Low Carbon Vehicle Technology Project
(LCVTP)
WS 7 Lightweight Structures - Technology Review
November 17th 2011
Gerain
t
Willia
ms
WMG
Developing robust predictive models for
composites CAE
Martin Gambling and Oliver Tomlin
17th November 2011
Presentation Overview





Composite Modelling Issues
Modelling Methods
Data Capture and Calibration
Design Optimisation Techniques
Applications and Examples
Presentation Overview
 The information presented here is largely derived from work
carried out on the LCVTP “CAE Case study and training based on
composite seat structure” project.
 This work was delivered in four stages:
1.
Creation of seat back FE model, definition of material orientation and
linear material model.
1.
Optimisation of thickness, orientation and cut-outs for linear loading.
2.
Development and correlation of non-linear material model.
3. Optimisation of design for crash (non-linear) loading.
 Each stage was disseminated to the LCVTP team via training and tutorials.
Composite Modelling Issues
• The complexity of different composite manufacturing techniques
and raw materials can make predictive modelling a challenge.
Ply stack
Ply orientation
Draping around
curvature
Resin system
Fibre supplier +
cross-section
Sizing
compounds
Mould release
agent and
compatibility
Processing
parameters
Composite Modelling Issues
• Generally two types of failure occur, depending on the loading rate
Low rate with fibre tensile
failure, pullout from resin matrix
and matrix failure. Easy to
distinguish mechanisms.
High rate loading, uses all
mechanisms plus heating and
pulverises the material.
Modelling Methods
 Modelling methods differ depending on which loading rate is of interest.
 The choice is guided by what future use the material may be put to.
 Available analysis codes offer different specialisms in loading rate and
varying support for composite materials:
> Nastran linear static
–
> Genesis linear static with composite optimisation support
–
> Abaqus non-linear static
–
> LS-Dyna non-linear dynamic and optimisation support
–
Modelling Methods
• Linear/non-linear static
Can either be modelled using a
simple model of the laminate with
a strain criteria for failure
interpretation by analyst (linear
model), or explicit failure
parameters for a non-linear model.
Modelling Methods
• Non-linear dynamic
Requires detailed material
parameters to create a nonlinear material model with
post-failure behaviour
captured.
Modelling Methods
• Given the correct choice of analysis code, there are options on
how to treat the composite ply stack to correctly account for the
laminate behaviour.
Short Beam Shear Simulation
Modelling Methods
• The option chosen for the LCVTP project work was to define a nonlinear static material model in LS-Dyna, giving access to the full
range of simulation types that might be required. This involved indepth understanding of the material’s mechanical properties and
calibration of the models to guarantee predictive capabilities.
Data Capture and Calibration
• The process for the predictive modelling of composite materials
in LS-Dyna was defined by GRM and WMG as:
Define material
data requirements
Carry out
coupon testing
Compare FE
result to test
Extract figures
from tests
Understand
differences
Check performance of material
model in component test.
Build FE
models of tests
Update input
figures where
necessary.
Data Capture and Calibration
Data required from coupon testing
Data specified by pre-processor for
orthotropic orientations
Data required from component
testing
Data Capture and Calibration
 Four tests designed to capture the key parameters
 The tests were recreated in LS-Dyna and correlated against the test
results
Values
Test and Fibre Orientation
EA, e11t, xt
UD Tension 0o
EB (=Ec), vBA, e22t, yt UD Tension 90o
GAB, τ1, γτ1, γ1, gms, sc Shear (in lamina plane) +/-45o
GCA, (=GBC)
Shear (interlaminar) 0o
Example Simulation
Data Capture and Calibration
• High quality correlation achieved to tension coupon tests
0 degree Tension
90 degree Tension
Data Capture and Calibration
• High quality correlation achieved to shear tests
Lamina Shear
Interlaminar Shear
Data Capture and Calibration
• Calibrated material model used to compare performance against
3-point bend test of closed top-hat section.
Data Capture and Calibration
• Some final calibration required.
XT found to increase
failure load.
Compression values
control gradient after
4mm
Changes to compression
values are realistic as these
were not correlated in our
coupon models.
Changes to tension values
are unrealistic as these
were correlated in our
coupon models.
Data Capture and Calibration
 Based on the 3-point bend test calibration results and use of
different modelling techniques for composites, the mis-match in
results was seen to be related to the shortcomings of shell
elements.
 It was noted that for most applications of the example structure
(automotive seat back) there would be little need to extend
beyond the shell element method.
 The cpu time required to model a Tshell element was significantly
greater than for a Shell element (3 times as long).
Design Optimisation Techniques
 With a calibrated FE models, there exists the opportunity to use
design optimisation techniques to define improve the structural
performance and/or reduce mass
 As part of the LCVTP activities the following optimisation studies
were considered:
> Linear seat frame optimisation using Genesis
> LS-OPT non-linear seat frame optimisation
> Preliminary non-linear optimisation studies using Genesis/LSDYNA interface
GENESIS Optimisation Methods
• The following methods are available within Genesis and other integrated
structural optimisation codes
LS-OPT Design Optimisation Techniques
 Successive Response Surface Method (SRSM) most
commonly used method, and has been used on seat
 Initial DoE matrix of LS-DYNA simulations run and RSM
generated
 Design variable ranges reduced about predicted
optimum and new DoE matrix of simulation run
Composite Seat Back Optimisation
 Target/allowable
changes – Thickness
– Ply orientation
– Hole placement
 Due to the limitation in the number of variables that can be used with LS-OPT hole
placement optimisation cannot practically be considered
 Allowing the introduction of weight reducing holes is possible using the GENESIS
coupling method
Targets
Achievable in:
LS-OPT
Constant thickness
only
GENESIS
Cupli
YES
Optimisation 1
Constant thickness and holes
(max. thickness 5mm)
Optimisation 2
Constant thickness and holes
(max. thickness 10mm)
-
YES
Optimisation 3
Variable thickness and holes
(max. thickness 10mm
-
YES
)
Composite Seat Back Optimisation
Torsional
Loading
Load applied rearward to one
corner of seat back, with
maximum deflection limits:
‒625N < 20mm
‒1112N < 50mm
Seat System
Moment
Moment applied rearward to
seat back. Requirements are:
‒No failure up to 723Nm
Composite Seat Back Optimisation
GENESIS
 Optimisation studies performed considering ply orientations and potential hole
positions

Process driven by strength based limits of seat back moment
Composite Seat Back Optimisation
GENESIS


Optimised design provided:
 Failure Index (composite stress) reduction from 1.2 to 1.05
 Mass reduction from 3.2kg to 2.8kg
Optimisation maintained uniform layup construction
Composite Seat Back Optimisation
LS-OPT
 Using LS-OPT, a Design of Experiments study was performed to determine the
optimum seat back thickness and ply orientation

Results
– The main objective of this optimisation was to determine the optimal number and
orientation of the plies in the seat back
– The following results show the optimised seat back design using LS-OPT –
A total of 24 plies giving a total thickness of 5.28mm (0.22mm per ply)
Optimisation
Thickness
Results (mm)
0° Ply
Interpreted
Number of
Individual Plies
1.90
9
45/-45° Plies 1.06
6
90° Ply
9
1.93
Composite Seat Back Optimisation
LS-OPT
–
–
–
Response surface plots can provide the user with an understanding to how
different variables can effect different objectives/constraints
The example plots below show the dominating effect of increasing the 0° ply
thickness over the 45° thickness
The red points are the infeasible models
LS-DYNA/GEN ESIS
Optimisation Interface
 LS-DYNA interface developed at GRM to couple to Genesis
 Using Equivalent Static Load Method (ESL) interface allows all methods of
optimisation to be undertaken linking to LS-DYNA implicit and explicit models
 As part of the LCVTP studies the method was extended to support composite
materials
Baseline Non-Linear
Model(s)
GENESIS
Interpretation of
Non-Linear Model
GENESIS
Optimisation
Updated Non-Linear
Solution
Benefits of Coupling
300

Internal sensitivity calculations break the link between the number of
design variables
2 0 0 and analysis calls
 Allows multiple non-linear and linear load cases to be considered in one
optimisation

Provides a quick route to the 'optimum' answer, without fully interrogating
1
design space
LS-OPT
Genesis Coupled Optimisation
Number of Variables
Composite Test Case
 Topometry (ply placement) optimisation performed on simple box
beam model, considering 2 LS-DYNA analysis models
> Torsion Loading
> Cantilever Bending
 Optimisation successfully developed minimum mass layup to meet LS-DYNA
requirements
GENESIS / DYNA Topometry Example
 Pole Impact Load Path Optimisation
 Utilising GENESIS' Topometry
Module, pseudo- topology results can
be obtained for impact based
solutions.
 Pole impact considered to
determine optimal load path creation
to minimise pole intrusion
 Direct load member developed,
effectively eliminating significant
deformations of the floor structure
Applications and Examples
LCVTP Correlation and
Development Studies
80
60
40
20
0
-20
0
Formula 1 Structural
Design
0.05
0.1
Summary
 Through existing experience and development work on the LCVTP
programme GRM have worked together with WMG to develop robust
analytical models of the glass/PA hot formed material
 The validated materials and methods have been applied to analysis of the
seat back design to allow predictive performance assessments
 With the validated models design optimisation studies have been
performed to reduce the mass of the seat back design
 The work with WMG has allowed GRM to further develop the methods in
the field of composite optimisation