Help make aviation more sustainable and climate-neutral. How? Quite simple: Join the alternative fuels team at the Institute of Combustion Technology and do research in the field of sustainable aviation fuels!

There are currently no technological alternatives to liquid fuels, particularly for medium- and long-haul flights. At the same time, a large part of the harmful climate and environmental impact of aviation occurs on these routes. To address this, sustainable aviation fuels (SAF) are being extensively researched at the DLR Institute of Combustion Technology. The institute is involved in a number of large European research projects, both in the field of fuel production and fuel use. The aim is to develop sustainable fuels with a particular focus on minimizing the impact on the climate to enable a world of tomorrow worth living in.

In the last few years, unique data sets on the effects of a wide variety of fuels have been collected in international ground and flight measurement campaigns. On this basis, our team in the department of Multiphase Flow and Alternative Fuels (MAT) develops and applies data-based methods (big data, machine learning, digital twin) to maximize the benefits of sustainable fuels. As part of the department, you will work in an international team of about 12 scientists from Germany and abroad.

Although the demand for SAF by the market and legislation is increasing, the introduction of alternative fuels is associated with major challenges for the aviation industry. On the one hand, it is desirable to approve the largest possible number of fuels. On the other hand, no risk must arise from the use of a new fuel. Therefore, a wide variety of fuel properties need to be tested and critically reviewed in an elaborate approval process. In this context, our team developed the ASTM-approved DLR prescreening process on the basis of probabilistic machine learning algorithms to evaluate new fuel candidates with regard to requirements for fuel approval. Furthermore, we provide partners with the information necessary for a holistic product development.

Machine learning models - such as those developed at DLR - are flexible algorithms, which demonstrated the capability of predicting a wide range of properties. The quality of the prediction however depends significantly on the training data set. This raises the question of how the training data must be distributed to achieve the desired predictive capability. Why is this important? Owing to novel production routes, the composition of new sustainable kerosene products will most likely differ significantly from conventional kerosene and other already approved fuels. In this context, your mission will be the evaluation and enhancement of current predictive capabilities with respect to new fuel production routes.

Your Tasks:

• training und validation of probabilistic machine learning models
• evaluation of the current predictive capabilities with respect to new fuel production routes and the influence of the distribution of the training data sets
• identification of gaps in the fuel data and closing of those gaps by defining data points for the training of the ML models in collaboration with the DLR-VT analytics department
• development of similarity metrics to identify risks and loss in accuracy, precision and reliability of the probabilistic ML models when being applied to novel fuel samples
• evaluation of the potential of embedding general physical laws into the ML methodology
• critical analysis and improvement of the methodology
• tasks as part of the maintenance of the fuel data base
• make specific data sets accessible via web-based dashboards
• screening of fuel products with respect to present and future (100% SAF) approval standards
• analysis and Identification of promising fuel candidates with respect to added values, e.g. minimize emissions and thus environmental impact
• tasks as part of research projects
• presentation and publication of (scientific) work results
• there is the opportunity to do a PhD