OptiMACS NSC3: The Challenge of Efficient Structural Design Optimisation

Similar to previous Network Short Course (NSC), this NSC will take the form of a mini-symposium on the theme ‘challenge of efficient structural design optimisation’ and explores the challenges and opportunities in this topic. Details are as follow:

  • Title: The challenge of efficient structural design optimisation – Outline of challenges and opportunities
  • Date: 2-3 December 2020
  • Location: Online
  • Entrance fee: free, but registration required at thanh-chung.dinh@risc-software.at

List of confirmed speakers includes:

This talk will present the issues and recent development made at Institut Clément Ader in design, manufacturing and damage modelling vs testing of aeronautic composite structures.

Concerning the design, the large choice of material compositions (i.e. matrix and fibres combinations), architectures and manufacturing processes makes the design process  complex and difficult, as the designers face a hyper-choice of materials and technologies that can be overwhelming. Most of the time, designing composite is understood as, and limited to, the choice of stacking and sizing using the TSAI method or derivative, with or without an optimization scheme. So the GAP composite design methodology (acronym of Geometry, Architecture, Process) which will be presented aims to be a starting point in a composite design process and, in this methodology, we would like to emphasize the importance of creating concepts in sufficient number and variety to tackle the issue of hyper-choice raised above.

Once a design have been selected, to reduce the cost, virtual manufacturing of the complete cycle of an autoclave curing must be develop. A calculation loop including thermo-kinetics, thermo-chemical and thermomechanical implementations has been developed. Refined experimental analysis of the different materials characterics during the cure must be conducted. CFRP stiffened panels are considered with a special focus on the effect on bonding of stiffeners in the distortion of the cured part.

Aeronautic composite part are certified according to a damage tolerance policy and one main issue is the modelling and the efficiency of testing damaged composite structures. The Discrete Ply Modelling (DPM) is based on a mesh following the orientation of the plies. This complex mesh allows taking into account naturally the coupling between intra and inter laminar damages but also splitting. Moreover, it is based only on 13 “true” parameters. This approach was applied successfully for impact and crash on laminates, CAI, residual dent computation, pull-through, edge impact and impact on tapered laminate. This approach was extended successfully to in-plane issues like open hole tension, scaling effects and recently large notches. So the confidence in this modelling strategy is high and the next step is to move from the scale of coupon under uniaxial loading to the scale of technological specimens under complex loadings. This investigation was made through the VERTEX research program. A significant step to Predictive Virtual Testing was achieved and a new pyramid of tests for the certification of aeronautic composite structures can be proposed.

Concerning the design, the large choice of material compositions (i.e. matrix and fibres combinations), architectures and manufacturing processes makes the design process  complex and difficult, as the designers face a hyper-choice of materials and technologies that can be overwhelming. Most of the time, designing composite is understood as, and limited to, the choice of stacking and sizing using the TSAI method or derivative, with or without an optimization scheme. So the GAP composite design methodology (acronym of Geometry, Architecture, Process) which will be presented aims to be a starting point in a composite design process and, in this methodology, we would like to emphasize the importance of creating concepts in sufficient number and variety to tackle the issue of hyper-choice raised above.

Once a design have been selected, to reduce the cost, virtual manufacturing of the complete cycle of an autoclave curing must be develop. A calculation loop including thermo-kinetics, thermo-chemical and thermomechanical implementations has been developped. Refined experimental analysis of the different materials characterics during the cure must be conducted. CFRP stiffened panels are considered with a special focus on the effect on bonding of stiffeners in the distorsion of the cured part.

Aeronautic composite part are certified according to a damage tolerance policy and one main issue is the modeling and the efficiency of testing damaged composite structures. The Discrete Ply Modelling (DPM) is based on a mesh following the orientation of the plies. This complex mesh allows taking into account naturally the coupling between intra and inter laminar damages but also splitting. Moreover, it is based only on 13 “true” parameters. This approach was applied successfully for impact and crash on laminates, CAI, residual dent computation, pull-through, edge impact and impact on tapered laminate. This approach was extended successfully to in-plane issues like open hole tension, scaling effects and recently large notches. So the confidence in this modeling strategy is high and the next step is to move from the scale of coupon under uniaxial loading to the scale of technological specimens under complex loadings. This investigation was made through the VERTEX research program. A significant step to Predictive Virtual Testing was achieved and a new pyramid of tests for the certification of aeronautic composite structures can be proposed.

No abstract available.

No abstract available.

Design optimization poses unique challenges as simulation is of often required to asses promising design candidates. Consequently, runtime becomes a limiting factor, preventing the optimization process from evaluating a large number of design candidates. To address those challenges, machine learning techniques can be applied to support the search processes by estimating the viability of a design candidate beforehand and guide the search towards promising designs, avoiding costly simulations of suboptimal solutions. Further, machine learning can reveal novel understanding of the design task at hand and make the optimization process and the results  more transparent and interpretable.

Structural composites are widely used in lightweight applications because of their competitive response as compared with more traditional metallic alloys. However, failure mechanisms observed in these laminates is multiscale and damage can be observed at different length scales ranging from fiber-matrix interface decohesion, fiber kinking to global delaminations, buckling, etc. Therefore, damage models should be developed with such multiscale perspective to fully account the complete energy absorption mechanisms and provide accurate predictions that can be used in a secure design. We present in this talk, a multiscale perspective of virtual testing of structural composites with a high emphasis on multiscale characterization of material components.

The traditional approaches to the numerical solution of initial-boundary value problems for parabolic or hyperbolic Partial Differential Equations (PDEs) are based on the separation of the discretization in time and space leading to time-stepping methods. This separation of time and space discretizations comes along with some disadvantages with respect to parallelization and adaptivity. To overcome these disadvantages, we consider completely unstructured finite element or isogeometric (B-spline or NURBS) discretizations of the space-time cylinder and the corresponding stable space-time variational formulation of the initial-boundary value problem under consideration. Unstructured space-time discretizations considerably facilitate the parallelization and simultaneous space-time adaptivity. Moving spatial domains or interfaces can easily be treated since they are fixed in the space-time cylinder. Beside initial-boundary value problems for parabolic PDEs, we will also consider optimal control problems constraint by linear or non-linear parabolic PDEs. Here unstructured space-time methods are especially suited since the optimality system couples two parabolic equations for the state and adjoint state that are forward and backward in time, respectively. In contrast to time-stepping methods, one has to solve one big linear or non-linear system of algebraic equations. Thus, the memory requirement is an issue. In this connection, adaptivity, parallelization, and matrix-free implementations are very important techniques to overcome this bottleneck. Fast parallel solvers like domain decomposition and multigrid solvers are the most important ingredients of efficient space-time methods.