Institution news
Q: Please briefly explain your role and how you use CFD.
Tom Dunstan (TD): I’m a Senior Scientist at the UK Met Office working on sub-grid process modelling with a focus on the effects of orography (hills and mountains). CFD is at the core of all our operational weather and climate models, as well as high-resolution research codes and more specialised applications such as atmospheric dispersion modelling and space weather.
Q: Are there any specific modelling challenges which arise in your sector or industry?
TD: The tails of the distribution are very important to us, ie. extreme events that are rare but can have a large impact, and so producing models that are correct in an average sense is not good enough. From a technical point of view we face the challenge of continuing to efficiently scale these increasingly complex models on non-standard or heterogeneous compute platforms.
Q: What particular developments or new applications are you excited about for the next 5 years?
TD: The potential for data driven methods to accelerate our existing models, produce more portable code and provide new insights into underlying physical processes.
Q: What are the key areas of focus for your presentation?
TD: Some examples of how we are fusing data-driven methods and machine learning with traditional simulation, and how we hope to develop this over the next few years.
Q: Are there other speakers on the programme who you are particularly excited to hear from, and why?
TD: I’m interested in wind energy and urban flow modelling so looking forward to hearing from Ruth Shilston from RWDI and Sergio Campobasso from Lancaster University. Outside work my real passion is sailing, so I’m excited to hear from Max Starr about his work with INEOS Team UK.
Q: What specifically are you hoping to get out of this seminar more generally?
TD: An insight into the CFD challenges faced in areas outside my own expertise.
Next-Generation CFD seminar takes place online on 23 March 2021.
The growth of cloud computing technology is unlocking the potential to perform more processor intensive models and analysis including using machine learning. This one-day online seminar brings together the foremost practitioners from multiple industries and academia to showcase advanced CFD applications and the innovative technologies that are enabling greater complexity and speed.
For more information, full programme and to book your place to get involved, please visit the event website.