Work Package 3

Compute-Driven Use-Cases towards Exascale

Leader:

Prof. Wolfgang Schröder, RWTH

This WP aims at integrating and further developing scaling AI methods from WP2 along representative use-cases from academia and industry that have a strong focus on compute-driven technologies, i.e., that employ high-scaling simulation technologies for various complex physical problems. The tasks of this WP are structured and laid out such that they cover a wide range of applications and that corresponding developments are transferable to other problems, also in industrial context. They concentrate on AI developments in turbulent boundary layer analysis, wind farm layout optimization, reacting flows, next-generation aircraft design, and wetting hydrodynamics, some of them aiming at full loops.

full loop.png

In each of the use-cases, the various experts collaborate closely to integrate all necessary aspects in obtaining Exascale-ready AI solutions for real-world problems. The overarching goals, which mirror to the use-cases, are

  • the development of in-situ feature detection techniques, i.e., Convolutional Neural Networs are trained in-situ to support automatic detection of important physical features in simulations to gain further insight into complex phenomena,

  • solving control and optimization problems, i.e., to have error-controlled extrapolation methods at hand,

  • perform Physics-Informed Deep Learning to use physical constraints as priors in the loss of Machine / Deep Learning algorithms to restrict learning to a space of physically reasonable / sound solutions,

  • replace expensive physical processes by AI models and surrogates to make simulations computationally more efficient,

  • to develop new numerical models with the help of AI, i.e., to derive at new models for multi- physics and/or multi-scale problems,

  • and to uncover non-obvious correlations of physical phenomena by means of AI methods to (i) increase computational performance and (ii) increase the accuracy of modern numerical simulation methods.

Tasks in WP3

Task
Task name
Leader
Contributors
Duration
T3.1
AI for turbulent boundary layers
RWTH
FZJ
M1-M36
T3.2
AI for wind farm layout optimization
BSC
RTU
M1-M36
T3.3
AI for data-driven models in reacting flows
CERFACS
UOI
M1-M36
T3.4
Smart models for next-generation aircraft engine design
SAFRAN
UOI, CERFACS, BULL
M1-M36
T3.5
AI for wetting hydrodynamics
CYI
UOI
M1-M36