Digital twins for improved dynamic design
Lead Research Organisation:
University of Sheffield
Department Name: Mechanical Engineering
Abstract
The aim of this proposal is to create a robustly-validated virtual prediction tool called a "digital twin". This is urgently needed to overcome limitations in current industrial practice that increasingly rely on large computer-based models to make critical design and operational decisions for systems such as wind farms, nuclear power stations and aircraft. The digital twin is much more than just a numerical model: It is a "virtualised" proxy version of the physical system built from a fusion of data with models of differing fidelity, using novel techniques in uncertainty analysis, model reduction, and experimental validation. In this project, we will deliver the transformative new science required to generate digital twin technology for key sectors of UK industry: specifically power generation, automotive and aerospace. The results from the project will empower industry with the ability to create digital twins as predictive tools for real-world problems that (i) radically improve design methodology leading to significant cost savings, and (ii) transform uncertainty management of key industrial assets, enabling a step change reduction in the associated operation and management costs. Ultimately, we envisage that the scientific advancements proposed here will revolutionise the engineering design-to-decommission cycle for a wide range of engineering applications of value to the UK.
Planned Impact
This project will have economic impact in the offshore wind, nuclear power, aerospace and automotive industries. The development of new digital twin technology will enable companies working in these sectors to design and operate their products and assets with lower design and operational costs. There may also be benefits in terms of extending operational life. In terms of societal impact, this will contribute to lower energy costs, reduced CO2 emissions, and employment security in the UK. The development of new knowledge, both in the academic domain and translated to industry will happen in parallel with the training and development of a cohort of expert early career researchers. These expert researchers are a key resource for the UK skills base, and they will contribute to the ongoing competiveness of the industrial sectors mentioned above.
Organisations
- University of Sheffield (Lead Research Organisation)
- Siemens (Germany) (Project Partner)
- Airbus (United Kingdom) (Project Partner)
- Stirling Dynamics (United Kingdom) (Project Partner)
- EDF Energy (United Kingdom) (Project Partner)
- Leonardo (United Kingdom) (Project Partner)
- LOC Group (London Offshore Consultants) (Project Partner)
- Romax Technology (United Kingdom) (Project Partner)
- Ultra Electronics (United Kingdom) (Project Partner)
- Schlumberger (United Kingdom) (Project Partner)
Publications
Hughes A.J.
(2020)
On decision-making for adaptive models combining physics and data
in Proceedings of ISMA 2020 - International Conference on Noise and Vibration Engineering and USD 2020 - International Conference on Uncertainty in Structural Dynamics
Hughes A.J.
(2021)
A risk-based active learning approach to inspection scheduling
in International Conference on Structural Health Monitoring of Intelligent Infrastructure: Transferring Research into Practice, SHMII
Hübler C
(2023)
Robust model updating in structural dynamics using a new non-implausibility-motivated optimisation approach
in Mechanical Systems and Signal Processing
Imtiaz Q
(2022)
CuO-based materials for thermochemical redox cycles: the influence of the formation of a CuO percolation network on oxygen release and oxidation kinetics.
in Discover chemical engineering
Jalali H
(2022)
A generalization of the Valanis model for friction modelling
in Mechanical Systems and Signal Processing
Jamia N
(2021)
An equivalent model of a nonlinear bolted flange joint
in Mechanical Systems and Signal Processing
Jamia N.
(2020)
Numerical and experimental investigations of nonlinearities in bolted joints
in Proceedings of ISMA 2020 - International Conference on Noise and Vibration Engineering and USD 2020 - International Conference on Uncertainty in Structural Dynamics
Lam B
(2021)
Ten questions concerning active noise control in the built environment
in Building and Environment
Lei CL
(2020)
Considering discrepancy when calibrating a mechanistic electrophysiology model.
in Philosophical transactions. Series A, Mathematical, physical, and engineering sciences
Liu X
(2019)
Simultaneous normal form transformation and model-order reduction for systems of coupled nonlinear oscillators.
in Proceedings. Mathematical, physical, and engineering sciences
Maizel T
(2022)
Advances in Service and Industrial Robotics - RAAD 2022
Marrocchio R
(2021)
Waves in the cochlea and in acoustic rainbow sensors
in Wave Motion
Martin R
(2021)
Response to the comment Confidence in confidence distributions!
in Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
Massa TB
(2023)
Fusel oil reaction in pressurized water: characterization and antimicrobial activity.
in 3 Biotech
Molés-Cases V
(2022)
Weighted pressure matching with windowed targets for personal sound zones.
in The Journal of the Acoustical Society of America
Nayek R
(2021)
On spike-and-slab priors for Bayesian equation discovery of nonlinear dynamical systems via sparse linear regression
in Mechanical Systems and Signal Processing
Nayek R
(2023)
Identification of piecewise-linear mechanical oscillators via Bayesian model selection and parameter estimation
in Mechanical Systems and Signal Processing
Neild S
(2020)
Accounting for Quasi-Static Coupling in Nonlinear Dynamic Reduced-Order Models
in Journal of Computational and Nonlinear Dynamics
Nicolaidou E
(2022)
Nonlinear mapping of non-conservative forces for reduced-order modelling
in Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
Nicolaidou E
(2020)
Indirect reduced-order modelling: using nonlinear manifolds to conserve kinetic energy.
in Proceedings. Mathematical, physical, and engineering sciences
Nicolaidou E
(2021)
Detecting internal resonances during model reduction
in Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences
PITCHFORTH D
(2023)
PHYSICS-INFORMED GAUSSIAN PROCESSES FOR WAVE LOADING PREDICTION
Poole J
(2022)
On statistic alignment for domain adaptation in structural health monitoring
in Structural Health Monitoring
Poole J.
(2021)
On Normalisation for Domain Adaptation in Population-Based Structural Health Monitoring
in Structural Health Monitoring 2021: Enabling Next-Generation SHM for Cyber-Physical Systems - Proceedings of the 13th International Workshop on Structural Health Monitoring, IWSHM 2021
Ritto T
(2022)
Reinforcement learning and approximate Bayesian computation for model selection and parameter calibration applied to a nonlinear dynamical system
in Mechanical Systems and Signal Processing
Ritto T
(2022)
A transfer learning-based digital twin for detecting localised torsional friction in deviated wells
in Mechanical Systems and Signal Processing
Rogers (T.)
(2018)
A grey box model for wave force prediction.
Rogers (T.J.)
(2019)
Identification of a Duffing oscillator using particle Gibbs with ancestor sampling.
Rogers T
(2020)
Probabilistic modelling of wind turbine power curves with application of heteroscedastic Gaussian Process regression
in Renewable Energy
Rooker (T.)
(2018)
Predicting geometric tolerance thresholds in a five-axis machining centre.
Ruffini V
(2020)
Real-Time Hybrid Testing of Strut-Braced Wing Under Aerodynamic Loading Using an Electrodynamic Actuator
in Experimental Techniques
Sadeghi J
(2020)
Robust propagation of probability boxes by interval predictor models
in Structural Safety
Sadeghi J
(2019)
Efficient training of interval Neural Networks for imprecise training data
in Neural Networks
Scott (D.)
(2018)
Machine learning for energy load forecasting.
Shalini S
(2016)
1000-fold enhancement in proton conductivity of a MOF using post-synthetically anchored proton transporters.
in Scientific reports
Description | In the first half of the project, the activities have been deliberately weighted towards developing the new science required to achieve the project objectives. Full details are given in the mid-term report, but the main scientific highlights so far are: A1. A logic-based framework for the analysis and synthesis of digital twin A2. Domain adaption and multi-task learning for digital twin applications A3. Development of new uncertainty propagation techniques in dynamic sub-structuring A4. Improved robustness of nonlinear parameter identification using control-based continuation As the project moves into its second half, there will be an increased emphasis on impact. In the mid-term report we highlighted the following achievements so far: I1. An "observational digital twin" for an aircraft ground-steering system I2. Work with Airbus on the Semi Aeroelastic Hinge (Albatross-one) I3. Code libraries & Puffin software for intrusive uncertainty quantification (UQ) objects and operations I4. Digital Twin Operational Platform (DTOP) for the 3-storey structure (now Use Case U0) In addition to the above points, fundamental research work has also been carried out in the key areas relating to verification & validation, uncertainty analysis, control, jointing, design and hybrid testing. |
Exploitation Route | The outcomes of this award are being used by the project industry partners, and we are in discussion with other potential users of the results from the project. In particular we are planning to further develop the open source software release on Github during 2021 https://github.com/Digital-Twin-Operational-Platform/Cristallo |
Sectors | Aerospace, Defence and Marine,Digital/Communication/Information Technologies (including Software),Energy |
URL | https://digitwin.ac.uk/ |
Description | Team members at Swansea have been collaborating with Airbus on the Semi Aeroelastic Hinge (Albatross-one). The main idea behind this project is the use of folding wingtip devices that increase the aspect ratio and therefore improve the aircraft efficiency & performance. Specifically, the linked PhD student started working on this in January 2019. There are continuing discussions with Dr AC about the comparison of results of the developed model at Swansea with the high-fidelity model at Airbus. There is also the possibility of using physical test data from Airbus. Digital Twin Operational Platform (DTOP): A web-based operational platform has been written for application of the digital twin concept to applications. This platform has been written using the Flask Python programming framework. It includes input from all partners and themes across the consortium. The code uses Flask/Python and Javascript to present real-time information on the web visualisation interface. This user-interface is used to showcase the algorithms developed in each of the Themes. We are now exploring use this platform for industry applications, and have released a early version as an open software project on Github: https://github.com/Digital-Twin-Operational-Platform/Cristallo Team members at the University of Sheffield are working with a Business Advisor to explore the possibility of creating a spin-out company |
First Year Of Impact | 2020 |
Sector | Aerospace, Defence and Marine |
Impact Types | Economic |
Description | Advice given on the development of the Gemini Principles for digital twins |
Geographic Reach | National |
Policy Influence Type | Participation in a guidance/advisory committee |
URL | https://www.cdbb.cam.ac.uk/DFTG/GeminiPrinciples |
Description | Advice on the research landscape for Digital Twins in the UK for the Futures Team at the Government Office for Science (BEIS) & UKRI. |
Geographic Reach | National |
Policy Influence Type | Participation in a guidance/advisory committee |
Title | Burst Random Amplitude Ramp : 0.4V |
Description | This is a zip file that contains 10 repetitions for burst random (multiple white noise) excitation at 0.4V LMS output voltage. This test is performed on the starboard wing of the BAE T1A Hawk housed at the Labratory for Verification and Validation (lvv.ac.uk). The wing is excited using a single shaker and data is recorded using 55 uni-accelerometers and the excitation load cell. The significance of the naming conventions is explained in the "ORDA_Readme.pdf" file. This file also contains the code needed to load the file into python for analysis called "hdf5_loader.py". This contains a function called load_hdf5 that takes the path of the hdf5 file and loads it into a single dictionary. The dictionary is separated into "Meta" for the testing parameters and the sensor names to identify the sensor of interest. For the sensor names, the naming convention is described in the readme file along with what types of data are included and the citation reference. The uploaded *.zip file contains all the repetitions for the testing parameters (nominally 10 repetitions). This test was performed under the Alan Turing Institute funded project Digital Twins for High-Value Engineering Systems (DTHIVE) with continuing support from the EPSRC funded project Digital Twins for Improved Dynamic Design (DigiTwin). For more information, please contact the PI, Professor David Wagg. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
URL | https://figshare.shef.ac.uk/articles/dataset/Burst_Random_Amplitude_Ramp_0_4V/21739466 |
Title | Burst Random Amplitude Ramp : 0.4V Repetition 1 |
Description | First repetition for burst random (white noise) excitation at 0.4V LMS output voltage. This test is performed on the starboard wing of the BAE T1A Hawk housed at the Labratory for Verification and Validation (lvv.ac.uk). The wing is excited using a single shaker and data is recorded using 55 uni-accelerometers and the excitation load cell. The significance of the naming conventions is explained in the "ORDA_Readme.pdf" file. This file also contains the code needed to load the files into python for analysis. The testing parameters is stored is the "BR_AR_1_test_info.pickle" file. The data is separated into one pickle file per accelerometer. These files contain both time and frequency information with more details in the readme file. This test was performed under the Alan Turing Institute funded project Digital Twins for High-Value Engineering Systems (DTHIVE) with continuing support from the EPSRC funded project Digital Twins for Improved Dynamic Design (DigiTwin). For more information, please contact the PI, Professor David Wagg. |
Type Of Material | Database/Collection of data |
Year Produced | 2022 |
Provided To Others? | Yes |
URL | https://figshare.shef.ac.uk/articles/dataset/Burst_Random_Amplitude_Ramp_0_4V_Repetition_1/21400008 |
Title | Burst Random Amplitude Ramp : 0.8V |
Description | This is a zip file that contains 10 repetitions for burst random (multiple white noise) excitation at 0.8V LMS output voltage. This test is performed on the starboard wing of the BAE T1A Hawk housed at the Labratory for Verification and Validation (lvv.ac.uk). The wing is excited using a single shaker and data is recorded using 55 uni-accelerometers and the excitation load cell. The significance of the naming conventions is explained in the "ORDA_Readme.pdf" file. This file also contains the code needed to load the file into python for analysis called "hdf5_loader.py". This contains a function called load_hdf5 that takes the path of the hdf5 file and loads it into a single dictionary. The dictionary is separated into "Meta" for the testing parameters and the sensor names to identify the sensor of interest. For the sensor names, the naming convention is described in the readme file along with what types of data are included and the citation reference. The uploaded *.zip file contains all the repetitions for the testing parameters (nominally 10 repetitions). This test was performed under the Alan Turing Institute funded project Digital Twins for High-Value Engineering Systems (DTHIVE) with continuing support from the EPSRC funded project Digital Twins for Improved Dynamic Design (DigiTwin). For more information, please contact the PI, Professor David Wagg. |
Type Of Material | Database/Collection of data |
Year Produced | 2023 |
Provided To Others? | Yes |
URL | https://figshare.shef.ac.uk/articles/dataset/Burst_Random_Amplitude_Ramp_0_8V/21878745 |
Title | Burst Random Amplitude Ramp : 1.2V |
Description | This is a zip file that contains 10 repetitions for burst random (multiple white noise) excitation at 1.2V LMS output voltage. This test is performed on the starboard wing of the BAE T1A Hawk housed at the Labratory for Verification and Validation (lvv.ac.uk). The wing is excited using a single shaker and data is recorded using 55 uni-accelerometers and the excitation load cell. The significance of the naming conventions is explained in the "ORDA_Readme.pdf" file. This file also contains the code needed to load the file into python for analysis called "hdf5_loader.py". This contains a function called load_hdf5 that takes the path of the hdf5 file and loads it into a single dictionary. The dictionary is separated into "Meta" for the testing parameters and the sensor names to identify the sensor of interest. For the sensor names, the naming convention is described in the readme file along with what types of data are included and the citation reference. The uploaded *.zip file contains all the repetitions for the testing parameters (nominally 10 repetitions). This test was performed under the Alan Turing Institute funded project Digital Twins for High-Value Engineering Systems (DTHIVE) with continuing support from the EPSRC funded project Digital Twins for Improved Dynamic Design (DigiTwin). For more information, please contact the PI, Professor David Wagg. |
Type Of Material | Database/Collection of data |
Year Produced | 2023 |
Provided To Others? | Yes |
URL | https://figshare.shef.ac.uk/articles/dataset/Burst_Random_Amplitude_Ramp_1_2V/21878766 |
Title | Burst Random Amplitude Ramp : 1.6V |
Description | This is a zip file that contains 10 repetitions for burst random (multiple white noise) excitation at 1.6V LMS output voltage. This test is performed on the starboard wing of the BAE T1A Hawk housed at the Labratory for Verification and Validation (lvv.ac.uk). The wing is excited using a single shaker and data is recorded using 55 uni-accelerometers and the excitation load cell. The significance of the naming conventions is explained in the "ORDA_Readme.pdf" file. This file also contains the code needed to load the file into python for analysis called "hdf5_loader.py". This contains a function called load_hdf5 that takes the path of the hdf5 file and loads it into a single dictionary. The dictionary is separated into "Meta" for the testing parameters and the sensor names to identify the sensor of interest. For the sensor names, the naming convention is described in the readme file along with what types of data are included and the citation reference. The uploaded *.zip file contains all the repetitions for the testing parameters (nominally 10 repetitions). This test was performed under the Alan Turing Institute funded project Digital Twins for High-Value Engineering Systems (DTHIVE) with continuing support from the EPSRC funded project Digital Twins for Improved Dynamic Design (DigiTwin). For more information, please contact the PI, Professor David Wagg. |
Type Of Material | Database/Collection of data |
Year Produced | 2023 |
Provided To Others? | Yes |
URL | https://figshare.shef.ac.uk/articles/dataset/Burst_Random_Amplitude_Ramp_1_6V/21878775 |
Title | Burst Random Amplitude Ramp : 2.0V |
Description | This is a zip file that contains 10 repetitions for burst random (multiple white noise) excitation at 2.0V LMS output voltage. This test is performed on the starboard wing of the BAE T1A Hawk housed at the Labratory for Verification and Validation (lvv.ac.uk). The wing is excited using a single shaker and data is recorded using 55 uni-accelerometers and the excitation load cell. The significance of the naming conventions is explained in the "ORDA_Readme.pdf" file. This file also contains the code needed to load the file into python for analysis called "hdf5_loader.py". This contains a function called load_hdf5 that takes the path of the hdf5 file and loads it into a single dictionary. The dictionary is separated into "Meta" for the testing parameters and the sensor names to identify the sensor of interest. For the sensor names, the naming convention is described in the readme file along with what types of data are included and the citation reference. The uploaded *.zip file contains all the repetitions for the testing parameters (nominally 10 repetitions). This test was performed under the Alan Turing Institute funded project Digital Twins for High-Value Engineering Systems (DTHIVE) with continuing support from the EPSRC funded project Digital Twins for Improved Dynamic Design (DigiTwin). For more information, please contact the PI, Professor David Wagg. |
Type Of Material | Database/Collection of data |
Year Produced | 2023 |
Provided To Others? | Yes |
URL | https://figshare.shef.ac.uk/articles/dataset/Burst_Random_Amplitude_Ramp_2_0V/21878787 |
Title | Real-Time Hybrid Testing of Strut-Braced Wing Under Aerodynamic Loading Using an Electrodynamic Actuator |
Description | Dataset for: "Real-Time Hybrid Testing of Strut-Braced Wing Under Aerodynamic Loading Using an Electrodynamic Actuator". Authors: V. Ruffini, C. Szczyglowski, D.A.W. Barton, M. Lowenberg, S.A. Neild Journal: Experimental Techniques. |
Type Of Material | Database/Collection of data |
Year Produced | 2020 |
Provided To Others? | Yes |
URL | https://data.bris.ac.uk/data/dataset/3r79185tdczil24hfny9m1i7y6/ |
Description | Digital Twin Workshop |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Study participants or study members |
Results and Impact | This workshop was hosted jointly between the University of Sheffield Advanced Manufacturing Research Centre and Faculty of Engineering, and was held in the state-of-the-art AMRC Factory 2050 on the Sheffield Advanced Manufacturing Park. The purpose of the workshop was to bring together leading UK based researchers currently working on topics related to digital twin. The contributions included both application specific, and basic research presentations. The event helped to promote the 'Gemini Principles', the 'National digital twin', and the EPSRC funded `DigiTwin' project. |
Year(s) Of Engagement Activity | 2019 |
URL | https://www.eventbrite.co.uk/e/digital-twin-workshop-tickets-77298831887# |
Description | Engineering Digital Twins in Practice |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Industry/Business |
Results and Impact | The purpose of the workshop was to bring together leading UK based researchers currently working on topics related to digital twins. We were particularly keen to showcase the latest research activities for the key research groups around the UK. We showcased both application specific, and basic research type talks |
Year(s) Of Engagement Activity | 2023 |
Description | The 3rd Sheffield Workshop on Structural Dynamics |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Industry/Business |
Results and Impact | Over 200 people attended a 3 day online workshop hosted by Prof Keith Worden (DigiTwin Investigator/Member of Management committee) The main body of the workshop had three distinct themes: physics-informed machine learning, population-based SHM, and digital-twins and their validation. The agenda for the workshop allowed time for question and discussions for all attendees. The workshop also provided an opportunity to showcase updates and research within the DigiTwin project. |
Year(s) Of Engagement Activity | 2020 |
URL | https://www.eventbrite.co.uk/e/the-3rd-sheffield-workshop-on-structural-dynamics-online-event-ticket... |
Description | The 4th Sheffield Workshop on Structural Dynamics |
Form Of Engagement Activity | Participation in an activity, workshop or similar |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Postgraduate students |
Results and Impact | Just under 100 people attended a 3 day online workshop hosted by Prof Keith Worden (DigiTwin Investigator/Member of Management committee) The main body of the workshop had three distinct themes: physics-informed machine learning, population-based SHM, and digital-twins and their validation. The agenda for the workshop allowed time for question and discussions for all attendees. The workshop also provided an opportunity to showcase updates and research within the DigiTwin project. |
Year(s) Of Engagement Activity | 2022 |