In the collaborative research project DIGIfly physically comprehensive digital models and simulation methods are being developed for the design of aircraft and engines. To this end, advanced modules of the European next generation CFD-solver will be further developed and machine learning methods and artificial intelligence will be adapted. The methods and modules will be evaluated based on realistic use cases and transferred into industrial processes.
In DIGIfly DLR combines its expertise and capabilities in the development and application of high-fidelity flow simulation processes for industrial problems with the knowledge of advanced methods from the fields of machine learning and artificial intelligence (AI) in its own proposal SMARTfly (Smart Modeling of flying Transport Vehicles). The research focuses on the further development of the CFD-solver CODA and the DLR software toolboxes AutoOpti (engines) and SMARTy (aircraft).
For the CFD-solver, among other things, the method of "Manufactured Solutions" for automatic code verification will be used and an augmented data-driven turbulence model with enhanced predictive performance will be implemented. In AutoOpti and SMARTy advanced surrogate models for high-dimensional problems, novel methods of dimension and model reduction and innovative procedures from the fields of machine learning, AI and statistics will be implemented. These methods should enable to generate data-driven models from simulation data, from measurement data or from a combination thereof and to provide confidence intervals. In addition, data fusion algorithms that allow a combination of numerical and experimental data with measurement inaccuracies will be further developed and tested. Based on the two aforementioned core topics, a process automation by means of FlowSimulator on modern high-performance computers and graphics cards will be implemented. Finally, the developed and automated simulation and data fusion capabilities will be tested for different application scenarios and evaluated in close cooperation with industrial partners in order to generate best practice guidelines. In addition, DLR will provide support to the partners to enable an efficient integration of these new methodologies into their individual workflows. The combination of these activities will allow an optimal alignment of DLR's numerical and data-driven simulation capabilities to answer complex questions of the aerospace industry in the future. In addition, current trends from research are systematically incorporated into the industrial aerospace environment.
The DLR Institute of Aerodynamics and Flow Technology is active in the development of CODA and SMARTy as well as in process automation and the investigation of various use cases. In addition, the Institute coordinates both, the network with external partners and the DLR work, with the other participating DLR Institutes