Support for Decision Making in Design of Composite Laminated Structures. Part 2: Reduced Parametric Model-Based Optimization

Dulcey, Gilberto Fontecha; Joyot, Pierre; Fadel, Georges

doi:10.1007/s10443-018-9742-9

Cited by 8 publications

(3 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Moreover, in the same equation, the model becomes parametric because the PGD method allows the introduction of functions (P1, P2, P3, P4, P5, P6, P7) in additional domains (p 1 , p 2 , p 3 , p 4 , p 5 , p 6 , p 7 )( F i g .10 Parametrization of the knowledge model). These domains representing parameters are explained in a similar way to previously published work by Fontecha-Dulcey et al [11]. Therefore, the resulting PRBM is reduced because the displacement field is no longer computed from tensor operations relating stresses and strains.…”

Section: Parametric and Reduced Behavior Model (Prbm)mentioning

confidence: 92%

See 1 more Smart Citation

Support for Decision Making in Design of Composite Laminated Structures. Part 1: Parametric Knowledge Model

2018

View full text Add to dashboard Cite

The design process of laminated composites faces two challenges: the engineer designs the product and its morphology, but also, simultaneously, the material. The number of design solutions can be huge since the solution space is very large. Standard CAE systems (CAD, Finite Element Simulation) do not offer to the designer an approach to explore design spaces easily handling design parameters that are intrinsic to the laminate structures: number of plies, layers' constitutive laws, viscoelastic capacity of the matrix and volume fraction of fibers. This paper provides a new model of behavior making explicit these design parameters. This Parametric and reduced Behavior Model (PRBM) allows engineers to make rapid simulations of the product they are creating. Integrated in a meta-model of Knowledge, it is combined to usual specific knowledge that are typically the domain of composite experts and manufacturing experts. Our PRBM is made from a separated numerical method, next enabling (1) a multiscale approach: the engineer can implement reasoning either at the scale of the fiber, or at the scale of the ply, at the scale of the plies interfaces, at the scale of the lamination or at the scale of the structure, or, (2) a multiphysical approach: the engineer can independently manage the mechanical effect of each ply and each interface, either in static or dynamic cases; in the latter, the creeping behavior can be considered. Two simple cases are presented to illustrate the relevance of the PRBM when simulating composite structures: one under a static load and the having a dynamic behavior.Keywords Separated behavior model . Multiscale vs multiphysical approach . Reduced and parametric model . Layer-by-layer simulation . Qualification of simulation model . Laminate composite structure . Meta-modelling of knwoedge * Gilberto Fontecha Dulcey

show abstract

Section: Parametric and Reduced Behavior Model (Prbm)mentioning

confidence: 92%

“…This equation contains both the elastic behavior and the viscous behavior: model with fractional derivatives as presented in Fontecha-Dulcey et al [11].…”

Section: Particularity Of Viscoelasticity At Ply Interfacesmentioning

confidence: 99%

Support for Decision Making in Design of Composite Laminated Structures. Part 1: Parametric Knowledge Model

2018

View full text Add to dashboard Cite

show abstract

“…Also in this case, the design space to be explored is often made of a few varying parameters. [51][52][53][54] In the above outlined situations, FE calculations can be replaced by surrogate analytical models, which allow to save time and reduce computational costs while keeping control of accuracy. [55][56][57][58][59] A spreading methodology consists of the use of radial basis functions (RBFs) to interpolate the numerical results corresponding to a few combinations of material properties (training sets).…”

Section: Introductionmentioning

confidence: 99%

Surrogate analytical models of damage localization in metal matrix composites

Bolzon

Pitchai

Talassi

2023

Journal of Composite Materials

View full text Add to dashboard Cite

Several parameters entering the constitutive laws of metal matrix composites are not amenable to direct measurement. Their determination is therefore demanded to indirect identification procedures, based on the comparison between the output of some experimental works and the results of the iterative simulation of the performed tests. Largely repetitive numerical analyses can also support the optimized design of composite microstructures, the understanding of the actual stress-transfer mechanisms, and virtual testing. Surrogate analytical models can replace effectively non-linear finite element (FE) computations in parametric studies and identification procedures, reducing execution times and costs without compromising the accuracy of the results. In this context, a frequently used approach consists of the interpolation, by means of radial basis functions (RBFs), of data filtered by proper orthogonal decomposition (POD). This methodology is often used to reproduce the smooth output of different mechanical systems. This work is rather focused on the homogenized response of damaging metal matrix composites subjected to strain localization phenomena, and explores the accuracy achievable in these situations by the POD-RBF approximation of more traditional FE analyses. In all considered cases, the POD-RBF results are accurate, and able to distinguish between apparently similar situations. The approach is flexible, and the performance of the surrogate models can be tailored to the requirements of each application. In particular, various analytical approximations can be introduced to support the design of new microstructures and material couplings, and to understand the role of any material and/or geometry imperfections resulting from the production processes.

show abstract

Multi-objective Optimization and Experimental Testing of a Laminated Vertical-Axis Wind Turbine Blade

Trivković

Svorcan

Baltić

et al. 2021

Current Problems in Experimental and Computational Engineering

View full text Add to dashboard Cite

Vertical-axis wind turbines, despite being somewhat uncommon and less efficient renewable energy converters, still offer many advantages to small consumers and possibilities for further improvement. Their rotor is usually made up of fiberglass composite blades that can be optimized both aerodynamically and/or structurally. Although the aerodynamics of vertical-axis wind turbine rotors is unsteady, complex and challenging to simulate, it is possible to make a sufficiently accurate estimation of variable aerodynamic loads acting on the blade and use them for its structural dimensioning. This research combines a multi-objective constrained optimization procedure by particle swarm and finite element method with experimental analyses with the purpose of defining the best blade, i.e. the blade of minimal mass, least tip deflection under the loading case corresponding to the aerodynamically most demanding operational regime, greatest difference between its natural frequencies and rated rotor angular frequency, lowest manufacturing cost and complexity, etc. Imposed constraints also include acceptable failure criteria along the blade. Design variables refer to ply lay-up scheme, i.e. lamina thicknesses and orientations. Final solution, chosen from the obtained Pareto set, was manufactured and experimentally validated. The consistency of strains, measured in several different cases of bending, and corresponding numerical values was mostly below 8%. The study demonstrates the applicability of the employed multi-objective optimization methodology in wind turbine blade design. It also proposes an affordable and structurally reliable composite lay-up scheme specifically designed for small-scale vertical-axis wind turbines.

show abstract

Support for Decision Making in Design of Composite Laminated Structures. Part 2: Reduced Parametric Model-Based Optimization

Cited by 8 publications

References 20 publications

Support for Decision Making in Design of Composite Laminated Structures. Part 1: Parametric Knowledge Model

Support for Decision Making in Design of Composite Laminated Structures. Part 1: Parametric Knowledge Model

Surrogate analytical models of damage localization in metal matrix composites

Multi-objective Optimization and Experimental Testing of a Laminated Vertical-Axis Wind Turbine Blade

Contact Info

Product

Resources

About