Thermal and Reactive Flow Simulation on High-End Computers

Thermal and reactive flows are cross-cutting fundamental disciplines that have found applications in technologies such as aerospace engineering, combustion engines for power generation and propulsion, geothermal energy, solar thermal energy, bioenergy, nanotechnology, chemical engineering and climate science, etc. Research in the field is a prime example where high-end computing (HEC) can have a crucial impact, as the reliability and accuracy of numerical prediction and diagnosis of thermal and reactive flows are directly linked to the computational grid resolution and the size of the time steps. The reason lies with the extremely wide range of time and length scales present in thermal and reactive flows, which are typically turbulent as well. There are 9 to 12 orders of magnitude change between the smallest and the largest length and time scales present in thermal and reactive flows of technical relevance, which should ideally be resolved by experimental measurement or numerical simulation. To study such complex phenomena by experiment alone would be prohibitively expensive and laborious if possible at all. Numerical simulation, on the other hand, offers non-intrusive, virtual "measurement" of all relevant quantities at desired resolution and accuracy, provided sufficient computing power is available. Over the past two decades, the world has first seen gigaflops supercomputers, then teraflops and more recently petaflops machines. The pace of development towards exa-scale HEC platforms has recently quickened. Only last autumn, Tianhe-1A caused a stir by reaching 2.566 petaflops maximum sustained calculation speed, but six months later the K computer achieved an astonishing 8.162 petaflops. At least two HEC machines with 20 petaflops are being built in the world and expected to enter service next year (http://www.top500.org/). The problem is that advance in supercomputing hardware and software, impressive as it appears, has barely kept pace with the research needs. Therefore, frontier research in computational thermal and reactive flows tends to be strongly associated with making use of the latest HEC available.

We believe that HEC is a key enabler of cutting-edge research in thermal and reactive flow flows. The main purpose of this application is to secure HEC resources on HECToR and its successors to support funded research projects in the field. These include: (a) K H Luo (P.I.), EPSRC grant No. EP/I016570/1 (09/2011 - 08/2014), "Tackling Combustion Instability in Low-Emission Energy Systems: Mathematical Modelling. Numerical Simulations and Control Algorithms"; (b) K H Luo (P.I.) and R W Eason, EPSRC grant No. EP/I012605/1 (05/2011 - 05/2014), "Laser-Induced Forward Transfer Nano-Printing Process - Multiscale Modelling, Experimental Validation and Optimization"; and (c) N D Sandham (P.I.), on-going LAPCAT II EU/FP7, "Long-term advanced propulsion concepets and technologies". In addition, the widely used SBLI code first developed by the applicants will be extended to incorporate capabilities for reactive flow simulation. By making use of the world-class computing facility HECToR, the above projects will fulfil the objectives of producing significant, world-leading research results. Examples of world-first simulations will include: (a) largest direct numerical simulation of a turbulent premixed flame interacting with acoustic waves (b) lattice Boltzmann simulation of the complete Laser-Induced Forward Transfer (LIFT) process; and (c) large-eddy simulation of a complete nose-to-tail scramjet engine. These projects are of direct interest to large research communities in aerospace engineering, combustion, nanotechnology, high-performance computing and so on, and will involve a dozen UK and EU companies, which will ensure wide and timely dissemination of research results.

The research consists of two EPSRC-funded and one EU-funded projects which have very close links to industry in the UK and EU. While the pathways to impact for each project are not all repeated here, it is worth highlighting the added impact of the proposed work, with the aid of HEC resources.

WP1 (EPSRC EP/I012605/1): The LIFT technology has the potential to become the next generation ultra-precision laser printer. There is worldwide interest in the topic but the UK and some EU countries are at the forefront of R& D. The work proposed is supported by TNO in Holland and is linked with the EU STREP project e-LIFT. As a result, the project involves four SMEs in the EU, a laser machining company in the UK and one larger company (which manufacture RFID tags for security, product labelling and tracking). The research of all these partners and collaborators is experiment-based, while the proposed simulations here will provide the only theoretical/modelling results. The requested HECToR resources will allow the complete LIFT process, rather than the individual sub-processes, to be simulated, which is essential for direct comparison with experimental results. The fundamental knowledge gained and the modelling tool developed from the integrated experimental and modelling studies will enable cutting-edge research to be exploited by industry in a timely fashion.

WP2 (EPSRC EP/I016570/1): Combustion instability is a major barrier to further improvement in the performance, energy efficiency and emission reduction of a wide range of combustion devices operating near the lean limit. Our industrial partners of the project, Rolls-Royce and Siemens, are keenly aware of the need to address both fundamental questions and applied problems in combustion instability in gas turbines. However, our initial proposal was focused on computationally less demanding fundamental studies due to limitations of both computer hardware and software at the time. With the rapidly enhanced capabilities of HECToR and especially the upgraded DSTAR code through the dCSE support by NAG, it is now possible to simulate combustion instability scenarios that are more relevant to the practical concerns of Rolls-Royce and Siemens. Such studies will be of great value to Rolls-Royce and Siemens, who will relate the research more readily to their design questions. As originally proposed, the research teams will meet representatives from Rolls-Royce and Siemens on a regular basis to discuss technical issues and exploit research results.

WP3 & WP4 (LAPCAT II EU/FP7): With the code developed within WP3 and the requested resources on HECToR, WP4 will be able to conduct world-first simulations of a complete functioning scramjet configuration as in the complete wind tunnel test. Such a capability is potentially of great value to ESA whose only alternative is to conduct a series of very expense wind tunnel tests. It is hoped that the simulations will eventually cut the cost and development time as well as providing enhanced fundamental understanding. The connections with companies such as Reaction Engines Ltd (REL), Fluid Gravity Engineering (FGE) and Gas Dynamics Ltd (GDL) will ensure that the expanded modelling and simulation capabilities are available to UK industry.

Users of SBLI will benefit from the enhanced capability which allows reacting hypersonic flow to be performed on HPC facilities. The in-depth knowledge gained about the flow physics will provide first-hand information to UK aerospace industries, assisting on-going design and future development of more efficient, reliable and environmentally-friendly future high-speed flight vehicles. The highly accurate data to be archived on the database server at the University of Southampton (http://www.dnsdata.afm.ses.soton.ac.uk/) could be used by industry as benchmarks for validating simplified design and diagnostic tools.