Prediction of nanofluid heat transfer characteristic and pressure drop in helical coil via artificial neural networks.

In this paper, experimental data of Nusselt number and pressure drop utilizing TiO2/water, ZnO/water, and Ag/water nanofluids in helical coils under isothermal conditions were used to design an artificial neural network to predict Nusselt number and pressure drop for experimental and predict data. Feed Forward Neural Network (FFNN) and Generalized Regression Neural Network (GRNN) were designed. All predicted results were compared with the original experimental data and the results showed that both networks could be used to predict Nusselt number and pressure drop. Root Mean Square Error (RMSE) was used to measure the accuracy of each neural network. Predicted Nusselt number average RMSE for FFNN and GRNN were 4.84 and 0.004, while expected Nusselt numbers were 4.009 and 0.63. Predicted pressure drop average RMSE for FFNN and GRNN were 6.78 and 0.005 respectively while the average RMSE for expected pressure drop were 6.173 and 0.105 respectively. GRNN showed more accuracy than FFNN. The maximum deviation between trained data and experimental data for FFNN and GRNN were ±15.5% and ±0.02%, respectively. The maximum deviation between predicted data and experimental data for FFNN and GRNN were ±13.1% and ±1.6% respectively. The generated networks to predict non-experimental data showed good and logical behavior and proved that they are sufficient in Nusselt number and pressure drop prediction. [ABSTRACT FROM AUTHOR]