Skip to Main content Skip to Navigation

Conception et optimisation multi-échelle de structures composites en intégrant une approche de modélisation global-local haute-fidelité

Abstract : This thesis focuses on the development of a suitable global-local modelling approach, based on higher-order theories, which is integrated into the multi-scale two-level optimisation strategy (MS2LOS) for the optimal design of composite structures developed by M. Montemurro and co-workers at the I2M laboratory in Bordeaux.The development of an appropriate global-local modelling approach is a fundamental brick to be integrated into the MS2LOS in order to correctly capture (at each relevant scale) the effective mechanical response of the structure with a considerable reduction in computational time and cost. The global-local modelling approach aims to replace the equivalent single-layer approach with a layered approach based on higher-order theories within the framework of the Carrera Unified Formulation (CUF) for local optimisation of the composite.The extension of the MS2LOS in designing CSC stiffened panels is also addressed in this work. The MS2LOS, in fact, is applied to solve the least-weight design problem of a stiffened composite panel. At the macroscopic level, the structure is modelled as an equivalent single-layer plate and the goal is to find the optimum value of the design variables (geometrical and mechanical) to minimise the mass of the panel meeting the set of imposed design requirements (feasibility, manufacturing, stiffness, buckling, etc.) without introducing simplified hypothesis on the mechanical behaviour. At the mesoscopic scale, the aim is to find at least one stacking sequence that meets the optimum design variables resulting from the structural optimisation carried out at the macroscopic scale.Then, the MS2LOS is applied to the optimisation of VSCs. In particular, the first-level problem of the MS2LOS is solved to maximise the first buckling load of the VSC structure determining the optimal distribution of the VSC stiffness properties at the macroscopic scale and satisfying the requirements of the problem. In this optimisation, a deterministic algorithm is used to find the optimum using determined analytically by exploiting the properties of the polar formalism and of the B-spline surfaces which are used in the framework of the MS2LOS.Lastly, the integration of the global-local modelling approach based on layer-wise higher-order theories in the MS2LOS is provided. The objective is to identify and isolate the regions of the model which require more precise investigations during the first optimisation step of the MS2LOS and to analyse these regions of interest using the global-local approach developed in the CUF framework. To this end, two FE models are developed and interfaced. The first one is a low-fidelity FE (LF-FE) model generated by means of commercial software. In the context of the LF-FE model, each laminate constituting the structure is modelled as an equivalent single-layer plate, whose macroscopic behaviour is described in the PPs space. The most critical regions of the LF-FE model, i.e. the so-called zones of interest (ZOIs), are identified by means of a suitable criterion formulated in the PPs space. Then, a high-fidelity FE (HF-FE) model based on layer-wise theories is generated for the critical ZOI to assess the local structural responses which are integrated into the problem formulation.
Complete list of metadata
Contributor : ABES STAR :  Contact
Submitted on : Thursday, June 23, 2022 - 2:58:11 PM
Last modification on : Friday, August 5, 2022 - 2:54:01 PM
Long-term archiving on: : Saturday, September 24, 2022 - 7:04:09 PM


Version validated by the jury (STAR)


  • HAL Id : tel-03702945, version 1


Giacinto Alberto Fiordilino. Conception et optimisation multi-échelle de structures composites en intégrant une approche de modélisation global-local haute-fidelité. Génie mécanique [physics.class-ph]. HESAM Université; Politecnico di Torino, 2022. Français. ⟨NNT : 2022HESAE004⟩. ⟨tel-03702945⟩



Record views


Files downloads