Multidisciplinary design optimization of a 3D shock control bump
Lead Research Organisation:
Imperial College London
Department Name: Aeronautics
Abstract
TBC
Organisations
People |
ORCID iD |
| Matthew Santer (Primary Supervisor) | |
| Dilaksan Thillaithevan (Student) |
Studentship Projects
| Project Reference | Relationship | Related To | Start | End | Student Name |
|---|---|---|---|---|---|
| EP/R513052/1 | 01/10/2018 | 30/09/2023 | |||
| 2297591 | Studentship | EP/R513052/1 | 01/10/2018 | 31/03/2022 | Dilaksan Thillaithevan |
| Description | Over the last three years my research has focussed on the application of multiscale methods to structural optimisation problems. In particular, I have focussed heavily on tackling issues which hamper the use of multiscale structures in real world applications. A common limitation of multiscale optimisation formulations presented in literature is the failure to consider the stresses during the optimisation process. Without considering stresses, there is no way of knowing whether an optimised design will be capable of sustaining the design loads once the part is manufactured. The inclusion of stress constraints in traditional topology optimisation frameworks has been widely considered and many techniques have been proposed to overcome the issues that arise when considering stresses. However, in multiscale methods, there are a limited number of established techniques for considering stresses during the optimisation and most of these utilise the homogenised stresses, which are insufficient to capture the behaviour of the individual microstructures. In my research I have proposed a method for overcoming this issue. By utilising linear superposition, the real microscale stresses are computed based on the macroscale strains to allow the inclusion of stress constraints in the optimisation process. By using full-scale finite element analysis, I have been able to validate my model and I have also demonstrated its efficacy using multiscale optimisation with the inclusion of stress constraints. Another area of interest has been the consideration of material uncertainties during the optimisation process. When designs are manufactured, uncertainties arise in the final geometry due to the tolerances of the manufacturing process. This is particularly concerning for optimised designs which by definition should be located within in a local minimum or maximum, which means that any small changes to the geometry can cause large unwanted changes in the behaviour of the design. To overcome this is it necessary to consider the uncertainties during the design process by performing robust optimisation. In my research I have developed a method for introducing material uncertainties within the multiscale optimisation framework by perturbing the radii of the microstructures directly to simulate the variations in geometry that occur during the manufacturing process. This is something that to my knowledge has not been considered previously. By coupling my uncertainty model with a robust optimisation framework, I have been able to generate designs which are up to 70% more robust compared to standard deterministic designs, without sacrificing the overall performance of the design. To reduce the large computational expense that is typically associated with performing robust optimisation I have been able to incorporate polynomial chaos expansions and random field discretisation within my optimisation framework. This has led to a 97% reduction in the time required to compute the low order statistics for the problems tackled in my research. |
| Exploitation Route | The multiscale optimisation method can be utilised in real world structural optimisation problems to generate designs which are more efficient (in terms of weight, cost or strength) and better suited to the application that traditional designs. In particular, these methods may be useful within the aerospace industry where the strength to weight ratio is of great concern. By utilising the multiscale stress model proposed, it will be possible to generate parts of the aerospace industry which are both lighter and stronger than traditional designs, while ensuring the designs are capable of sustaining the required loads. Furthermore, by considering the material uncertainties during the optimisation process, the designers can be confident that the behaviour of the part or design will not be significantly compromised by the errors that occur during manufacturing stage. |
| Sectors | Aerospace, Defence and Marine |