Honeywell launches plant AI at Dangote's Lekki refinery

By James Calloway

Honeywell estimates that AI-enabled optimisation at Dangote Petroleum Refinery in Lekki, Nigeria, could produce more than $60 million a year in gross benefits once the refinery runs at full-capacity throughput.

The estimate follows preliminary data showing gross operational benefits above $0.29 per barrel at Lekki. Honeywell says the figure excludes implementation costs and is not guaranteed, leaving the difference between modelled upside and refinery margin unstated.

Honeywell Technologies launched Forge Plant Cognition as an industrial AI platform designed to improve refinery performance, operational reliability and profitability. Developed with NVIDIA’s AI and accelerated computing capabilities and Dell Technologies’ AI infrastructure, it targets refining, petrochemical and liquefied natural gas operations. Dangote’s Lekki refinery will host the first plant deployment.

What changes for refinery operators?

Built on Honeywell Technologies’ Forge Cognition platform, Forge Plant Cognition combines industrial automation, process technology expertise and AI capabilities. It brings real-time operating data together with engineering knowledge, process expertise and physics-based models. The system aims to help operators identify performance constraints, anticipate problems and turn insights into practical actions.

Unlike conventional AI tools that mainly analyse data and generate recommendations, the platform connects live operating information with the physical context of the plant. For operators, its value rests on turning those inputs into specific decisions about throughput, yields, equipment and process performance.

The platform focuses on three operating areas:

• Identifying bottlenecks to improve throughput and product yields.
• Detecting emerging equipment and process issues to reduce unplanned downtime.
• Embedding specialised industrial knowledge into AI-powered operating workflows.

For operators, the test will be higher throughput and product yield with fewer unplanned outages.

What did Honeywell say about the rollout?

Rajesh Gattupalli, President of Honeywell Technologies UOP, said:

“Industrial plants are among the most complex operating environments in the world, with enormous potential to realise greater performance and value using digital technologies and connected solutions. By collaborating with Dangote to launch Forge Plant Cognition, we are demonstrating how AI can help transform industrial operations, driving new levels of productivity, reliability and profitability at significant scale.”

The collaboration gives Honeywell its first refinery operating environment for testing the platform’s performance claims. Dangote will also determine whether the estimated per-barrel benefit survives contact with implementation requirements and daily plant operations.

The Dangote rollout will show whether the platform can turn a joint technology launch into repeatable refinery gains.

What technology sits behind the system?

Forge Plant Cognition uses the Dell AI Factory with NVIDIA and receives support from Dell Services. This combines Dell’s AI infrastructure with NVIDIA’s accelerated computing tools for refinery data, models and simulations.

NVIDIA Nemotron open models power AI agents that interpret operational data. NVIDIA Omniverse libraries support digital twins of plant operations for simulation-based training and planning.

NVIDIA Cosmos models, combined with the NVIDIA Metropolis VSS Blueprint, enable vision AI agents to generate real-time alerts and detect anomalies. The NVIDIA PhysicsNeMo framework supports development and training of physics-based AI models for digital-twin-integrated plant optimisation.

The technical stack links live plant data with simulation and physics-based models so operators can test and act on operating changes.

Why did Dangote choose the platform?

David Bird, CEO and Managing Director of Dangote Petroleum Refinery and Petrochemicals FZE, said:

“We chose Honeywell Technologies’ Forge Plant Cognition because the platform can deliver unrivaled industrial expertise combined with differentiated AI capabilities to help us optimize performance, unlock greater value from our operations and evolve how we operate our assets.”

Beyond refinery performance, the deployment supports Dangote’s goal of developing a workforce that is 70 per cent Nigerian. Providing engineers and operators with access to embedded process knowledge should accelerate problem-solving and reduce reliance on external expertise over time, according to Honeywell.

That makes the rollout a test of industrial software and the transfer of refinery knowledge to local engineers. The deployment therefore tests both refinery performance and the transfer of process knowledge to Nigerian engineers.

What should operators watch next?

Honeywell did not disclose a startup date or implementation cost for the Lekki deployment, and the announcement gave no refinery capacity figure in barrels per day. The company also did not commit to the preliminary $0.29-per-barrel benefit. The estimate therefore remains a gross operating opportunity rather than a demonstrated improvement in net margin.

Several measures will indicate whether the deployment delivers the stated economics:

• The startup date and scope of the platform’s use at Lekki.
• Measured changes in throughput, product yield and unplanned downtime.
• Realised gross benefits per barrel against Honeywell’s preliminary estimate.
• Implementation spending and the resulting effect on refinery margin.
• Progress toward Dangote’s 70 per cent Nigerian workforce goal.

The item to watch is whether Dangote can turn the preliminary $0.29-per-barrel gross-benefit estimate into measured operating gains after implementation costs, with a clear effect on refinery margin.