(420c) Machine Learning Based Prediction of Oilfield Scale Formation Kinetics.

Several millions of dollars have been invested into researching the mitigation of scale formation in oilfield industries, a persistent problem that leads to decreased operational efficiency and increased maintenance costs. Here, we utilize a proactive approach, harnessing the power of machine learning to predict scale formation kinetics using data obtained from a novel continuous stirred tank laser setup. This setup offers precise control over experimental conditions and enables real time monitoring of scale formation processes at simulated oilfield conditions. This is fundamental approach, the use of machine learning models can often outperform traditional empirical or mechanistic approaches by capturing subtle trends and nonlinear relationships in the data, leading to more accurate predictions. This informs accurate inhibitor application and is a less costly approach that can be harnessed to eradicate oilfield scale formation altogether. The continuous stirred tank laser setup is used to create an extensive data set that captures the nuances of the formation of different types of scales at different replicated oilfield conditions and a machine learning algorithm is developed from it that accurately predicts the kinetics of oilfield scale formation.