DLR Prescreening of new fuel candidates

The approval for new types of fuel is extremely complex because safety is a top priority in aviation. This prioritisation is unavoidable, but also represents an enormous hurdle for innovations. This is because fuel manufacturers lack important information about the suitability of their fuel for approval in the early stages of development, when sufficient quantities are not yet available for experimental tests. Against this background, the researchers at the institute have developed a method based on probabilistic machine learning algorithms to evaluate new fuel candidates with regard to the requirements for approval: the DLR Prescreening. We make the resulting information available to fuel-producing partners to enable holistic product development and reduce the risk of subsequent approval. With the Technical Fuel Assessment, we offer a further, scientifically in-depth and comprehensive service aimed at the targeted optimisation of a fuel for the respective purpose.

Our team uses supervised machine learning to predict the properties for the comparative analysis. The algorithms are trained with the data from the SimFuel database. They are then used to simulate and evaluate the properties of a new candidate based on the detailed gas chromatography results. The visualisation of the results in specially developed, complex diagrams makes it possible to see at a glance what behaviour a fuel will exhibit later - and, above all, whether these properties are within or outside the specifications set by the ASTM.