Abstract The aim is to estimate the stability of Al2O3-ethylene glycol (EG) nanofluids using the particle size distribution and velocity ratio. The thermal conductivity and viscosity were measured under ultrasonic conditions for various time intervals, mass fraction (from 0 to 2.0 wt%), and temperature range (from 25 to 60 °C). Moreover, various criteria were presented to estimate the thermal performance in the convective heat transfer. Based on different sets of experimental data, new correlations and optimal artificial neural network models (ANN) were proposed. The results showed that Al2O3-EG nanofluids obtained by ultrasonation for 60 min exhibits the most encouraging properties. Moreover, the correlations for the experiment and ANN models can predict these two parameters. However, the ANN model is more precise. It is expected that the results to be useful for other studies of nanofluids stability especially since it recommends suitable selecting criteria based on heat transfer behavior before real applications.

中文翻译：

---

Al2O3-乙二醇纳米流体粘度和热导率的稳定性、热性能和人工神经网络建模

摘要 目的是利用粒度分布和速度比来估计Al2O3-乙二醇(EG) 纳米流体的稳定性。在超声波条件下测量不同时间间隔、质量分数（从 0 到 2.0 wt%）和温度范围（从 25 到 60 °C）的热导率和粘度。此外，还提出了各种标准来估计对流传热中的热性能。基于不同的实验数据集，提出了新的相关性和最优人工神经网络模型（ANN）。结果表明，通过超声处理 60 分钟获得的 Al2O3-EG 纳米流体表现出最令人鼓舞的特性。此外，实验和 ANN 模型的相关性可以预测这两个参数。但是，ANN 模型更精确。