AI-DRIVEN OPTIMIZATION OF CRUDE OIL REFINING PROCESSES

Bulgyn Mailykhanova, Shamil Koshimbayev, Anar Khabay, R. A. Jamasheva, Sadratdin Abdukarimov

Bulletin of Shakarim University Technical Sciences, 2025

Full text Open access: Gold Metadata (CSL-JSON)
DOI 10.53360/2788-7995-2025-2(18)-4
Data confidence 70 medium
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

The integration of Artificial Intelligence (AI) in industrial automation has led to significant improvements in efficiency, predictive maintenance, and cost reduction. This study investigates the application of AI-based control systems in crude oil refining, focusing on optimizing process efficiency, minimizing maintenance costs, and improving system reliability. Traditional control methods, which rely on pre-defined rules and manual intervention, often lead to inefficiencies and unplanned downtime. In contrast, AI-driven automation enables real-time data analysis, predictive decision-making, and adaptive control mechanisms. Our research utilizes advanced machine learning models, including artificial neural networks (ANNs) and gradient boosting algorithms, to optimize process parameters. These models were trained using historical operational data and validated through simulation-based testing. Results demonstrate that AI-driven systems reduce maintenance costs by up to 30%, improve predictive accuracy by 25%, and enhance energy efficiency by 15%. Furthermore, intelligent control systems show high adaptability to variations in crude composition, enabling more robust and sustainable operations. To address the challenge of AI model transparency, the study incorporates explainable AI (XAI) techniques such as SHAP and LIME to improve interpretability and support trust in automated decision-making – particularly in safety-critical refinery processes. These tools provide insights into feature importance and model behavior, facilitating better understanding by engineers and operators. Despite the performance benefits, the adoption of AI in industrial environments faces challenges, including high initial investment costs, integration with legacy systems, and cybersecurity risks. The paper proposes strategies to mitigate these barriers, such as phased deployment, secure system architecture, and hybrid control models combining AI with rule-based logic. This research underscores the transformative potential of AI in refining operations and contributes to the development of reliable, transparent, and cost-effective automation solutions for the energy sector.