The activities in CoEC werte focused on the development of new methodologies for computational combustion in (pre-) Exascale systems, testing the codes in new architectures, defining Exascale Challenge Demonstrators (ECDs), creating training and services, and building a community around the project. A dedicated effort has been given to conduct training, where several training courses in virtual format have been conducted and additional courses are in preparation. A common language was established in the CoEC Consortium, between partners from Academic and Research Institutions, HPC centers, Scientific organisations and Industrial and Governmental stakeholders. In the transition from Demonstrators (ECDs) to Services, end-users were involved in the co-design of Services by defining, in synergy with the CoEC teams and the ECD leaders, the appropriate use-cases and requirements for the validation of the ECDs. This was consolidated in the Final Workshop that took place in Barcelona in December 2023 were representative members from the Advisory Board, and invited industries participated in the event with more than 40 participants from industry and academia.
On the technical side, the project focused on the development of new methodologies for combustion. The activities started with a successful collaboration with the CoE POP to obtain a profiling of all the CoEC flagship codes, which was used to identify the requirements and define the development roadmaps of the codes. From this point, four key methodologies were addressed in CoEC to adapt the codes for Exascale. It includes (1) high-order methods and low-dissipation numerical schemes to improve the accuracy of the numerical simulations and better exploit the parallelism of the heterogenous systems with accelerators, (2) error estimators for spray flames and more generally for turbulent two-phase flows, (3) adaptive chemistry and chemistry reduction strategies through the development and optimization of methodologies for the on-the-fly reduction and optimize ODE solvers adapted to accelerators, and (4) methodologies to deal with two phase flows including Eulerian-Eulerian and Eulerian-Lagrangian methods. The use of Machine Learning with data processing and visualization has been identified to be an enabling tool to tackle some of the fundamental challenges in CoEC, while exploit efficiently the upcoming Exascale architectures, and this has been an active area of development.
CoEC has proposed 13 ECDs that deal with fundamental problem in propulsion and power generation. Those include simulations of hydrogen flames, spray flames, pollutant formation, thermoacoustics, sparks and plasma, and internal combustion engines. The selection of the ECDs has been done in collaboration with the Industrial and Scientific Advisory Boards and has ensured the relevance of the proposed problems to the industrial sectors.
