Process Simulation

of Composites

Dr. Roland Hinterhölzl

Process Simulation – Why?
Aviation industry
approx. 50% of the structure are made out of
composite: fuselage, frames, stringers, skins,…
 High volume, bigger parts [Airbus]

 Need for process optimization, for automation

Automotive industry
Serial production with very high volume, short cycle
time, robust processes
[BMW]

From hand lay up to (partly) automated serial production

 Automation new developments and optimization
 Shorter cycle times of manufacturing processes
 Process reliability
 Short design and development time

15.5.2014 Process Simulation of Composites 2

Manufacturing in Aerospace and Automotive Industry

Aerospace Automotive

[FACC] [BMW]

Hours, Days Seconds, Minute(s)

15.5.2014 Process Simulation of Composites 3

Overview
• Motivation
• Introduction of LCC
• Forming Simulation
• Draping Simulation
• Automated Fiber Placement (AFP) Process Simulation
• Braiding Process Simulation

• Simulation of Liquid Composite Moulding (LCM) processes

• Determination of Process Induced Deformations
(PID) & Curing Simulation
• Simulation Platform
• Conclusion - Outlook

15.5.2014 Process Simulation of Composites 4

Institute for Carbon Composites

• Founded in Mai 2009 financially supported by

the SGL Carbon Group

• Professor Dr.-Ing. Klaus Drechsler

• ~60 Researchers

• 1 Team Assistant

• 5 Technical Employees

• 2 Office locations in Garching

• 3 lab locations in Garching and Ottobrunn

15.5.2014 Process Simulation of Composites 5

Research Groups
Process Technology Process Technology Simulation Material Behavior
for Fibers + Textiles for Matrix Materials and Testing

Preform + Flow Process

Automated Fiber Placement Hybrid Materials +Structures Simulation
Composite Testing Lab

Tooling Systems Compaction, Curing and

Braiding Deformation Test Method Development

Production Systems Material Modeling and Structural

Tailored Textiles Analysis High Strain Rate Behavior

15.5.2014 Process Simulation of Composites 6

Simulation of composites

Semi-finished part structure:

FE draping simulation
semi-finished part
micro-/mesomechanics
preforming (here: draping)

LCM simulation:
matrix-injection
flow front

[FACC AG]

Curing- and distortion

Structural component simulation
curing
simulation
component

15.5.2014 Process Simulation of Composites 7

Preform Process Simulation

15.5.2014 Process Simulation of Composites Kap.-8

Overview
• Motivation for Process Simulation:
• Analyze the influence of process parameters on produced preforms
 increase knowledge of the process

• Optimization of the process and of produced geometry

• Generate input for further analysis (LCM, Curing/PID, Material Modeling)

• Main research activities

• Draping Simulation

• Automated Fiber Placement

• Braiding Process Simulation

15.5.2014 Process Simulation of Composites 9

Draping Process Simulation

15.5.2014 Process Simulation of Composites Kap.-10

 lay-up of the preform

Goals of Draping Simulation

• Process development / optimization
tool preform
• Geometry and alignment of the flat preform
• Drapeability (feasibility study)
• Prediction of draping defects
• …

• Part design (design to process) Forming (draping)

• Prediction of fiber alignment in required shape

• Prediction of fiber volume fraction

• …

• Embedding in process simulation platform:

• LCM simulation
• Curing and PID simulation
• Structural simulation
BMW i3Airbus
– fullyA380
automated draping
– Pressure process [BMW]
Bulkhead

15.5.2014 Process Simulation of Composites 11

Requirements for Draping Simulation
• Textiles
• woven, Non-Crimp Fabrics (NCF), UD, etc.
• dry or pre-impregnated
plain weave twill weave NCF
• Deformation modes

Shear Bending Elongation Inter-ply Motion Intra-ply Motion

• Defects

Wrinkling / Folds Loops Gaps / Voids Fiber Pull-Out Damage of Stitching

15.5.2014 Process Simulation of Composites 12

 Numerical Approaches for Draping Simulation
P
Kinematic Approach
• pin-joint method [Mack 1950]
• purely geometrical approach
• computation of the fiber direction
• direct cut geometry generation

level of detail
L2 L1

Finite Element Approach

• modeling of the preform on ply-level with shell elements
• simulation of the draping process and the tools
• consideration of forces, friction, velocities, etc.
• prediction of deformation modes and defects in the preforms

shear
elongation
bending

15.5.2014 Process Simulation of Composites 13

Comparison of the Different Approaches

Kinematic Approach Finite Element Approach

+ fast & easy to use + accurate results even for complex shapes
+ many software available + accounts for material and process influence
+ good interface to structural analysis + prediction of draping defects possible
(wrinkles, gaps, etc.)
- no material behavior
- no process influence - higher computational effort
- poor results for complex shapes - material testing required

Catia CPD Experiment PAM-FORM

(Kinematic Simulation) (FE Simulation)

15.5.2014 Process Simulation of Composites 14

 Finite Element Approaches for Draping Simulation
Macro-FE Approach
• modeling of the preform on ply-level with shell elements
• simulation of the draping process and the tools
• consideration of forces, friction, velocities, etc.
• prediction of deformation modes and defects in the preforms
Meso-FE Approach
• modeling of the preform on yarn-level with 2D/3D elements
• simulation of Unit Cells deformations
• consideration of friction between yarns, weaving structure,…
• prediction of deformation modes and behavior of units cells
• Not yet suitable for forming simulation (computational cost)
on part scale

shear elongation
bending

[Voith]

15.5.2014 Process Simulation of Composites 15

 Material characterization (simulation of the material
behavior)
Simulation Procedure • Picture Frame Test
•…

Parameter Validation on test geometry

Determination
• Hemisphere
• Double Dome
•…

Process simulation (design and optimization)

• Diaphragm draping
• Drape forming
• Hand lay-up
•…

Validation

ECD
Process
Simulation
Saertex

15.5.2014 Process Simulation of Composites 16

Material Characterization Devices at LCC

Bias Extension Test

Picture Frame Test

Tensile Test

Friction Test
Cantilever Test

15.5.2014 Process Simulation of Composites 17

Validation
Material card has to be validated before applying the simulation on actual processes
1. Set up of validation process
2. Simulation of validation process
3. Measurement method to compare simulation and test results

Validation Process (INSA Lyon) Simulation Measurement

15.5.2014 Process Simulation of Composites 18

Example:
FE Simulation of Diaphragm Process

15.5.2014 Process Simulation of Composites Kap.-19

 Example – Simulation of Single Diaphragm Process

Principle:
Draping of a textile with a silicone membrane (diaphragm)
Membrane Textile Table

Tool

Placing the fabric over Lowering the edges of Applying a vacuum Thermal binder
the tool the membrane between tool and activation
membrane

Demoulding
[AHD]

15.5.2014 Process Simulation of Composites 20

 Simulation of the Diaphragm Process

Modeling
– Macro-scale simulation with PAM-Form (ESI Group)
– Material 140 – „biphase shell element material with thermo-
visco-elastic matrix and elastic fibers“

Setup
Components of MAT 140 (ESI)
– Draping of a biaxial NCF over a generic helicopter frame
– Modeling of membrane, fabric, tool and table
– Same sequence as in experiment

applying vacuum

initial model

lowering edges

15.5.2014 Process Simulation of Composites 21

Characteristics of the Diaphragm Process

• Relative movement between membrane and textile

--> Results in high friction forces

x-displacement y-displacement

high friction

• Anisotropic friction between membrane and textile

--> Orthotropic friction coefficient needs to be defined
Fiber
Fiber dirc1
dirc1

Fiber Fiber low friction

dirc2 dirc2

15.5.2014 Process Simulation of Composites 22

Simulation Results for Single Ply Draping

+/-45 NCF

Experiment Simulation

-/+45 NCF

(Eurocopter DE)

15.5.2014 Process Simulation of Composites 23

Example:
FE Simulation of locally stitched NCF Forming

15.5.2014 Process Simulation of Composites 24

Local Stitching
NCF preform stacks are locally stitched together for better handling

Stitching can influence the draping result

Influence of the local stitching needs to be investigated

Final Part: Pitch Horn

Sewing of Preform Stack

Draping of Stitched Preform Stack

Lay-up on Tool

15.5.2014 Process Simulation of Composites 25

Single-Lap Shear Test
Purpose of test
• Investigation of initial state:
pretensed stitch <> slack in stitch
• Influence of the stitching on the textile
• Stress-Strain behavior of the textile
in the stitching vicinity

Local stitching

Lap Shear Test Setup

Clamping areas

Tensile Force applied on Lap Shear Test

Test Setup of Lap Shear Test

15.5.2014 Process Simulation of Composites 26

Modeling the Stress-Strain behavior of the Stitching
• Only normal stress are applied on PLINKS
• Definition of a Load-Displacement curve for PLINKS elements

Phase I:
Fmax Slack phase -> Lap Shear Test

Phase II:
III Fibers deformation -> Lap Shear Test
I II
Phase III:
Stitching thread elongation -> Bias Extension Test

Fmax:
Maximum bearable force -> Lap Shear Test

Load-Displacement Curve for PLINKS elements

15.5.2014 Process Simulation of Composites 27

Simulation of double diaphragm forming process

Cut-out line
Stitching shape

Stitching and cut-out lines Overview of the simulation model

• Comparison on a real forming process : generic double curved helicopter frame

• FE meshing automatically created from CAD data

15.5.2014 Process Simulation of Composites 28

Comparison

Region a Region c
Region b

Average angle
Layup Region Standard deviation
difference

a 1,89° 1,07
[0/90 // -/+45] b 4,7° 5,28
c 13,15° 6,85

Margossian et al, Comp. Part A, 2014

15.5.2014 Process Simulation of Composites 29

Automated Fiber Placement (AFP) Process Simulation

15.5.2014 Process Simulation of Composites Kap.-30

Overview – Simulation of the AFP – Process

Placement – scenario

Therm. and mech. material

Validation Simulation Model models
[LCC]

15.5.2014 Process Simulation of Composites 31

Research Activities at the LCC
0.4

ε [Δl / l]
1D – Simulation: 0.35

0.3

• Calculation of the time-dependent 0.25

mech. compaction behaviour 0.2

• Simulation of the transient heat

0.15

0.1
transfer 0.05

0
Zeit [s]

2D – Simulation:
• Prediction of the mech. and therm. behaviour
of the process
• Simulation of bridging of the tape
• Modeling of tack

3D – Simulation:
• Mech. placement on complex geometry
possible
• Prediction of the complete thermal history
of the placement process

15.5.2014 Process Simulation of Composites 32

AFP – Process Simulation: Example
• Gap- / Overlap detection
• Compaction / Debulking
Fig. 2: Compaction behaviour of a Slit-Tape
• Bridging
• Temperature distribution

Fig. 3: Sim. Bridging

 Define reliable
process window

Fig. 1: Gap in Simulation

Fig. 4: Surface temperature distribution

15.5.2014 Process Simulation of Composites 33

Braiding Process Simulation

15.5.2014 Process Simulation of Composites Kap.-34

 Braiding Process Simulation
• Goals of braiding process simulation
• Process parameters have a strong influence on
braided structure
• Mandrel shape defines contour of structure
• Modeling the real manufacturing process using
an explicit FE-Solver (e.g. Abaqus/Explicit,
PAM-Crash) to analyze correlation between
process parameters and braided structure

• Rovings are represented by truss- or beam-

elements

• Consideration of friction between

• roving-roving
• rovings-guide ring
• rovings-mandrel
diamond/ half regular/ full triaxial

15.5.2014 Process Simulation of Composites 35

Simulation of Liquid Composite Moulding (LCM)
processes

15.5.2014 Process Simulation of Composites Kap.-36

Motivation
Goals

• Prediction of process parameters

• Tool pressure
• Flow front velocity
• Fill time
• Prediction of defects (dry spots, pores)
• Cure kinetics / Heat transport Fig.1: Sportscar Monocoque

• Optimization of injection schemes

• Tool design

• Design of robust processes

Challenges

• Material characterization
• Preform
• Resin

• Capturing influences of preceding

process steps (such as textile handling, preforming)
Fig.2: Filling simulation (flow front plot) of a Sportscar Monocoque

15.5.2014 Process Simulation of Composites 37

Physics of LCM Processes
• Flow processes in porous media
can be described by the Navier-Stokes equation
𝜕𝑣
𝜌 = −𝛻𝑃 + 𝛻 ∙ 𝑇 + 𝑓
𝜕𝑡

• By homogenization of the media (e.g. by volume averaging)

and introducing some assumptions
the flow can be described by Darcy‘s law
𝐾 𝐾
𝑣= 𝛻𝑃 − 𝜌𝑔 ≈ 𝛻𝑃 Fig.1: Illustration of flow fronts (Real vs. Darcy) within an textile preform
𝜇 𝜇

• Darcy‘s law relates the flow front velocity 𝑣

to the applied pressure gradient 𝛻𝑃
with the proportionality constants permeability 𝐾
and viscosity 𝜇.
• Newtonian Fluid
• Permeabiliy is a material parameter of the fiber preform • Constant Viscosity
• Homogeneous, Inelastic Porous Medium
• Material
• Neglecting of Gravity and Capillary Forces
• Layup • Low Flow Velocity (𝑅𝑒 < 1)
• Fiber orientation • Saturated Flow
• Compaction state • Neglecting Intra Yarn Flow
Tab.1: Summary of assumptions for validity of Darcy’s law

15.5.2014 Process Simulation of Composites 38

Permeability Testing for LCM Processes
(lat.: permeare = ‚to pass‘) ln-Plane components / Off-Plane component/
2D tensor through thickness perm.

Description 𝐾11 0 0
In the 3D case the permeability can be noted as a 2nd order tensor. 𝐾∗ = 0 𝐾22 0
0 0 𝐾33
The tensor can be diagonalized with a rotation
of the global coordinate system Fig.1: 3D permeability tensor
to a coordinate system spanned
by the major permeabilities (cf. Eq.1).

Determination

• Experimental
• 1D-flow method
• Radial-flow method (2D)
• 3D-flow method

• Numerical (Multi-scale approach)

• CFD-calculation on meso-level
based on representative volume elements (RVE)
• Abstraction of the results to the
macro level as permeability tensor
Fig.2: In-plane test setup Fig.3: Off-plane test setup

15.5.2014 Process Simulation of Composites 39

Numerical Method - Overview

Fabric Scan „Grey Box“ Material Card RTM Process Simulation

Software Tool (PAM-RTM)
Info (Matlab)
• Image Analysis
• Textile Modeling Draping Simulation
• Permeability Calculation
• Material Data
Process Parameter

CAD

Fig.1: Fabric Scan Fig.2: Image Analysis Fig.3: Textile Model [WiseTex, KU Leuven]

15.5.2014 Process Simulation of Composites 40

Determination of Process Induced Deformations
(PID) & Curing Simulation

15.5.2014 Process Simulation of Composites Kap.-41

Motivation: Dimensional Control on Stiff Structures
• Geometrical deviations on stiff structures result in
 Additional rework effort and / or high mounting forces required,
 Expensive iterative tooling adaption.
Deformed frame section
Magnification factor: 30

Picture: DLR

Deformed frame after curing

Magnification factor: 30

15.5.2014 Process Simulation of Composites 42

Curing simulation - goals
• Primary:
• Determination of process induced deformations (dimensional accuracy)
• Evaluation of the effects of residual stress (dimensioning)

• Secondary:
• process optimization (cycle time)
• process control (interaction between parallel simulation of physical curing process)

Spring-In / Spring -
Shrinkage Back

PID
Fernlund et al, 2003

Warpage

Svanberg, 2002 Twigg, 2001

15.5.2014 Process Simulation of Composites 43

Methodologies for determination of process induced
deformations (PID) Increasing level
of detail and
modeling effort

Curing simulation:
Simulation of all
relevant
Phenomenological:
mechanisms (e.g.
Summarization of
reaction kinetics,
different
modulus
mechanisms with
development,…)
Experimental: one actuating
Iterative variable (enhanced
optimization of final CTE)
part shape

Analytical:
Estimation of Evaluation of Residual
deformation for Stresses, Optimization of
simple
geometries Process Control

PID Quantification (Dimensional Control)

15.5.2014 Process Simulation of Composites 44

Simplified Approaches for PID Determination
Utilization of simplified approaches for strain anisotropy

Cure
m  m eq therm
 mchem
Fig. 1: Thermal and effective chemical shrinkage are
projected on the matrix coefficient of thermal expansion

 x  z
0

1 z →
m eq

"Reiner" Spring-In, MS1-17, MS1-18, MS1-19

1.350

1.300

1.250

1.200
Fig. 2: Experimental Spring-In

 [°deg]
1.150
data presents target variable
Fig. 3: Verification of αm,eq on linear-
1.100

for determination of αM,eq

Fig. 4: Application of αm,eq for particular material 1.050

and process on a frame structure elastic FEA of cool down step 1.000
Position1 Position2 Position3 Position4 Position5

15.5.2014 Process Simulation of Composites 45

Process simulation including curing: application flow on the
example COMPRO
Virtual autoclave Thermo-chemical module
• heat transfer autoclave gas → part-tooling-assembly • reaction kinetics
• feedback control autoclave • heat conduction

control routine – status

Figure: Experimental determination of effective HTC variables (,T) Figure: Time-Temperature-Transition-(TTT) Diagram

Flow Module
Stress module • resin flow due to laminate compaction (viscosity, permeability)
• modulus development related to → variations in wall thickness and fiber volume content
resin degree of cure, temperature
• resin cure shrinkage and thermal expansion coefficients

Johnston A., Hubert P.,

1996

COMPRO is a commercial composite

processing software: Convergent
Manufacturing Technologies, Canada
Figure: modulus development Figure: Force-Displacement-Diagram Fiber Bed Compaction

15.5.2014 Process Simulation of Composites 46

Characterization for Process Simulation
Viscosity
Resin Cure Development
Kinetics

Karkanas, Partridge: Cure Modeling and Monitoring of Epoxy/Amine Resin

Systems. II. Network Formation and Chemoviscosity Modeling

Cure
Shrinkage

8
-4
x 10 dalpha / dt of the model solution
Database
7

5
dalpha / dt [1/s]

3 alpha of the model solution

0.9

2
0.8

1 0.7

0 0.6
0 50 100 150 200 250
time [min]

Modulus
alpha [1]

0.5

0.4

0.3

0.2

0.1
3
-5
x 10 residual dalpha / dt of the model solution
Development
0
0 50 100 2 150 200 250
time [min]
1

0
Residual dalpha / dt [1/s]

-1

-2

-3

-4

-5

-6

-7
0 50 100 150 200 250
time [min]

15.5.2014 Process Simulation of Composites 47

Curing / PID simulation - results
Deformed frame section
• Quantities: Magnification factor: 30

• Deformations
• Residual stress
• Spatial distribution of fiber volume fraction
• Time history of temperature evolution on the part
• Final degree of cure distribution
• Glass transition temperatures

• Assessment a priori concerning: Fig. 1: Spring-In of a generic C-frame after

curing
• Mounting tolerances, dimensions of tooling
• Load bearing capacity
• Part’s quality (sufficient degree of cure,
glass transition temperatures,
hot spots (resin degradation))

• Optimization via variation of

• Geometry
• Material selection
• Lay-up
• Process cycle Fig. 2: Temperature Distribution on a Frame Section at the End of 2nd Ramp
[simulated utilizing Compro2D and Raven, Convergent Manufacturing
Technologies, Vancouver, Canada]

15.5.2014 Process Simulation of Composites 48

Simulation Platform

15.5.2014 Process Simulation of Composites Kap.-49

Simulation Platform

Forming Simulation

Material Modeling

Flow Process Simulation

[FACC AG]

Compaction,
Structural Analysis Curing and Consolidation

15.5.2014 Process Simulation of Composites 50

Process Simulation Platform - Example

Preforming
• Draping, Braiding, Filament winding, …
• Simulation of the preform process

Preform Characterization
• Experimental
• Numerical

LCM-Simulation
• Tool design, Process time prediction
(Flow front velocity, pressure distribution)
• Process optimization
• Curing simulation (for PID investigations)

15.5.2014 Process Simulation of Composites 51

Conclusion - Outlook

15.5.2014 Process Simulation of Composites Kap.-52

Geometry In-service,
Manufacturing Recycling
Geometry, Quality Allowable Damages
(Waviness, Porosity, etc.) Repair

Robust Design
via Simulation as Built
Effects

Product Development
Design, Process Simulation,
Structural Analysis, Testing, Thank you
Quality Assurance,…
Material for your attention!

15.5.2014 Process Simulation of Composites 53