
Publication Timeline
Submitted: 2026-09-30
Accepted: 2026-10-05
Published: 2026-10-21
Abstract
Turning of EN31 alloy steel (equivalent to AISI 52100) a high-carbon, high-chromium bearing-grade alloy prized for its hardenability, wear resistance, and dimensional stability remains a challenging machining operation because of the steel's high hardness and abrasiveness. A growing body of literature has applied machine learning (ML) and desirability-function-based optimization to hardened-steel and AISI 52100/EN31 turning, typically reporting a single “optimal” parameter combination without checking whether that combination lies within the region actually sampled by the underlying experimental design. This paper addresses that gap directly. Using a 200-observation EN31 turning dataset (cutting speed, feed, depth of cut, coolant condition, and tool material as inputs; surface roughness, tool wear, cutting force, temperature, material removal rate, power, and chip reduction coefficient as responses), five regression algorithms linear regression, random forest, gradient boosting, support vector regression, and an artificial neural network were trained and compared via 5-fold cross-validation, achieving cross-validated R² up to 0.998 for the best model per response. A desirability-function-based multi-objective optimization over a continuous parameter grid identified a model-predicted optimum (cutting speed ≈ 389 rpm, feed ≈ 0.127 mm/rev, depth of cut = 0.25 mm, MQL cooling, carbide tool; desirability D = 0.520). Cross-checking this recommendation against the raw experimental design revealed that the tested cutting speeds were confined to six discrete levels with a 304 rpm gap (196–500 rpm) containing no trials, so the model-predicted speed of 389 rpm falls inside an unsampled region and is, strictly, an unvalidated interpolation. Repeating the same desirability analysis using only the 200 physically measured rows identified a data-validated optimum of 500 rpm cutting speed, 0.168 mm/rev feed, 0.35 mm depth of cut, MQL cooling, and a carbide tool (D = 0.497), which trades a modest surface-finish/tool-wear penalty for substantially higher measured MRR (23.5 vs. 12.7 mm³/s) than the unvalidated prediction. To the best of our knowledge, the finding that a continuous ML-based optimizer can recommend an operationally untested parameter combination even when its underlying response models are locally accurate has not been reported in prior EN31/AISI 52100 ML-optimization studies, and constitutes the principal methodological contribution of this work. The findings corroborate and extend prior EN31-specific and hardened-steel machining literature and offer a validation protocol directly transferable to other ML-based machining-optimization studies. A confirmatory turning-trial protocol, a discussion of how tree-based models behave across unsampled gaps, and constraint-aware (convex-hull and density-based) strategies for automated optimization workflows are also presented.
Keywords
EN31 alloy steel • AISI 52100 • Machine learning • Desirability function • Multi-objective optimization • Model validation • Surface roughness • Tool wear • Minimum quantity lubrication
Downloads
References
1. Fang, W., Lai, T., Song, Z., Cai, L Wang, J., Han, Y., Lyu, B.: Artificial intelligence for process modeling and optimization in precision machining: a review. Int. J. Adv. Manuf. Technol. (2026). [Google Scholar] [Crossref]
2. https://doi.org/10.1007/s00170-026-18834-8 [Google Scholar] [Crossref]
3. Jiang, Y., Huang, J., Liu, K., Zhao, D., Liu, H., Wang, Y.: Intelligent machining technology for online monitoring and control of tool condition: a review. Front. Mech. Eng. 20(6), 50 (2025). [Google Scholar] [Crossref]
4. https://doi.org/10.1007/s11465-025-0866-y [Google Scholar] [Crossref]
5. Ko, J.H., Yin, C.: A review of artificial intelligence application for machining surface quality prediction: from key factors to model development. J. Intell. Manuf. 37, 775–798 (2026). [Google Scholar] [Crossref]
6. https://doi.org/10.1007/s10845-025-02571-y [Google Scholar] [Crossref]
7. Almomani, O., Venkatesh, B., Chaudhary, S.P., Mishra, A., Sujai, S., Juneja, S., Pradhan, P., Venkatesan, S.P., Bhowmik, A., Tamene, Y.: Intelligent tool wear monitoring using XGBoost, SVR, and DNN models in NMQL environment. Sci. Rep. 16, 10030 (2026). https://doi.org/10.1038/s41598-026-40968-8 [Google Scholar] [Crossref]
8. Kosarac, A., Mladjenovic, C., Zeljkovic, M., Tabakovic, S., Knezev, M.: Neural-network-based approaches for optimization of machining parameters using small dataset. Materials 15(3), 700 (2022). https://doi.org/10.3390/ma15030700 [Google Scholar] [Crossref]
9. Batish, A., Bhattacharya, A., Kaur, M., Cheema, M.S.: Hard turning: parametric optimization using genetic algorithm for rough/finish machining and study of surface morphology. J. Mech. Sci. Technol. 28(5), 1629–1640 (2014). https://doi.org/10.1007/s12206-014-0308-y [Google Scholar] [Crossref]
10. Misra, S., Kumar, Y., Paul, G.: Study on effects of flow rate of coolant during CO2-assisted MQ turning of EN31 steel. J. Inst. Eng. India Ser. C 105(1). https://doi.org/10.1007/s40032-023-01011-1 [Google Scholar] [Crossref]
11. Kumar, R., Rafighi, M., Özdemir, M., Şahinoğlu, A., Kulshreshta, A., Singh, J., Singh, S., Prakash, C., Bhowmik, A.: Modeling and optimization of hard turning: predictive analysis of surface roughness and cutting forces in AISI 52100 steel using machine learning. Int. J. Interact. Des. Manuf. (IJIDeM) 19, 5749–5778 (2024). https://doi.org/10.1007/s12008-024-02166-6 [Google Scholar] [Crossref]
12. Mane, S., Patil, R.B., Al-Dahidi, S.: Predictive modeling of surface roughness and cutting temperature using response surface methodology and artificial neural network in hard turning of AISI 52100 steel with minimal cutting fluid application. Machines 13(4), 266 (2025). [Google Scholar] [Crossref]
13. https://doi.org/10.3390/machines13040266 [Google Scholar] [Crossref]
14. Khatai, S., Kumar, R., Panda, A., Sahoo, A.K.: WASPAS based multi response optimization in hard turning of AISI 52100 steel under ZnO nanofluid assisted dual nozzle pulse-MQL environment. Appl. Sci. 13(18), 10062 (2023). https://doi.org/10.3390/app131810062 [Google Scholar] [Crossref]
15. Vyasa, R., Brahmbhatt, P., Sonawane, C., Lakkimsetty, N.R., Pavithra, G.: Investigation on the multi- objective optimization of machining parameters and prediction for EN series materials. Eng. Technol. Appl. Sci. Res. 14(5), 16427–16437 (2024). https://doi.org/10.48084/etasr.7953 [Google Scholar] [Crossref]
16. Abhang, L. B., & Hameedullah, M. (2011). Modeling and analysis for surface roughness in machining EN-31steel using response surface methodology. International Journal of Applied Engineering Research. [Google Scholar] [Crossref]
17. Abhang, L. B., & Hameedullah, M. (2012). Optimization of machining parameters in steel turning operation by Taguchi method. Procedia Engineering, 38, 1686–1691. [Google Scholar] [Crossref]
18. https://doi.org/10.1016/j.proeng.2012.06.204 [Google Scholar] [Crossref]
19. Abhang, L. B., & Hameedullah, M. (2014). Parametric investigation of turning process on EN-31 steel. Procedia Engineering. [Google Scholar] [Crossref]
20. Abhang, L. B., & Hameedullah, M. (2015). Simultaneous optimization of multiple quality characteristics in turning EN-31 steel. Procedia Materials Science. [Google Scholar] [Crossref]
21. Chaudhari, T. M., & Karunamurthy, L. (2017). Determining the effect of coated and uncoated tool insert during turning of EN31 steel. International Journal of Mechanical Engineering and Technology, 8(11). [Google Scholar] [Crossref]
22. Rajaparthiban, K., et al. (2018). Experimental investigation of cutting parameters influence on tool wear and surface roughness in turning of EN-31 steel using carbide insert in CNC turning center. International Journal of Engineering Research & Technology (IJERT). [Google Scholar] [Crossref]
23. Machining of EN31 Steel Using Carbide Insert – A Statistical Approach. (2020). Materials Today: Proceedings. https://doi.org/10.1016/j.matpr.2020.03.386 [Google Scholar] [Crossref]
24. Experimental Investigation and ANFIS-Based Modelling During Machining of EN31 Alloy Steel. (2020). Materials, 13(15), 3351. https://doi.org/10.3390/ma13153351 [Google Scholar] [Crossref]
25. Sahu, S., Sivarajan, R., & Padmanabhan, K. K. (2021). Optimization of machining parameters in turning of EN31 steel with TiAlN coated cutting tool. Materials Today: Proceedings. [Google Scholar] [Crossref]
26. Singh, G., et al. (2021). Experimental investigations and optimization of machining performance during turning of EN-31 steel using TOPSIS. Materials Today: Proceedings. [Google Scholar] [Crossref]
27. Multi-response optimization of process parameters using desirability function analysis during machining of EN31 steel under different machining environments. (2021). Materials Today: Proceedings. [Google Scholar] [Crossref]
28. Multi-response optimization of process variables in MQL-assisted face milling of EN31 alloy steel using grey relational analysis. (2021). Materials Today: Proceedings. [Google Scholar] [Crossref]
29. Sivarajan, R., et al. (2022). Prediction of surface roughness in hard machining of EN31 steel with TiAlN coated cutting tool using fuzzy logic. Materials Today: Proceedings. [Google Scholar] [Crossref]
30. https://doi.org/10.1016/j.matpr.2022.04.161 [Google Scholar] [Crossref]
31. Khatai, S., et al. (2022). Investigation on tool wear and chip morphology in hard turning of EN31 steel using AlTiN-PVD coated carbide cutting tool. Materials Today: Proceedings. [Google Scholar] [Crossref]
32. Jagadeesan, S., & Ramya Devi, D. (2022). Comparison of novel TiN coated and uncoated carbide tool in CNC dry turning of EN31 to analyse chip profile. Materials Today: Proceedings. [Google Scholar] [Crossref]
33. https://doi.org/10.1016/j.matpr.2022.07.235 [Google Scholar] [Crossref]
34. Verma, R., & Srivastava, A. (2022). Investigation about machining issues in turning process of EN-31 steel. Materials Today: Proceedings. https://doi.org/10.1016/j.matpr.2021.10.238 [Google Scholar] [Crossref]
35. Experimental investigation of tool wear in face milling of EN-31 steel under different machining environments. (2022). Materials Today: Proceedings. https://doi.org/10.1016/j.matpr.2023.04.473 [Google Scholar] [Crossref]
36. Vamsidhar, et al. (2023). Comparison of novel TiN coated and uncoated carbide tool in CNC wet turning of EN31 to maximize material removal rate. AIP Conference Proceedings. [Google Scholar] [Crossref]
37. https://doi.org/10.1063/5.0173215 [Google Scholar] [Crossref]
38. Synergetic effect of cryogenic treatment and laser surface texturing of tungsten carbide cutting inserts on machinability of EN31 steel. (2025). Journal of Engineering Research. [Google Scholar] [Crossref]
39. Friedman, J.H. (2001). Greedy function approximation: a gradient boosting machine. The Annals of Statistics, 29(5), 1189–1232. [Google Scholar] [Crossref]
40. Barber, C.B., Dobkin, D.P., & Huhdanpaa, H. (1996). The Quickhull algorithm for convex hulls. ACM Transactions on Mathematical Software, 22(4), 469–483. [Google Scholar] [Crossref]