Dynamically parameterising CAD models using sensitivities for optimisation
Lead Research Organisation:
Queen's University Belfast
Department Name: Sch Mechanical and Aerospace Engineering
Abstract
To design complex products, engineers need to consider and optimise many different attributes. In aerospace, optimisation mainly considers both structural (e.g. displacements, accelerations) and fluid (e.g. pressures acting on a body) attributes. One of the main factors which can impact performance is product shape, which affects a number of disciplines. When changing the shape of the design the options are to change the analysis model (i.e. a mesh) or the geometry model which represents the design. The preferred option is to optimise the geometry model as the result is integrated with the wider design enterprise (e.g. it can also be used for manufacturing considerations). This is particularly true if the geometry model is a feature based CAD model (e.g. Catia V5 or Siemens NX). In a feature based CAD system, the object shape is modified using the parameters which define the features that make up the model itself.
One challenge is that the variables which define the shape of the design and control how it can change, may not actually be well suited for the disciplines driving the optimisation. This means that regardless of how much effort the optimiser puts in, it will not be possible to reach a truly optimum design. This three year project will ensure the parameterisation is suited to optimisation by investigating robust methodologies to automatically insert new features into the CAD model, for which the associated parameters will be new optimisation variables. This will rely on robust and efficient new methods for computing multi-disciplinary sensitivities. The project benefits from collaboration with a major UK industrial partner (Airbus) and developers of key analysis software (DLR). They will assist in researching a new capability with the overall aim of "delivering a step change in the configuration, time to market and performance of new designs." The following objectives have been set:
1. Implement strategies for improving CAD parameterisations for multi-disciplinary optimisation by automatically inserting features into the model based on sensitivity.
2. Investigate efficient and robust methodologies for computing aero-structural sensitivities. This will see a novel approach to the calculation of the sensitivities.
3. Develop strategies for coupling and coherently meshing solid and fluid models. This is a key piece of research required in any aero-structural analysis.
4. Combine aero-structural sensitivities with CAD parameterisation strategies, in an automated optimisation framework, for a range of test cases. This is where the benefits of the work will be demonstrated to industry.
5. Quantify the decrease in time to market and increase in performance due to this research.
Application areas for this research include the design of products which require the optimisation of complex shapes. It will be particularly relevant in industries where feature based CAD systems underpin the design process, and where the physics of the problem may identify the need for shape features which may not be apparent when the CAD models are being setup. An example may be where the surface sensitivities suggest the need for a winglet, but where the parameterisation of a basic wing does not include the parameters to allow such a feature to form. Benefits include:
1. the ability to discover new, optimum, configurations. This is a route to innovative design solutions which will help to keep the UK as a world leader in the design and manufacture of complex products;
2. improved product performance due to the improved optimisation variables (CAD parameters) created based on the requirements of the physics of the problem. For air travel this will result in more environmentally friendly aircraft and lower travel prices;
3. reduced development times due to an automated and efficient optimisation processes, leading to new, better performing, products being available sooner;
One challenge is that the variables which define the shape of the design and control how it can change, may not actually be well suited for the disciplines driving the optimisation. This means that regardless of how much effort the optimiser puts in, it will not be possible to reach a truly optimum design. This three year project will ensure the parameterisation is suited to optimisation by investigating robust methodologies to automatically insert new features into the CAD model, for which the associated parameters will be new optimisation variables. This will rely on robust and efficient new methods for computing multi-disciplinary sensitivities. The project benefits from collaboration with a major UK industrial partner (Airbus) and developers of key analysis software (DLR). They will assist in researching a new capability with the overall aim of "delivering a step change in the configuration, time to market and performance of new designs." The following objectives have been set:
1. Implement strategies for improving CAD parameterisations for multi-disciplinary optimisation by automatically inserting features into the model based on sensitivity.
2. Investigate efficient and robust methodologies for computing aero-structural sensitivities. This will see a novel approach to the calculation of the sensitivities.
3. Develop strategies for coupling and coherently meshing solid and fluid models. This is a key piece of research required in any aero-structural analysis.
4. Combine aero-structural sensitivities with CAD parameterisation strategies, in an automated optimisation framework, for a range of test cases. This is where the benefits of the work will be demonstrated to industry.
5. Quantify the decrease in time to market and increase in performance due to this research.
Application areas for this research include the design of products which require the optimisation of complex shapes. It will be particularly relevant in industries where feature based CAD systems underpin the design process, and where the physics of the problem may identify the need for shape features which may not be apparent when the CAD models are being setup. An example may be where the surface sensitivities suggest the need for a winglet, but where the parameterisation of a basic wing does not include the parameters to allow such a feature to form. Benefits include:
1. the ability to discover new, optimum, configurations. This is a route to innovative design solutions which will help to keep the UK as a world leader in the design and manufacture of complex products;
2. improved product performance due to the improved optimisation variables (CAD parameters) created based on the requirements of the physics of the problem. For air travel this will result in more environmentally friendly aircraft and lower travel prices;
3. reduced development times due to an automated and efficient optimisation processes, leading to new, better performing, products being available sooner;
Planned Impact
By introducing automated design optimisation procedures, this research will impact industrial productivity and accelerate the development of new products. The automatic placement of new features in a feature based CAD model, guided by multi-disciplinary sensitivity information, means that many of the design decisions about where to place features and parameters, which are currently made by a designer, will now be informed by physics based methods. As such, the skillset required by, and role of, the designer will change. This will reduce the expense of designing new, complex products which will reduce overall product cost. As the positioning and choice of the features to be added will be based on rational metrics, they may indicate a route to performance improvement not previously considered. The benefit of such information could be considerable if an innovative solution is produced which offers a large performance improvement. The new parameters will provide optimisation variables capable of producing innovative designs that can outperform alternatives obtained using the original, immutable, parameters, resulting in a commercial advantage and reduce time to market.
The UK will benefit because the processes will mean its position as a world leader in the design and manufacture of complex products (such as aircraft) will be enhanced. The automated addition of new features will mean that innovative design solutions will emerge, resulting in a commercial advantage. The improved integration of structural and fluid domains will have a significant impact on all complex engineering projects, where the consideration of both disciplines is essential. The ability to automate the addition of features, and the removal of the roadblocks when integrating structural and fluid models, will mean that the time to run an optimisation cycle for a new product will be shorter. This benefit can be exploited by products undergoing more optimisation cycles, thereby improving performance, or by products reaching the market more quickly, resulting in significant commercial advantage. As manufacturing is an essential part of the UK economy, the advantages brought about by this increased capability will be significant. This could translate into a significant competitive advantage for Airbus and UK PLC in the commercial airline market and allow subsequently lower airfares.
Environmental benefit will be generated through better performing products, which in terms of aerospace translates into lower emissions and noise. The global nature of aerospace, and the contribution it makes to environmental pollution, means that a percentage reduction in the emissions from an aerospace product will have a significant global impact.
Societal benefit will be realised not just through the UK economic and environmental benefits, but also by the improvement in performance. The use of automated toolsets will result in much lower product and usage costs, therefore a reduction in travel fares and the associated increase in social benefits that more accessible transport systems create.
The PDRAs will benefit from being involved in an innovative, industrially relevant and well supported project. This will develop them as individuals, researchers and future technology leaders.
The PhD student will have the significant benefit of conducting a PhD in a collaborative environment, with the PDRAs playing a key role in developing and mentoring the PhD student. The association of the project with key partners, and the opportunity to be seconded to them, will add another dimension to the student's knowledge and skill set.
The UK will benefit because the processes will mean its position as a world leader in the design and manufacture of complex products (such as aircraft) will be enhanced. The automated addition of new features will mean that innovative design solutions will emerge, resulting in a commercial advantage. The improved integration of structural and fluid domains will have a significant impact on all complex engineering projects, where the consideration of both disciplines is essential. The ability to automate the addition of features, and the removal of the roadblocks when integrating structural and fluid models, will mean that the time to run an optimisation cycle for a new product will be shorter. This benefit can be exploited by products undergoing more optimisation cycles, thereby improving performance, or by products reaching the market more quickly, resulting in significant commercial advantage. As manufacturing is an essential part of the UK economy, the advantages brought about by this increased capability will be significant. This could translate into a significant competitive advantage for Airbus and UK PLC in the commercial airline market and allow subsequently lower airfares.
Environmental benefit will be generated through better performing products, which in terms of aerospace translates into lower emissions and noise. The global nature of aerospace, and the contribution it makes to environmental pollution, means that a percentage reduction in the emissions from an aerospace product will have a significant global impact.
Societal benefit will be realised not just through the UK economic and environmental benefits, but also by the improvement in performance. The use of automated toolsets will result in much lower product and usage costs, therefore a reduction in travel fares and the associated increase in social benefits that more accessible transport systems create.
The PDRAs will benefit from being involved in an innovative, industrially relevant and well supported project. This will develop them as individuals, researchers and future technology leaders.
The PhD student will have the significant benefit of conducting a PhD in a collaborative environment, with the PDRAs playing a key role in developing and mentoring the PhD student. The association of the project with key partners, and the opportunity to be seconded to them, will add another dimension to the student's knowledge and skill set.
Publications
Agarwal D
(2022)
CAD-based Adjoint Optimization Using Other Components in a CAD Model Assembly as Constraints
in Computer-Aided Design and Applications
Agarwal D
(2022)
Aerodynamic Shape Optimisation Using Parametric CAD and Discrete Adjoint
in Aerospace
Lecallard B
(2021)
Mesh and Geometry Manipulations for Optimization and Inverse Design
Lecallard B.
(2021)
Mesh and geometry manipulation for optimization and inverse design
in AIAA Scitech 2021 Forum
Marques S
(2023)
Nonintrusive Aerodynamic Shape Optimisation with a POD-DEIM Based Trust Region Method
in Aerospace
Marques S.
(2021)
Non-intrusive aerodynamic shape optimisation with a discrete empirical interpolation method
in AIAA Scitech 2021 Forum
Shannon T
(2022)
Generalized Bezier components and successive component refinement using moving morphable components
in Structural and Multidisciplinary Optimization
Description | 1. Discovered a new surface mesh movement tool which is robust against boundary topology changes. This overcomes a significant issue in mesh movement techniques for industrial CAD models. 2. Demonstrated the use of a single parameter set to link and control the aerodynamic and structural models of an industrially complex wing model. 3. Demonstrated how to run an optimisation process to make a parametric CAD model fit to an existing CAD model shape. This simplifies the re-parametrization of design models. 4. Demonstrate how computing the shape change between two CAD models can be used to compute models for intermediate variants and similar models with different levels of simplification. 5. Extended the application of Galerkin projection type ROMs to transonic problems. 6. New LU-SGS adjoint solver based on the AUSM flux function 7. Established a machine learning based mapping between a parametric shape and physical quantities. 8. Developed a non intrusive class of ROMs for aerodynamic shape optimisation, based on the Euler and RANS equations. |
Exploitation Route | We will build upon the outcomes of this project through further research projects. We will also work with our industrial partners to attempt to exploit the work in their setting. We have published our work and made codes available for other researchers to pick up and use the outcomes of this project. |
Sectors | Aerospace Defence and Marine |
Description | This project has demonstrated how performance sensitivity information can be used to drive a generative design process. Such learning has fed into out understanding of Bio inspired design processes in follow-on grants such as Re-imagining Engineering Design Programme Grant and Biohaviour - Building the Blind Watchmaker. All of these projects have contributed to "Smart Design" being a key pillar of the ~£100m Advanced Manufacturing Innovation Centre (AMIC) which has part of the Belfast Region City Deal. |
First Year Of Impact | 2021 |
Sector | Aerospace, Defence and Marine,Agriculture, Food and Drink,Digital/Communication/Information Technologies (including Software),Electronics,Manufacturing, including Industrial Biotechology,Transport |
Impact Types | Economic |
Description | COLIBRI - Collaboration Across Business Boundaries |
Amount | £9,010,939 (GBP) |
Funding ID | 113296 |
Organisation | Innovate UK |
Sector | Public |
Country | United Kingdom |
Start | 04/2020 |
End | 05/2023 |
Description | Re-Imagining Engineering Design: Growing Radical Cyber-Physical-Socio Phenotypes |
Amount | £7,355,902 (GBP) |
Funding ID | EP/V007335/1 |
Organisation | Engineering and Physical Sciences Research Council (EPSRC) |
Sector | Public |
Country | United Kingdom |
Start | 04/2021 |
End | 04/2026 |
Title | CATIA V5 parametric CRM model |
Description | This is a CAD model of the CRM, with a novel parameterisation. |
Type Of Material | Computer model/algorithm |
Year Produced | 2020 |
Provided To Others? | No |
Impact | None to date. |
Description | Project partnership with Airbus |
Organisation | Airbus Group |
Country | France |
Sector | Academic/University |
PI Contribution | The research team demonstrated state of the art methods and techniques on their models. |
Collaborator Contribution | The partners attended project meetings, provided an industrially relevant test case and invited us to internal meetings and conferences. |
Impact | It produced a paper presented at an AIAA conference. |
Start Year | 2017 |
Description | Project partnership with DLR |
Organisation | German Aerospace Centre (DLR) |
Department | DLR Braunschweig |
Country | Germany |
Sector | Public |
PI Contribution | We were table to give feedback on their research tools and make suggestions for future methods of working. |
Collaborator Contribution | License for Tau including aero-structural adjoint capability and the provision of training and support in the use of Tau. Also, participation at steering group meetings and workshop, up to 6 days per year over the three years |
Impact | NA |
Start Year | 2017 |
Title | Model reduction method for CFD solvers |
Description | A methodology was developed to generate reduced models of large CFD (computational fluid dynamics) solver and update them within an optimisation framework. The method is based on the Galerkin projection technique and further employs algorithmic differentiation to efficiently build and solve a system of equations based on the fluid Euler equations. |
Type Of Technology | Software |
Year Produced | 2019 |
Impact | This work resulted in two conference papers and a journal paper is currently under review. |
Title | Model reduction method for CFD solvers |
Description | A new methodology was developed to generate reduced models of large CFD (computational fluid dynamics) solver that does not require modification of the underlying full order model. A trust region model management was built around the new technology to perform highly effective surrogate based aerodynamic shape optimisation. The method is based on the DEIM technique and results so far show a halving to computational costs with respect to state-of-the-art adjoint based methods. |
Type Of Technology | Software |
Year Produced | 2020 |
Impact | A conference paper and journal paper are under development. |
Title | Surface mesh movement |
Description | This software can be used to move the surface mesh on the boundary of a model to the boundary of a perturbation of that model. |
Type Of Technology | Software |
Year Produced | 2019 |
Impact | None to date. |
Description | High lift workshop in Cranfield University |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Industry/Business |
Results and Impact | Workshop to discuss CFD for high lift devices, design, test and optimisation. |
Year(s) Of Engagement Activity | 2019 |
Description | Invited Notes from the Front presentation |
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 | I gave a presentation on the QUB work on generative design, including this research project. |
Year(s) Of Engagement Activity | 2022 |
URL | https://www.assessinitiative.com/congress/assess-2022/#program |
Description | Presentation at Airbus DiPart conference 2017 |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | National |
Primary Audience | Industry/Business |
Results and Impact | We presented the project and its context. |
Year(s) Of Engagement Activity | 2017 |
URL | https://cfms.org.uk/airbus-dipart-2017/ |
Description | Presentation at Airbus DiPart conference 2018 |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Industry/Business |
Results and Impact | Presented the latest work from the project |
Year(s) Of Engagement Activity | 2018 |
Description | Studey visit and presentations at Zhejiang University China |
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 | This was a visit to one of the leading institutions for CAD modelling in China. The PDRA visited three departments to learn from this and to talk about this work. |
Year(s) Of Engagement Activity | 2019 |
Description | Tau use group workshop |
Form Of Engagement Activity | A talk or presentation |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Professional Practitioners |
Results and Impact | Overview of project given and offers of support received. |
Year(s) Of Engagement Activity | 2018 |
Description | Two presentations at DiPart 2019 |
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 | This is an Airbus workshop. This project had two presentations at the conference. |
Year(s) Of Engagement Activity | 2019 |
Description | Visit to State Key Lab of CAD&CG and the School of Computer Science and Engineering, Zhejiang University |
Form Of Engagement Activity | A formal working group, expert panel or dialogue |
Part Of Official Scheme? | No |
Geographic Reach | International |
Primary Audience | Other audiences |
Results and Impact | A visit was made to the State Key Lab of CAD&CG and the School of Computer Science and Engineering, Zhejiang University. This is the main department on China into CAD modelling research, and so was an important dialogue for the research being conducted. |
Year(s) Of Engagement Activity | 2019 |