Utilizing machine learning techniques to predict the evaporation and falling of a droplet

Droplet evaporation plays a significant role in various natural and industrial phenomena, such as raindrops, combustion sprays, evaporative coolers, and in the food and pharmaceutical industries. Consequently, diverse research studies have been conducted in this field. This research evaluates the capabilities of machine learning (ML) models to predict the evaporation rate and fall of a droplet. In the first step, a series of simulations were conducted using the front-tracking method, and the results obtained from these simulations were utilized as inputs for training and testing ML methods. Subsequently, based on the coefficient of determination (R2) and root mean square error, the performance of five models Random Forest (RF), Artificial Neural Network (ANN), Classification and Regression Trees (CART), Support Vector Regression (SVR), and Linear Regression (LR) was assessed in predicting the evaporation rate and fall of droplets. The ANN algorithm demonstrated the best performance in predicting both the evaporation and fall rates with values of 0.999 for the R2 index, followed by CART, RF, LR, and SVR in that order. After selecting the ANN model as the best model, it was used to predict the falling and evaporation rates of droplets at density ratios higher than the training and testing data range. According to the results, the model consistently provided accurate and appropriate predictions, confirming the high efficiency of the model and indicating that by utilizing ML models, not only the computational costs can be reduced, but results related to certain problems that challenge numerical methods and are difficult to simulate, can also be predicted.