Aramco and Emerson Deploy AI Models to Boost Refining Accuracy
Aramco has integrated Emerson's AI-powered optimization system into its refinery planning operations, achieving prediction accuracy of up to 98.5% in critical units. The deployment marks one of the largest multi-site, multi-period optimization frameworks in the energy sector.
The system combines physics-based engineering models with industrial AI trained on operational data. This hybrid approach captures complex relationships between feedstock composition, processing conditions, and product yields-relationships that traditional models struggle to predict accurately.
Where the technology is running
Aramco has deployed the models across Continuous Catalyst Regeneration (CCR) and Platformer Units. The company is now extending the approach to hydrocracker units, where operators expect further gains in prediction accuracy.
These units represent some of the most complex operations in a refinery. Small changes in feedstock or operating conditions can produce outsized effects on product quality and yield. Better predictions mean better decisions about blending strategies and processing parameters.
Operational impact
Higher prediction accuracy is reducing the gap between planning models and actual plant performance. This cuts the need for manual adjustments and lets operators respond faster to changing conditions.
The system also enables more flexible feedstock selection. Operators can now evaluate how different crude oils or intermediate streams will perform before committing them to processing, improving both profitability and resource efficiency.
Ahmad Alkudmani, Director of the Global Optimiser Department at Aramco, said the improved model accuracy is "enhancing planning decisions, reducing manual adjustments and uncovering new value across our global assets."
Why this matters for operations teams
The deployment shows how AI can address real operational constraints. Rather than replacing domain expertise, the models codify it-capturing decades of engineering knowledge about how refineries actually behave under different conditions.
This approach reduces the time operators spend on routine tuning and manual optimization. It also makes their decisions more defensible: when a model predicts a specific outcome with 98% accuracy, operators can justify their choices to management and stakeholders.
The system maintains performance across Aramco's global refinery network, suggesting the approach can scale across different facility designs and configurations.
For operations professionals implementing similar systems, the key lesson is straightforward: AI works best when it's built on top of established engineering principles, not as a replacement for them. AI for Operations professionals working on optimization problems should understand how to combine data-driven models with physical constraints and domain knowledge.
Managers overseeing refining operations or supply chain optimization may find relevant guidance in the AI Learning Path for Operations Managers, which covers process optimization and workflow automation in detail.
Emerson's Claudio Fayad said the deployment demonstrates "the tangible value of combining deep domain expertise with advanced AI." The company plans to expand the relationship as Aramco extends the approach to additional refining units.
Your membership also unlocks:
