Self-adaptive Model Order Reduction Models for Large Nonlinear Finite Element Models

Research proposals

  • Research area

    Operations and remote autonomous monitoring

  • Institution

    University of Sheffield

  • Research project

    Self-adaptive Model Order Reduction Models for Large Nonlinear Finite Element Models

  • Lead supervisor

    Dr Charles Lord (Lecturer – Mechanical Engineering, University of Sheffield)

  • Supervisory Team

    Dr Jem Rongong (Senior Lecturer - Mechanical Engineering, University of Sheffield)
    Dr Charles Lord (Lecturer – Mechanical Engineering, University of Sheffield)

Project Description:

Project Description: Model order reduction is a process by which mathematical models are reduced in the number of degrees of freedom to simplify the numerical expense while at the same time still maintaining the rich content of the model. Several model order reduction techniques already exist, but are generally limited to a particular dominant mode or set of modes. These models are further limited by their bias toward linear representations which do not lend their usefulness or efficiency to nonlinearities. For structures, such as those comprised of thin components (e.g. wind turbine blades), this is particularly important due to their inherent nonlinearities. The proposed research here is the development of a self-adaptive reduced order model. The novelty lies in two main parts: (i) in the ability of the reduced order model being self-adaptive switching between which model and the coupling of models to account for extended complex mode shapes (mixed bending and torsion) and (ii) that it can be used and directed to localised areas and does not have to contain global information or accountability. From an industrial standpoint, reduced order models are used frequently. Currently, there is little standardisation or agreement on how, when, and to what extent a reduced order model should be utilised; the decision gate usually comes from ‘trial and error’. As part of this research, it would be expected that the self-adaptivity of the model would define these guidelines.

Download all research proposals here