Desenvolvimento de aplicativo para tratamento de dados de emulsões de petróleo

Abstract

The formation of water-in-oil emulsions during petroleum production operations represents a significant technical challenge, as it increases costs and processing time. Understanding the formation and stability of these systems is of paramount importance for mitigating issues and optimizing production. Parameters associated with emulsion stability, such as the Droplet Size Distribution (DSD) and the mean droplet diameter, are fundamental for characterizing and understanding emulsion behavior. Traditional analysis methods, such as manual counting in micrographs, are labor-intensive, subjective, and poorly scalable. In this context, the present work developed a Python-based application capable of automating the processing of emulsion micrographs through the Hough Transform for circle detection, enabling the determination of the mean droplet diameter (D32) and the corresponding DSD. A total of 325 emulsions were produced using 13 distinct crude oils, with variations in temperature, salinity, and water cut; after filtering for immediate instability, 265 emulsions were analyzed. The implemented method presented an average absolute error of approximately 1.3 μm when compared to manual processing performed using the ImageJ software, while reducing processing time from approximately 30 minutes to less than 5 seconds per image set. The application also incorporated machine learning models to predict emulsion stability, viscosity, and D32, following the methodology proposed by Rolim et al. (2026). The use of Python programming enabled fast analyses, providing significant advances in standardization and in the handling of large volumes of experimental data.