This post is also available in:
עברית (Hebrew)
Aviation faces mounting pressure to lower fuel use and emissions, but improving engine efficiency remains an expensive and time-consuming challenge. Engine components must be tested across countless configurations, and full flow simulations often take hours or days to complete. Researchers at TU Graz have now introduced an AI-supported approach that dramatically accelerates this process, offering a pathway to more efficient aircraft engines that better meet Europe’s long-term sustainability targets.
The team focused on a critical but under-optimized component: the intermediate turbine duct. This section sits between the high-pressure and low-pressure turbines, guiding hot airflows that rotate at different speeds. Although essential to engine performance, the duct is heavy and difficult to redesign. Length and shape changes can impact efficiency, weight, and manufacturability, and traditional simulations make rapid iteration almost impossible.
According to the Innovation News Network, to overcome this limitation, researchers drew on years of measurement data and computational flow models, building an extensive database of how different duct geometries affect performance. They then applied machine-learning techniques to identify which parameters matter most. The most successful method was a reduced order model, which is an AI approach that extracts only the most influential features from the data and uses them to approximate flow behavior.
This technique allows engineers to evaluate design changes orders of magnitude faster than full simulations. Although the model sacrifices some precision, it reliably predicts performance trends, helping researchers identify promising design directions before running more detailed tests. The system can also flag how efficiency shifts when individual parameters—such as duct length—are adjusted, enabling targeted optimization across the entire engine architecture.
While the work is intended for civilian aviation, improved engines also benefit the defense sector. Military aircraft operate in extreme conditions where fuel efficiency directly affects range, endurance, and logistical demands. Faster modeling tools could accelerate the development of engines for next-generation aircraft, drones, and long-endurance platforms, helping reduce the operational footprint of defense aviation.
Next, the team plans to extend its model from two-dimensional to three-dimensional simulations, enabling more realistic predictions. The database and modeling tools will be shared with other research groups, accelerating progress across the aerospace community.
As aviation shifts toward stricter efficiency targets and more demanding operational profiles, AI-assisted engine design may become essential, offering a cost-effective way to explore innovations that traditional engineering alone would struggle to uncover.
