Quantification of Uncertainty Toolkit for Engineering (QUTE) | Computing the unknown
04 Apr 2018
Yes
-  

 

 

Simpler, cheaper and faster methods that can reproduce the results of a more accurate and more computationally demanding model to within a defined tolerance level

Yes

​​​​

 

Abstract internet computer technology-dreamstime-Vs1489_34481865 (1900x1728).jpg

​For many engineering applications, simple, fast and cheap simulation methods can reproduce the results of a more accurate and more computationally demanding model to within a defined tolerance level. To make virtual product design a reality we must be able to quantify and control the size of the errors made by the model, understanding the sensitivity outputs to the values of input parameters.

Our QUTE toolkit can build ‘surrogate models’ – simpler, cheaper and faster methods capable of reproducing results of more accurate and computationally demanding models to within a defined tolerance level.  It automates the process of parametric uncertainty studies, including selection of parameters to sample (using methods such as Polynomial Chaos and multi-level Monte Carlo), submission of simulation tasks to computing platforms, retrieval and analysis of results. QUTE is available under a commercial software license, or can be deployed on our high performance computing (HPC) resources​​ via our Platform-as-a-Service offering. Customisation of QUTE to support specific HPC systems can be carried out through collaborative R&D projects.​

​Applications

  • Virtual product design and optimisation
  • Modelling the operation of industrial processes​

​Benefits

  • ​Better understanding of performance variation
  • Discovery of rare product failure models
  • Cost reduction by reducing the number of simulations required


To explore and improve the accuracy and efficiency of your computational models – contact us​ 

Publications

​​Zimon M, Elisseev V, Sawko R, Antao S, Jordan K. Uncertainty Quantification-as-a-Service. In 28th Annual International Conference on Computer Science and Software Engineering (CASCON 18), Toronto, Canada, 29-31 Oct 2018. https://epubs.stfc.ac.uk/work/42016095​​




Contact: