An AI-driven framework for property-oriented fuel design toward sustainable engine combustion

Co-optimizing renewable synthetic fuels and internal combustion engines is critical for decarbonization, with fuel design as a crucial approach. However, the combinatorial explosion of fuel compositions renders composition-driven design infeasible. While artificial intelligence (AI) offers optimization tools, data scarcity hinders its application in engine research. Therefore, we propose an AI-driven fuel design framework using physicochemical properties to bridge fuel composition and engine performance. Treatin…

By the Truvace deskOctober 16, 20262 min read1 sourcePeer-reviewedTRV-2026-1518· cite & verify

In brief

Researchers propose an AI framework that designs fuels by their physicochemical properties rather than by chemical composition, linking those properties to engine performance. They trained a predictive model on a dataset mined from published literature and used it in a proof-of-concept demonstration to select a tailored property combination.

The approach matters because the huge number of possible fuel compositions makes composition-driven design impractical, and scarce data has limited AI use in engine research. The reported gains come from a demonstration of a property combination, so it remains to be seen how they translate into real synthesized fuels, which the authors say the framework could help guide.

Main points

Treating properties as direct design variables reduces the dimensionality of the fuel design problem.
Property-based variables allow standardized integration of data from different studies.
The predictive model was trained on a literature-mined dataset.
Versus diesel, the identified property combination improved thermal efficiency by 6.62%.
NOx fell by 19.28% and soot by 67.76% relative to diesel in the demonstration.

The gain

By using physicochemical properties as design variables, the AI framework found a property combination that, relative to diesel, raised thermal efficiency and cut NOx and soot emissions in a proof-of-concept test.

The rundown

The work starts from the need to co-optimize renewable synthetic fuels and internal combustion engines for decarbonization, where fuel design is a key lever. The authors note that the combinatorial explosion of possible fuel compositions makes composition-driven design infeasible.

To get around this, the framework uses physicochemical properties as a bridge between fuel composition and engine performance. Because data are scarce in engine research, pooling literature data under a common property-based format is central to the method.

In the proof-of-concept, the trained model was applied to identify a tailored property combination. The authors say such property-oriented design could instruct upstream synthesis and help reduce trial-and-error in fuel development.