Exascale Computing for System-Level Engineering: Design, Optimisation and Resilience

The arrival of exascale supercomputers will open up new frontiers around our ability to simulate highly complex engineered and natural systems. This will expose new opportunities for the design and optimisation of new, highly integrated engineered systems for the future. Enabling paradigm-shifting, high-fidelity simulations on novel, highly heterogeneous exascale hardware will require a joint community effort between mathematicians, computer scientists, software engineers and domain specialists. For the development of mathematical methods and efficient software implementations for coupled problems targeting current petascale and future exascale systems it is absolutely essential to consider the following:

  1. The mathematical foundations of how to accurately and robustly couple different candidate methods for different physical systems so as to achieve extreme levels of parallelism.
  2. How system utilisation can be maximised when simulating each physical process with differing and potentially quite disparate time scales together with a consideration of the directionality of couplings. 
  3. How different methods for different physical processes vary in suitability with respect to hardware in a heterogeneous exascale system (e.g. CPU vs GPU).
  4. How software libraries can be developed to be sustainable and performance portable, and capable of running single-physics models at exascale in orchestration with other models (a requirement typical of the interacting physical processes representing sophisticated engineered and natural systems).

Objectives include:

  • Mapping out of future research and development needs that will enable unprecedented simulations of complex, coupled physics systems of high industrial and societal relevance in order to exploit the first generation of exascale computing systems.
  • The creation of a community, supporting close engagement between research software engineers and researchers, to foster skills development and collaboration around open source software tools that will target the effective exploitation of exascale computers.
  • Demonstration, through proof-of-concept software implementations and petascale testing, the viability of simulating challenging, system-level problems from engineering and nature, again focusing upon on the impending arrival of the exascale.

The technical challenges addressed by the working group have been framed and guided by four high-priority Application Themes (or use cases):

  1. Coupled airframe-propulsion simulation of electrified air transport vehicles.
  2. Integrated, offshore wind farm simulation and optimisation.
  3. Digital twins of the human body. 
  4. Resilience against sea level rises and flooding: regional-level simulation of engineered and natural flood defences.

In order to support the use cases, a series of workshops will be held. Leaders will collaboratively investigate, review and assess the challenges around the selected use cases, including an assessment of the start-of-art in each area, dependencies for advancing the area and the relationship to other themes. The workshop topics that are being explored include:

  • Simulation of integrated engineering systems at extreme scale
  • Inverse problems and optimisation at the exascale
  • Data assimilation and uncertainty quantification at the exascale
  • Data visualisation and data flows for the exascale
  • Software engineering for exascale exploitation

Latest news