CFD @ NWC13
The following is a summary, prepared by the NAFEMS CFD Working Group and the German NAFEMS CFD Advisory Board, of the CFD content and discussions at the recent World Congress.
On Tuesday afternoon there was an open discussion on topics relating to Computational Fluid Dynamics. The session was chaired by Anthony Mosquera, member of the CFD working group, with Althea de Souza, leader of the CFD working group, and Gerhard Müller, moderator of the DACH region CFD advisory board (CAB). The session started with a brief presentation by the session chairman, Anthony Mosquera, about the NAFEMS CFD working group. This was then followed by a presentation from the moderator of the new DACH regional CFD Advisory Board, Gerhard Muller. The following topics were proposed for discussion: technical trends, requirements, role of CFD in the development process, CFD used by designers, ROI, processes, qualifications, training, quality, infrastructure, IT, cloud computing, Open Source in CFD.
The first question was about current areas that the groups are working on. For the CFD working group these include: two books on heat transfer - one a general why do and the other on radiation, a designers guide to CFD, a book on how to validate industrial CFD, a re-write of the now rather out of date guide to meshing, a book on how to analyse and process CFD output, work on the next edition of the NAFEMS International Journal of Industrial Case Studies and of course ongoing work to encourage CFD-related articles for Benchmark. There are also two seminars being planned in the UK; one on Acoustics planned for autumn 2013 and the other on Quality and Reliability for Industrial CFD (the 6th in the series) planned for spring 2014. There are also two seminars planned for Germany: CFD in system simulation, planned for 12/13th November 2013 and the next seminar will be 20/21st March 2014 and is likely to be one one of the following topics: CFD for managers, CFD in the cloud or CFD for designers. Both seminars will be held in Bamberg in Bavaria.
It was explained that one of the greatest challenges faced by the NAFEMS working groups is finding authors who are suitably qualified and experienced to write publications but that also have time to do the writing. NAFEMS publications are typically written on a part-time basis, fitted around full-time work commitments and as a result, usually take several years to be completed. In recent years, as some of the more long standing members of the NAFEMS community have reached retirement, they have continued to be involved with the analysis community through NAFEMS. Anyone in the room who was either themselves in this category or who had colleagues in this category were asked to consider writing for NAFEMS. Authoring is a paid activity and most of the working groups will consider proposals from authors regardless of current invitations to tender, if the author proposes a topic they wish to write about.
There was some discussion around meshing. Some years ago, it was suggested that the CFD working group was wasting their time writing about meshing because the problem would go away (and at the time it was suggested that this was already the case for much structural analysis). However, we haven't seen this and it is still a hot topic, affecting both the quality of results and the ability to achieve solutions too. There have been changes though, many software tools have become a lot more robust and able to handle lower quality meshes, others now use semi or even fully automated meshing, sometimes entirely hiding the mesh and meshing activity from the user. We are also starting to see mesh-less tools (such as lattice-boltzmann solvers) become more widely used. However, it was agreed that some of the issues have diminished significantly, such as handling poor quality CAD, coping with large scale differences and creating sufficiently large meshes for large problems or higher resolutions results (including those required for LES or DNS).
The group comprised of both industrial users of CFD tools and representatives of several CFD vendors and the difficulty in comparing software was raised. There was some discussion around CFD benchmarks for detailed specific applications and it was suggested that blind benchmark cases could be provided as a challenge to vendors. This will be raised as a topic for the CFD Working Group to investigate. An idea was proposed for a NAFEMS forum to allow users to post problems (applications) and ask which software could be used to simulate these cases. There was mixed enthusiasm for this as several vendors felt that users could approach them directly they would be able to work with them on their specific requirements. However, there is a challenge for the users as there are now very many CFD codes available, both commercial tools and open source or free tools and to consider all of them would be a huge task. It was pointed out that the NAFEMS website allows member companies to provide details of their software and capabilities and that it is a good place for users to start to find out more information.
Simulation quality and accuracy are important considerations when selecting the most appropriate tool to use and the importance of simulations being fit for purpose was raised. Sometimes users or even vendors can be distracted by validating simulation results, forgetting that experimental data may not be accurate or reliable or that the information available from a CFD simulation is much greater than that usually provided by measurements (all the variables at all the points are in the CFD result). Often a high level of accuracy is less important than a quick result, so a lower fidelity result that can be obtained in less time may be more useful. Between them, the range of CFD tools available encompass this spread of requirements and it should be taken into account when selection the tools to use for any specific set of circumstances. There is a NAFEMS book available titled, 'How to ensure your CFD simulation is fit for purpose'. A possible seminar around a call for challenges could ask users to submit challenging cases and then vendors could be asked to present their fit for purpose solutions and to justify how they are fit for purpose.
The increasing demand for computing resources was noted. For aero-acoustics, in particular, turn-around times were too long and a two orders of magnitude reduction in time to solution was required. Unless there are algorithm improvements to speed up aero-acoustic calculations that are fit for purpose, reducing time for the computation would require access to more computing cores than currently used. Availability of resources on the ‘cloud’ was raised more generally as an opportunity for analysis calculations with both software vendors and industrial users having mixed views on how open they would be to accessing a public resource.
These topics will be considered in the meetings of the NAFEMS CFD Working group and the German CFD Advisory Board.
All participants were thanked for their participation and contributions and both the CFD working group and DACH CAB will consider the points raised in their future activities.If you would like to contact either of the groups, they can be e-mailed at CFD@nafems.org or at CAB@nafems.de