Honeywell Brings AI Anomaly Detection to Refineries
Honeywell Technologies has launched Forge Plant Cognition, an industrial AI platform designed to detect operational anomalies and anticipate equipment problems. The technology combines AI models, real-time plant data, and digital twins to support refinery operations.
Honeywell announced this platform will first be introduced at Dangote’s petroleum refinery in Lekki, Nigeria. Honeywell developed the solution using NVIDIA AI technologies and infrastructure from Dell Technologies. Forge Plant Cognition targets refineries, petrochemical facilities, and liquefied natural gas (LNG) operations. It combines industrial automation, engineering expertise, and AI to identify process constraints and support operational decisions.
AI Detects Emerging Equipment and Process Issues
The platform analyzes operating data alongside engineering calculations and physics-based models. This approach is designed to help operators identify developing equipment problems before they cause unplanned downtime. Honeywell identifies three main applications: optimizing production throughput, detecting emerging process issues, and embedding industrial expertise into AI-supported workflows.
Forge Plant Cognition combines real-time industrial data, physics-based models, and vision AI to detect anomalies, anticipate equipment problems, and support operational decisions at complex processing facilities.
The system can recommend corrective actions and help engineers respond to changing operating conditions. “By collaborating with Dangote to launch Forge Plant Cognition, we are demonstrating how AI can help transform industrial operations,” said Rajesh Gattupalli, President, Honeywell Technologies UOP.
Digital Twins and Vision AI
Forge Plant Cognition runs on Dell AI Factory with NVIDIA and incorporates several NVIDIA technologies. NVIDIA Nemotron models support AI agents that interpret industrial operating data. NVIDIA Omniverse libraries enable digital twins for simulation-based training and operational planning.
The platform also uses NVIDIA Cosmos models with the NVIDIA Metropolis VSS Blueprint to enable vision AI agents. These capabilities support real-time alerts and anomaly detection.
NVIDIA PhysicsNeMo provides tools for developing physics-based AI models that support digital twin applications and plant optimization. The combination allows operators to evaluate industrial processes using operational data and simulation capabilities.
First Application at Dangote Refinery
Dangote Petroleum Refinery will integrate its operating data with Honeywell’s process expertise and AI optimization tools. The deployment aims to improve troubleshooting, production efficiency, and operational decision-making. Honeywell estimates that Forge Plant Cognition could deliver more than $60 million in annual gross operational benefits at full refinery capacity.
The preliminary estimate is specific to the Lekki facility and excludes implementation costs. Honeywell states that actual results will depend on operating conditions and other factors. The project will also support technical training for local engineers and operators, contributing to Dangote’s goal of developing a workforce that is 70% Nigerian.
The launch follows Honeywell’s broader introduction of Forge Cognition, an AI platform for buildings and industrial operations. The company is also testing that technology at the Fairmont Hotel in Dubai, as we have previously reported.
