Turning a Complex Problem into an AI Challenge

The system, developed within IAG’s London and Barcelona-based AI Labs, is designed to tackle a particularly complex issue: how to schedule engine maintenance in a way that simultaneously meets regulatory requirements, part availability, labor constraints, and operational continuity.

Every commercial jet engine must adhere to strict regulatory intervals while also fitting around flight schedules, parts inventory, and shop capacity. Planners manage thousands of variables, yet one delayed part or an unexpected route change can disrupt months of planning.

By running millions of “what-if” scenarios daily, IAG’s new system re-plans in minutes instead of weeks, helping the airline avoid Aircraft On Ground emergencies serious enough to ground the airplane until it's repaired and preventing maintenance-related passenger delays. The system dynamically updates maintenance schedules, adapting in real time as new data arrives.

“By applying advanced algorithms, we’re making our engine maintenance programme more efficient. We are avoiding unnecessary maintenance delays to ensure that our fleet is available and in service,” says Ben Dias, IAG’s chief AI scientist. “The system gives our people the data and tools they need for smarter planning and better teamwork.”

An In-House Approach to AI System Development

Many organizations purchase predictive-maintenance dashboards from OEMs or software vendors. IAG chose a different approach: retain control of the data and code, and tailor the algorithms to its own mixed fleet. Dias’ team started with the widely used CFM56 engine, common in narrow-body aircraft, to validate the concept before expanding to other engine types.

Owning the intellectual property is important for two reasons. First, IAG can continuously improve the model as its network, fleet composition, and shop capacity evolve. Second, the group avoids vendor lock-in, which is crucial when an engine swap between BA and Iberia depends on data portability.

AI Making an Increasing Impact in the Airline Industry

IAG’s efforts mirror similar changes across aviation. Lufthansa Technik uses its Aviatar platform for predictive diagnostics that identifies recurring fault codes and suggests solutions, now used by more than 100 airlines. Delta Air Lines’ APEX engine-health system analyzes real-time sensor data; the carrier reports that parts-demand accuracy has improved from 60% to 90%. Air France-KLM is collaborating with Google Cloud to integrate generative-AI tools into its existing “Prognos” analytics stack for both maintenance and network planning.

What sets IAG apart is its focus on prescriptive optimization. Rather than simply predicting when an engine might require service, the model selects the optimal slot that minimizes ground time across a 700-aircraft fleet.

Looking at the bigger picture, the financial benefits become evident. With the industry expected to spend over $100 billion annually on maintenance, repair, and overhaul (MRO) by 2030 according to Strategic Market Research’s Aircraft MRO Market Size & Forecast report, even small efficiency gains have significant implications. McKinsey estimates that AI-driven maintenance could reduce costs by 20% and cut unscheduled repairs by up to half.

There is also an environmental benefit. By minimizing last-minute swaps and repositioning flights, the system reduces emissions, helping airlines meet sustainability goals while cutting costs. A more streamlined shop schedule means fewer repositioning flights and emergency charters, lowering fuel consumption and CO? emissions.

Obstacles on the AI Taxiway

However, there are challenges ahead. AI systems depend on clean, consistent data, but aviation data is often inconsistent. Airlines still deal with irregular logbook entries, paper-based records, and parts labeled under multiple naming conventions. IAG spent months cleaning historical files and standardizing schemas before training the model. Integrating these systems with existing workflows, especially under strict safety regulations, adds another level of complexity.

Managing change is equally difficult. Engineers accustomed to planning on whiteboards may resist a probabilistic recommendation engine. That’s why the system presents its schedule along with the reasoning behind each decision, allowing human oversight. Trust grows when planners can question the AI, adjust a variable, and see the plan update instantly.

Getting the data right and gaining trust from frontline teams will be essential for long-term success.

Where the Airline Industry Is Heading

Future AI developments could take things even further. Technicians could share anonymized model insights among member airlines through a federated-learning loop. This would allow datasets from different airlines and locations to enhance each other without revealing sensitive business details. Over time, this could feed the optimization layer with live flight-ops and crew-roster data so that disruption management and maintenance planning come from a single source of truth.

If that seems ambitious, remember that pilots once carried over 30 pounds of binders to the cockpit in large black roll-aboard suitcases. The electronic flight bag (EFB), a tablet that stores charts, manuals, and performance calculators digitally, revolutionized that process. Today, it's standard. A decade from now, an AI-based scheduler that treats engines, slots, and spares as a dynamic puzzle may seem just as routine, and IAG will have secured a multi-year head start.

The above is the detailed content of How IAG's Home-Grown AI Could Save Airlines Millions. For more information, please follow other related articles on the PHP Chinese website!