Strategic Package: Centre for Predictive Modelling in Science and Engineering

Lead Research Organisation: University of Warwick
Department Name: Sch of Engineering

Abstract

Predictive modelling and uncertainty quantification are more than just another research direction relevant to science and engineering. They constitute a different way of thinking that impacts practically all aspects of scientific and engineering analysis and design. Rather than deriving deterministic answers to complex problems, distributions (error-bars) are obtained that account for our incomplete and often inaccurate information about the problems of interest.
Modelling of uncertain physical and geometric properties in the large scales, intrinsic randomness in the small scales or incorporation of stochastic closures necessitate the use of stochastic simulations in many multiscale problems, heavily taxing computational resources. The emerging exascale computers will allow stochastic simulations of realistic systems but analysing petabyte data bases with existing methods can be too inefficient, and full-scale models cannot be used in optimisation, design or real-time control. To this end, the University of Warwick proposes to establish a new centre focused on Predictive Modelling and uncertainty quantification with the mission to develop computational, mathematical and statistical methodologies for uncertainty quantification (UQ) and predictive modelling applied to a broad range of applications.

The new centre will emphasises a dynamic integration of several fundamental research areas including the following:
1) Formulation, analysis and numerical solution of stochastic multiscale/multiphysics systems exploring synergies between mathematical, statistical and machine learning approaches.
2) Data-driven development of certified stochastic reduced-order models.
3) Development of a probabilistic approach to systems synthesis, design, optimisation and control under uncertainty.
This proposed work will develop methods for quantifying uncertainty in multiscale simulations driven by experimental data collected at different scales. In order to account for the high-dimensional nature of the input uncertainties and resulting stochastic outputs, it is proposed to develop reduced-order surrogate models both for the input uncertainties and for the output of multiscale systems. Topics (1)-(3) are unifying themes that are common to many disciplines. Bringing together existent strengths within the University of Warwick to address these problems would allow simultaneously to impact multiple applications and move forward the problem of resolving bottlenecks in uncertainty quantification (e.g. high-dimensionality, limited data, long term time integration, rare events).

Planned Impact

Predictive modelling is a key enabler for industry; it speeds up the design process, reduces the need for physical testing, reducse risk and enables rapid optimisation. The net effect is a significant reduction in cost and improved product development. The broad remit of the centre will ensure impact into a wide range of industrial sectors such as aerospace, energy, automotive and healthcare.

This research is of particular relevance and importance to the aerospace industry and most major corporations, e.g. General Electric, United Technologies, Boeing, Rolls Royce USA, Siemens USA, have significant activity in the area. We note that aerospace is a strategic priority for the UK government; in March 2013 the department of Business Innovation and Skills published its aerospace strategy, committing £1B to fund a research partnership with industry, with an additional £1B matched by industry. The importance of predictive modelling is now generally widely accepted by UK industry as highlighted by the recent TSB call 'Towards Zero prototyping" programme that is designed to promote the use of predictive modelling in engineering design and manufacturing and therefore reduce the need for physical prototypes. Additionally, 'advanced materials' is one of the UK government's 'eight great technologies' and an area where the UK has an international lead. UQ is already established in the US in the area of 'predictive materials science' and will increasingly impact on UK industry.

The centre will focus on the development of generic data-driven techniques and methodologies; impact will be delivered through tailored projects in collaboration with a wide range of industrial partners. The School of Engineering and WMG have extensive links with over 500 companies and hence rapid uptake of relevant methodologies can be facilitated. The centre will also proactively engage with industry to promote the use of predictive modelling techniques.

Publications

10 25 50

publication icon
Bianchini F (2016) Modelling defects in Ni-Al with EAM and DFT calculations in Modelling and Simulation in Materials Science and Engineering

publication icon
Bianchini F (2016) Modelling defects in Ni-Al with EAM and DFT calculations in Modelling and Simulation in Materials Science and Engineering

publication icon
Bitzek E (2015) Atomistic aspects of fracture in International Journal of Fracture

 
Description • Developed innovative techniques to quantify uncertainties in electronic structure calculations for computing material properties. This allows us to provide error bars on computed properties (e.g. bulk properties, band gaps, phase diagrams, etc.) induced by limited data to train surrogate models (cluster expansions). Such techniques are valuable in not only reliable property calculation but also for the design of new materials with desired properties.
• Inverse problems were addressed for computing properties with multiresolution features (e.g. permeability or porosity fields in a reservoir engineering problem). The scale of estimation for such problems is computed directly from data. The developments include new mathematical models and algorithms using deep learning with wavelets but also potential new applications with technological significance (e.g. calculating properties of reservoirs using limited pressure or flow data, computing microstructures that result in desired properties, etc.)
• We have extended the use of surrogate models in electronic structure calculations of bulk systems to open systems (surfaces and interfaces). This has allowed to compute properties of nanowires, as well to design nanostructures with desired properties (e.g. electrical conductivity).
• We explored synergies between machine learning in the context of probabilistic graphical models and approximate inference algorithms to address uncertainty quantification of dynamical systems with multiple time scales. These techniques are important on predicting macroscopic behaviour from fine scale data as for example is the case in atomistic and molecular simulations of materials.
• We are exploring the use of deep learning in the context of uncertainty quantification of multiscale systems using limited data. The challenges are many but there are potential rewards such as automatic recognition of the coarse grained variables and their distribution.
• Organized two international meetings on Predictive Modelling (Predictive Multiscale Materials Modelling, Isaac Newton Institute, Cambridge, December 1-4, 2015 http://www.turing-gateway.cam.ac.uk/uqm_dec2015 and Big Data and Predictive Computational Modeling, IAS-TUM, Munich, May 18-20, 2015 http://www.tum-ias.de/bigdata2015/)
Exploitation Route Developed algorithms for computing material properties with error bars can be used to design new materials with unprecedented properties (e.g. nanowires with minimal electrical conductivity). Uncertainty quantification algorithms can be used as black box simulators for building data-driven predictive models in multiple applications using both experimental and simulation data.
Sectors Aerospace, Defence and Marine,Construction,Digital/Communication/Information Technologies (including Software),Electronics,Energy,Healthcare,Manufacturing, including Industrial Biotechology

URL http://www2.warwick.ac.uk/wcpm/
 
Description ARCHER eCSE
Amount £43,747 (GBP)
Funding ID eCSE11-7 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start 09/2017 
End 02/2018
 
Description EPSRC Standard Mode
Amount £739,715 (GBP)
Funding ID EP/P022065/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start 04/2017 
End 03/2021
 
Description EPSRC standard grant
Amount £442,958 (GBP)
Funding ID EP/R043612/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start 10/2018 
End 03/2022
 
Description Industrial CASE award
Amount £83,296 (GBP)
Funding ID 17000185 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start 10/2017 
End 09/2021
 
Description Leverhulme Research Project Grant
Amount £385,476 (GBP)
Organisation The Leverhulme Trust 
Sector Academic/University
Country United Kingdom
Start 04/2018 
End 03/2022
 
Description Royal Society Research Grant
Amount £14,764 (GBP)
Funding ID RG160691 
Organisation The Royal Society 
Sector Academic/University
Country United Kingdom
Start 04/2017 
End 03/2018
 
Description Standard grant
Amount £823,000 (GBP)
Funding ID EP/R012474/1 
Organisation Engineering and Physical Sciences Research Council (EPSRC) 
Sector Academic/University
Country United Kingdom
Start 03/2018 
End 02/2021
 
Description Participation in Innovate UK/KTN Special Interest Group in Uncertainty Quantification and Management 
Organisation Knowledge Transfer Network
Country United Kingdom 
Sector Charity/Non Profit 
PI Contribution Warwick Centre for Predictive Modelling is hosting an upcoming Industry Study Group on uncertainty quantification and management (UQ&M) for high value manufacturing in December 2017. Members of the Centre have also participated in other events organised by the UQ&M SIG, including: - JR Kermode, invited talk at KTN Uncertainty Quantification and Management for Industry, CFMS, Bristol, Aug 2017 - JR Kermode, Keynote at KTN event on Compound Semiconductor Simulation and Modelling, Birmingham, April 2017 - JR Kermode, Session chair at KTN event on Uncertainty Quantification and Management in High Value Manufacturing, London, June 2015
Collaborator Contribution The KTN facilitated the events listed above, and is co-funding the WCPM-based Study Group to take place in December 2017.
Impact Multi-disciplinary: academic, government and industrial members. Industrial sectors include aerospace, automotive, infrastructure.
Start Year 2015