PyFR: Towards Industry and Exascale

Lead Research Organisation: Imperial College London
Department Name: Aeronautics

Abstract

This is an extension of the Fellowship: 'Developing Software for High-Order Simulation of Transient Compressible Flow Phenomena: Application to Design of Unmanned Aerial Vehicles' - EP/K027379/1.

Over the past decades, computer simulations of fluid flow have emerged as an important tool for design of complex systems across a range of sectors. It is apparent, however, that for a range of flow problem current generation software is not fit for purpose. Newer software is required, that can make effective use of current and future computing platforms, to perform highly accurate so called 'scale-resolving' simulations of unsteady flow phenomena over complex geometric configurations. Such capability would lead to design of more efficient and capable technology across a range of sectors, including aerospace, defense, architecture, automotive, and green energy.

Current activities under award EP/K027379/1 have led to development of PyFR (www.pyfr.org), a new software that can effectively leverage capabilities of massively-parallel computing platforms, with a view to undertaking hitherto intractable simulations of unsteady airflow over complex Unmanned Aerial Vehicle (UAV) configurations. The proposed Fellowship extension will address a range of outstanding issues currently blocking wider industrial adoption of PyFR, taking it further "Towards Industry", as well as addressing a range of issues that will block exploitation of PyFR on next-generation exascale supercomputers, taking it further "Towards Exascale". The proposed Fellowship extension will also look to expand the application space of PyFR beyond just UAVs to a range of sectors, and includes test cases involving flow over turbine blades, missiles, buildings, and submarines.

The research program will be lead by Dr. Peter Vincent, a Reader in the department of Aeronautics at Imperial College. It will be undertaken in collaboration with various industrial partners including MTU Aeroengines, MBDA, Arup, BAE Systems Submarines, BAE Systems MAI, NASA Glenn, Nasa Langley, NVIDIA, Pointwise, Kitware, Zenotech, and Oak Ridge National Lab, and with various academic partners including Stanford University, and the Massachusetts Institute of Technology. This assembled team of project partners, comprising a selection of the world's leading companies and elite research institutions, will ensure the project successfully delivers its objectives.

Planned Impact

Beneficiaries

Beneficiaries of the research include UK Government and the general public (via its impact on UK defence capabilities, green aviation, and knowledge/skills - see Societal Impact below), and UK business/industry (via its impact on high-tech exports, and knowledge/skills - see Economic Impact below). Academic beneficiaries and impact are discussed in Academic Beneficiaries.

Societal Impact

It is becoming increasingly important that the UK Government leverage technological advances to maintain and improve defence capabilities on a reduced budget. Outcomes from the proposed Fellowship extension will help advance a range of UK defence capabilities. These include Unmanned Aerial Vehicles (UAVs), missile systems, and submarines. UAVs will increasingly support key components of UK defence strategy and policy, and hence UK national security. Their importance was highlighted in the 2010 Strategic Defence and Security Review which, whilst cutting several more established capabilities, explicitly mentioned UAV technology as an area to be explored further. Missile systems are employed across all services, and underpin the effective function of many platforms. They are a critical technology in terms of UK national security. Submarines form a vital component of the UK's nuclear deterrent, which is currently being renewed. Via collaboration with BAE Systems Submarines, PyFR (www.pyfr.org) is likely to play an important role in this project "Submarines are seen by the HMG as having great importance especially Vanguard replacement, i.e. Dreadnought ... We see PyFR and its future development as one of a number of very important tools in a very important part of national defence." - David Hankey, Future Capability Manager, BAE Systems Submarines. Taken together, the proposed work will have a significant impact on various important UK defence capabilities, and hence have significant societal impact.

Air transportation moves over 3.5Bn passengers annually and produces upwards of 700M tonnes of CO2 and other Greenhouse Gases (GG), which are a significant driver of climate change. Global demand for air transportation is rising, particularly in emerging markets. The total number of passengers increased by 6.5% in 2015 alone, with similar growth forecast for the foreseeable future. Outcomes from the proposed Fellowship extension will help reduce GG emissions from commercial aircraft via e.g. design of more efficient low-pressure turbines. This will help make growth in commercial aviation environmentally viable, and will hence have significant societal impact.

Knowledge/skills generated in the areas of numerical methods, and massively-parallel computing can be applied in various other fields of research, including weather prediction, and climate modelling, both of which have significant societal impacts.

Economic Impact

Outcomes from the proposed Fellowship extension will help advance a range of high-tech high-value export industries. These include aerospace (turnover £32.0Bn in 2016, exports of £28.0Bn), architecture (turnover £2.5Bn in 2016, exports of £474M), defence (turnover £23.0Bn in 2016, exports of 8.7Bn per year for the period 2012-2015), automotive (turnover £77.5Bn in 2016, exports of £60.5Bn), and green energy (turnover £43.1Bn in 2015, exports of £4.1Bn). This will consequently have a significant economic impact.

Knowledge/skills generated in the areas of numerical methods and massively-parallel computing will also benefit other important UK industries, such as the financial sector, which relies heavily on efficient algorithms and is keen to explore the benefits of modern massively-parallel computing technology. Having the aforementioned knowledge/skills in the UK will also drive further international R&D investment in the UK, across all the above-mentioned sectors.

Publications

10 25 50
 
Description Development of software for very accurate simulations of unsteday turbulent fluid flow using supercomputers:

- PyFR https://pyfr.org

Development of techniques to:

- Inject turbulence into flow simulations https://arc.aiaa.org/doi/abs/10.2514/1.J061046

- Speed up flow simulations using cache blocking https://www.sciencedirect.com/science/article/abs/pii/S0010465521003052

- Speed up flow simulations using locally adaptive pseudo time stepping https://www.sciencedirect.com/science/article/pii/S0021999119306187

Use of technology for both scientific and industrially relevant test cases with project partners, specifically:

- Application to simulation of flow over a low pressure turbine blade (part of a jet engine) with MTU Aeroengines https://www.sciencedirect.com/science/article/pii/S0045793021001560

- Application to simulation of flow over high-rise buildings with Arup https://www.sciencedirect.com/science/article/pii/S0167610522002653

- Application to fundamentals of turbulent flow https://www.cambridge.org/core/journals/journal-of-fluid-mechanics/article/identifying-eigenmodes-of-averaged-smallamplitude-perturbations-to-turbulent-channel-flow/EF03E2FFFF1A481FC55448B3F11496F2
Exploitation Route The software that was developed (PyFR) is now being used around the world by academics and industrial partners. See: https://cassyni.com/s/pyfr
Sectors Aerospace, Defence and Marine

URL https://pyfr.org
 
Description PyFR now has an active international user/developer base, as evidenced by activity on the the PyFR Discourse Forum https://pyfr.discourse.group/ and the PyFR Seminar Series https://cassyni.com/s/pyfr in which users and developers from around the world have presented their latest work using PyFR e.g. https://doi.org/10.52843/cassyni.536kkl (Agency for Defense Development, Korea) https://doi.org/10.52843/cassyni.nqp2sp (Concordia, Canada)) https://doi.org/10.52843/cassyni.wnnxmp (Kyushu, Japan)) https://doi.org/10.52843/cassyni.bq69kg (ISAE-SUPAERO, France)) https://doi.org/10.52843/cassyni.t2ms3w (Texas A&M, USA) https://doi.org/10.52843/cassyni.pw9418 (Aachen, Germany) https://doi.org/10.52843/cassyni.nd09lk (Technion, Israel) https://doi.org/10.52843/cassyni.hkqms2 (University of Hannover) https://doi.org/10.52843/cassyni.br5jn1 (University of Science and Technology of China) https://doi.org/10.52843/cassyni.w15qsz (Leonardo Company)
First Year Of Impact 2021
Sector Aerospace, Defence and Marine
Impact Types Societal,Economic

 
Description BAE Systems 
Organisation BAE Systems
Country United Kingdom 
Sector Academic/University 
PI Contribution Visits. Share findings.
Collaborator Contribution Share findings.
Impact Multi-disciplinary - mathematics, computer science, computational fluid dynamics, aerospace engineering.
Start Year 2013
 
Description Pointwise 
Organisation Pointwise
Country United States 
Sector Private 
PI Contribution Provided them with meshing challenges! And insight into PyFR mesh format.
Collaborator Contribution Added PyFr mesh output to Pointwise meshing software, and have started to make us meshes.
Impact -
Start Year 2015
 
Title PyFR v1.10.0 
Description High-order flow solver. 
Type Of Technology Software 
Year Produced 2020 
Open Source License? Yes  
Impact Ongoing use of the solver. 
URL https://www.pyfr.org/
 
Title PyFR v1.11.0 
Description High-order flow solver. 
Type Of Technology Software 
Year Produced 2021 
Open Source License? Yes  
Impact Ongoing use of the software. 
URL https://www.pyfr.org/
 
Title PyFR v1.12.0 
Description High-order flow solver. 
Type Of Technology Software 
Year Produced 2021 
Open Source License? Yes  
Impact Ongoing use of the software. 
URL https://www.pyfr.org/
 
Title PyFR v1.13.0 
Description High-order flow solver. 
Type Of Technology Software 
Year Produced 2022 
Open Source License? Yes  
Impact Ongoing use of the software. 
URL https://www.pyfr.org/
 
Title PyFR v1.7.6 
Description PyFR is an open-source Python based framework for solving advection-diffusion type problems on streaming architectures using the Flux Reconstruction approach of Huynh. The framework is designed to solve a range of governing systems on mixed unstructured grids containing various element types. It is also designed to target a range of hardware platforms via use of an in-built domain specific language derived from the Mako templating engine. 
Type Of Technology Software 
Year Produced 2018 
Open Source License? Yes  
Impact Further funding. Preliminary industrial adoption. 
URL http://www.pyfr.org
 
Title PyFR v1.8.0 
Description PyFR is an open-source Python based framework for solving advection-diffusion type problems on streaming architectures using the Flux Reconstruction approach of Huynh. The framework is designed to solve a range of governing systems on mixed unstructured grids containing various element types. It is also designed to target a range of hardware platforms via use of an in-built domain specific language derived from the Mako templating engine. 
Type Of Technology Software 
Year Produced 2018 
Open Source License? Yes  
Impact Further funding. Preliminary industrial adoption. 
URL http://www.pyfr.org
 
Title PyFR v1.8.5 
Description PyFR is an open-source Python based framework for solving advection-diffusion type problems on streaming architectures using the Flux Reconstruction approach of Huynh. The framework is designed to solve a range of governing systems on mixed unstructured grids containing various element types. It is also designed to target a range of hardware platforms via use of an in-built domain specific language derived from the Mako templating engine. 
Type Of Technology Software 
Year Produced 2019 
Open Source License? Yes  
Impact
URL http://www.pyfr.org
 
Title PyFR v1.9.0 
Description PyFR is an open-source Python based framework for solving advection-diffusion type problems on streaming architectures using the Flux Reconstruction approach of Huynh. The framework is designed to solve a range of governing systems on mixed unstructured grids containing various element types. It is also designed to target a range of hardware platforms via use of an in-built domain specific language derived from the Mako templating engine. 
Type Of Technology Software 
Year Produced 2019 
Open Source License? Yes  
Impact
URL http://www.pyfr.org