Numerical study and artificial neural network modeling of the tube banks arrangement considering exergetic performance

The present study investigates numerically heat transfer rate (HTR) and generation of entropy (GOE) for turbulent NF flow in an electronic cooler equipped with four sinusoidal pipes. The heat transfer fluid (HTF) is water-based MWCNT nanofluid (NF) in different NVFs (ϕ = 0.2, 0.4, 0.6 and 0.8%). Four different wave amplitudes (α = 2, 4, 6 and 8 mm), three different wavelengths (λ = 40, 60 and 80 mm) are studied in this paper. According to obtained results, the average Nusselt number (Nu) always increases by increase in wave amplitude or decrease in wavelength. Therefore, at α = 8 mm or λ = 40 mm the highest Nu value is achieved. Besides, Reynolds numbers (Re) variation has a significant influence on Nu distribution in tube and the Nu values always intensify by augmentation of Re. The optimum Reynolds number in thermal viewpoint is 11,000 and in second law viewpoint is 5000. Furthermore, therefore higher nanoparticles volume fraction (NVF) always leads to higher HTR in channel. But it also increases dynamic viscosity and pressure drop penalty in channel. The results of the models obtained from the neural network algorithm show that the variation of the Nusselt number with respect to the Reynolds number, wave amplitude and wavelength is mitotically so that heat transfer increases by increasing the amount of these three parameters. An increase in the wave amplitude leads to a 38% increase in the average Nusselt number and a 50% increase in the total Nusselt number. However, the effect of the wavelength is small. An increase in the wavelength results in an enhancement in the average Nusselt number by 6% and the total Nusselt number by 16%.