Machine learning‑based prediction of thermal performance in pipe flows equipped with conical turbulators using CFD‑generated data
Journal of Thermal Analysis and Calorimetry, cilt.1, sa.1, ss.1-27, 2026 (SCI-Expanded, Scopus)
- Yayın Türü: Makale / Tam Makale
- Cilt numarası: 1 Sayı: 1
- Basım Tarihi: 2026
- Doi Numarası: 10.1007/s10973-026-16180-1
- Dergi Adı: Journal of Thermal Analysis and Calorimetry
- Derginin Tarandığı İndeksler: Academic Search Ultimate (EBSCO), Engineering Source (EBSCO), Scopus, Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest), Aerospace Database, Science Citation Index Expanded (SCI-EXPANDED), Chemical Abstracts Core, Chimica, Compendex, Index Islamicus, INSPEC
- Sayfa Sayıları: ss.1-27
- Erciyes Üniversitesi Adresli: Evet
Özet
This study presents a machine learning-based prediction framework for evaluating the thermo-hydraulic performance of turbulent pipe flows equipped with conical turbulators using CFD-generated data. Three different turbulator configurations, namely converging, diverging, and converging–diverging geometries, were numerically investigated under various pitch-to-diameter (P/D) ratios and rotation angles over a wide range of Reynolds numbers. The obtained CFD dataset was used to predict the Nusselt number (Nu), friction factor (f), and thermal performance criterion (THP) through Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) models. The input parameters of the models consisted of Reynolds number, turbulator geometry type, P/D ratio, and rotation angle. The prediction performances were evaluated using the coefficient of determination (), root-mean-square error (RMSE), mean absolute error (MAE), and mean absolute percentage error (MAPE). The ANN model produced acceptable estimations with values generally ranging between 0.88 and 0.92 depending on the flow configuration and target parameter. However, larger deviations were observed in several operating conditions, particularly for high P/D ratios. In contrast, the ANFIS model demonstrated excellent agreement with the CFD data for all cases, yielding values close to 1.000 and MAPE values below 0.31%. In addition, SHAP-based explainability analysis revealed that Reynolds number and turbulator geometry were the most influential parameters affecting the thermo-hydraulic behavior of the system. The results demonstrate that ANFIS provides a highly reliable and computationally efficient alternative to repeated CFD simulations for rapid prediction of thermal performance in enhanced heat transfer applications involving conical turbulators.