Dr Jens-Dominik Mueller
Dipl-Ing, MSc, PhD
Research Group News
Unfortunately listed as 'former U of Bath' student in the article. https://www.bbc.co.uk/news/uk-england-somerset-58083385
Salvatore Auriemma has passed his PhD viva on "Development of parametric CAD models for gradient-based aerodynamic shape optimisation" with minor corrections. Gradient-based methods are the most efficient approach for soptimisation with very many design variables and has become a standard in aerodynamic shape optimisation of e.g. aircraft wings and … [more]
Rejish Jesudasan successfully defended his thesis on "An Adaptive Parametrisation Method for Shape Optimisation Using Adjoint Sensitivities". Owing to COVID-19 / SARS-CoV-2, the viva was conducted over video conferencing, but even with noise of his ceiling fan (essential in his Indian home at this time), the finer details of his algorithm … [more]
Siamak Akbarzadeh successfully defended his thesis on using Automatic Differentiation software tools to obtain adjoint variants of the incompressible SIMPLE scheme. Adjoint methods are essential for numerical optimisation of many-parameter systems with expensive models, such as CFD. Siamak's work is a step forward in deriving robust adjoints for the widely … [more]
Orest Mykhaskiv has successfully defended his PhD on gradient-based shape optimisation. His thesis significantly advanced the field by including the geometric CAD model description inside the design loop. This breakthrough allows to integrate effective gradient-based optimisation seamlessly into industrial design workflows. Orests PhD was funded by the European Commission through … [more]
Researchgate has tallied up 4500 reads for the papers of the group. Many thanks to my co-authors, well done all!
Our groundbreaking paper Algorithmic differentiation of the Open CASCADE Technology CAD kernel and its coupling with an adjoint CFD solver the first to differentiate a full CAD system and a breakthrough for gradient-based optimisation with CAD-based parametrisations has achieved 100 reads on researchgate. Jens. [more]
Pavanakumar Mohanamuraly successfully defended his thesis on "Fast Adjoint-assisted Multilevel Multifidelity Method for Uncertainty Quantification of the Aleatoric Kind" His work uses Monte-Carlo methods to compute the effect of uncertainties due to manufacturing. Current methods work well for up to handful of uncertain parameters, but their computation cost increases dramatically … [more]
The paper by P. Mohanamuraly (QMUL), L. Hascoet (INRIA, France) and J.-D. Mueller (QMUL) on Seeding and adjoining zero-halo partitioned parallel scientific codes has been accepted for publication. The paper resolves the important issue of how to initialise the computation of a gradient with modern Automatic Differentiation software tools … [more]
Researchgate says: 50 reads for this paper. Such reads are the first steps to get cited, it is very important to make pre-prints available online. All the best, Jens. Adjoint-Based Aerodynamic Optimisation of Wing Shape Using Non-uniform Rational B-Splines September 2019 DOI: 10.1007/978-3-319-89890-2_10 In book: Evolutionary and Deterministic Methods … [more]
The EU has agreed provide EUR225,000 to fund the KAFKA project: "Efficient Computational Methods for Active Flow Control Using Adjoint Sensitivities" Active flow control aims to change the character of the flow by suction, blowing or pulsating jets. Such jets may e.g. be able to replace bulky car spoilers … [more]
Prof. Shenren Xu of NPWU has visited QMUL to discuss continued collaboration around the the open-source STAMPS code that originates from Jens' group.
Dr. Jens-Dominik Mueller gave an invited presentation on "Shape optimisation using AD of complete CFD workflows, including CAD geometry" at the "Algorithmic Differentiation Workshop" organised by the Institut du Physique du Globe of the Sorbonne University, Paris. [more]
Airbus UK is a partner for a 3.5y CASE studentship with J.-D. Mueller's fluid optimisation group. The PhD project for Gordon Lindsay will apply our CAD-based gradient-enabled shape parametrisation tools to perform multi-disciplinary shape optimisation of winglets. [more]
Researchgate says, our paper CAD-based shape optimisation with CFD using a discrete adjoint Xu, Jahn, Mueller is now cited 30 times. I'd encourage all to post your preprints on ResearchGate, this is important for dissemination and being cited. All the best, Jens. [more]
Paper published in Desalination, impact factor 6.6: Sensitivity analysis and gradient-based optimisation of feed spacer shape in reverse osmosis membrane processes using discrete adjoint approach YangWang, Wei He, Jens-DominikMüller Desalination, 2019, Volume 449, Pages 26-40 Well done to Yang and Wei. Jens. [more]
Jens-Dominik Mueller was invited to present at the Von Karman Institute Lecture Series on "Introduction to Optimization and Multi-disciplinary Design", September 10-14, 2018, in Brussels. The von Karman Institute organizes each year 8 to 12 one-week Lecture Series on specialized topics in the field of aerodynamics, fluid mechanics and heat transfer with application … [more]
Jens gave a 3-day Summer School, "An introduction to gradient-based optimisation, adjoint methods and Automatic Differentiation", organised by the Special Interest Group on Numerical Optimisation with Fluids at Cambridge, 8-10 August 2018. 20 students from the UK and the EU attended. The SIG is offering the training material at http://signumopt.sems.… [more]
8 May 2014:
Next group meeting, 10:30-13:00 in the SEMS Seminar Room, SCR afterwards.
5 May 2014:
Welcome to Mattia Oriani, ex-MSC from QMUL, now ESR on the About Flow project hosted by partner ESI. Mattia will be working on higher-order accurate mimetic finite differences for incompressible flow equations, and will have his 8M progress examination while he is here.
Arriving: 05 May 2014, departing: 29 May 2014
14 April 2014:
Welcome to Alaa Taftaf, ESR on the About Flow project hosted by partner INRIA. Alaa wll be with us from 14 April to 6 June, working on fixed-point loop support in Tapenade. She will be looking at our gpde and mgopt codes, understand how we implement fixed-point loops in them, and develop Tapenade's capability to offer this 'out of the box'.