Performance Estimation of Hybrid PV–Savonius Turbines Using Artificial Neural Networks and Computational Fluid Dynamics: Effects of Panel Position and Configuration on Turbine Aerodynamics

Creative Commons License

YILMAZ M., Yilmaz Y., YAVUZ H.

Journal of Applied Fluid Mechanics, cilt.19, sa.7, ss.1676-1696, 2026 (SCI-Expanded, Scopus)

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 19 Sayı: 7
  • Basım Tarihi: 2026
  • Doi Numarası: 10.47176/jafm.19.7.4172
  • Dergi Adı: Journal of Applied Fluid Mechanics
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Applied Science & Technology Source, Directory of Open Access Journals, Materials Science & Engineering Collection (ProQuest), Technology Collection (ProQuest)
  • Sayfa Sayıları: ss.1676-1696
  • Anahtar Kelimeler: Artificial neural network, Computational fluid dynamics simulation, Heywood, Photovoltaic panel, Renewable energy, Savonius turbine
  • Açık Arşiv Koleksiyonu: AVESİS Açık Erişim Koleksiyonu
  • Çukurova Üniversitesi Adresli: Evet

Özet

This study examines the aerodynamic performance of a Savonius-type vertical axis wind turbine integrated with photovoltaic panels that function simultaneously as flow deflectors. Two-dimensional transient simulations were performed using the k-ω Shear Stress Transport (SST) Unsteady Reynolds-Averaged Navier-Stokes (URANS) turbulence model to evaluate the influence of panel length (L), distance (x), and tilt angle (α) on the turbine's torque coefficient (C_m) and power coefficient (C_p). The configuration with L = 2m, x = 0.5m, and α = 35° achieved the highest aerodynamic efficiency, yielding a maximum power coefficient of C_p = 0.3514 at a tip speed ratio (TSR) of 1, representing an approximately 40.8% improvement over the baseline design. To minimize the number of required CFD simulations and enable performance prediction across untested configurations, Artificial Neural Network models are developed and trained using datasets obtained from the transient CFD analyses. The optimal Elman Neural Network model, employing five neurons with a tansig activation function and the trainlm learning algorithm, achieved high predictive accuracy with a Mean Absolute Percentage Error (MAPE) of 3.18%, Root Mean Square Error (RMSE) of 0.014, and Mean Absolute Error (MAE) of 0.0113. The developed ENN-based framework provides a computationally efficient and accurate alternative to CFD simulations, effectively capturing the nonlinear and time-dependent aerodynamic interactions of the turbine-panel system and supporting design optimization of hybrid renewable configurations.