Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

Finite element (FE) analysis of curved aerospace laminates has the potential to offset the cost of the expensive experimental testing that is required for certification. However, the number of degrees of freedom necessary to model entire parts (e.g. a wing spar) or to sufficiently resolve defects in smaller parts is extremely high. In order to solve these large problems efficiently a high level of parallelism is needed. For this reason a preconditioner suitable for large anisotropic problems has been implemented in the High Performance Computing library DUNE (Distributed and Unified Numerics Environment). GenEO coupled with an iterative solver such as CG can solve large problems quickly on hundreds of computer cores. This preconditioner can then be used to test the effect of varying the slope of a simple wrinkling defect on the initiation of damage. A convergence analysis of this example problem shows that models with fewer than 4 elements through-thickness in each ply and interface layer do not yield a good approximation of the stresses.

Original language

English

Title of host publication

21st International Conference for Composite Materials

title = "Multiscale Modelling of Lamination Defects in Curved Structures",

abstract = "Finite element (FE) analysis of curved aerospace laminates has the potential to offset the cost of the expensive experimental testing that is required for certification. However, the number of degrees of freedom necessary to model entire parts (e.g. a wing spar) or to sufficiently resolve defects in smaller parts is extremely high. In order to solve these large problems efficiently a high level of parallelism is needed. For this reason a preconditioner suitable for large anisotropic problems has been implemented in the High Performance Computing library DUNE (Distributed and Unified Numerics Environment). GenEO coupled with an iterative solver such as CG can solve large problems quickly on hundreds of computer cores. This preconditioner can then be used to test the effect of varying the slope of a simple wrinkling defect on the initiation of damage. A convergence analysis of this example problem shows that models with fewer than 4 elements through-thickness in each ply and interface layer do not yield a good approximation of the stresses.",

author = "Anne Reinarz and Timothy Dodwell and Timothy Fletcher and Richard Butler and Robert Scheichl",

year = "2017",

month = "8",

language = "English",

booktitle = "21st International Conference for Composite Materials",

}

TY - GEN

T1 - Multiscale Modelling of Lamination Defects in Curved Structures

AU - Reinarz, Anne

AU - Dodwell, Timothy

AU - Fletcher, Timothy

AU - Butler, Richard

AU - Scheichl, Robert

PY - 2017/8

Y1 - 2017/8

N2 - Finite element (FE) analysis of curved aerospace laminates has the potential to offset the cost of the expensive experimental testing that is required for certification. However, the number of degrees of freedom necessary to model entire parts (e.g. a wing spar) or to sufficiently resolve defects in smaller parts is extremely high. In order to solve these large problems efficiently a high level of parallelism is needed. For this reason a preconditioner suitable for large anisotropic problems has been implemented in the High Performance Computing library DUNE (Distributed and Unified Numerics Environment). GenEO coupled with an iterative solver such as CG can solve large problems quickly on hundreds of computer cores. This preconditioner can then be used to test the effect of varying the slope of a simple wrinkling defect on the initiation of damage. A convergence analysis of this example problem shows that models with fewer than 4 elements through-thickness in each ply and interface layer do not yield a good approximation of the stresses.

AB - Finite element (FE) analysis of curved aerospace laminates has the potential to offset the cost of the expensive experimental testing that is required for certification. However, the number of degrees of freedom necessary to model entire parts (e.g. a wing spar) or to sufficiently resolve defects in smaller parts is extremely high. In order to solve these large problems efficiently a high level of parallelism is needed. For this reason a preconditioner suitable for large anisotropic problems has been implemented in the High Performance Computing library DUNE (Distributed and Unified Numerics Environment). GenEO coupled with an iterative solver such as CG can solve large problems quickly on hundreds of computer cores. This preconditioner can then be used to test the effect of varying the slope of a simple wrinkling defect on the initiation of damage. A convergence analysis of this example problem shows that models with fewer than 4 elements through-thickness in each ply and interface layer do not yield a good approximation of the stresses.

UR - http://www.iccm-central.org/Proceedings/ICCM21proceedings/papers/3771.pdf

M3 - Conference contribution

BT - 21st International Conference for Composite Materials