Data-driven estimation of flow around cactus-inspired cylinders using ANN

OpenAlex İngilizce

Flow past cactus-inspired cylinders has been examined by Particle Image Velocimetry (PIV) and Artificial Neural Network (ANN) at Reynolds numbers (Re) of 4000 and 6000. Apex angles including β = 7.5°, β = 10°, β = 15°, β = 30° and β = 45° have been considered. The results exhibited a good agreement between measurements and predictions. Increasing β led to wake shrinkage and mitigation of flow separation. Drag coefficients (CD) are 1.1364 and 1.1483 for experimental and predicted results of Re = 6000. Drag reduction has been observed up to β = 30° except for β = 10° with ANN. For both PIV and ANN, the highest percentages of drag reduction are 18.83% and 17.64% at β = 30°, respectively. At β = 15°, drag coefficients have been reduced by 11.29% for PIV and 9.74% for ANN. However, for β ≤ 10°, the maximum percentage of drag reduction is 2.92% for PIV. Between these results, the peak deviation is 4.76% for β = 10°. The minimum deviation is 0.53% for β = 45°.