Presentation Ale UP Crashmat Final Long

See discussions, stats, and author profiles for this publication at: https://www.researchgate.
net/publication/273118545
An overview on airbag simulation with LS-DYNA: Uniform pressure

and ALE-modeling
Conference Paper · April 2004
CITATIONS READS
0 1,051
2 authors, including:
André Haufe
DYNAmore GmbH Stuttgart
258 PUBLICATIONS 848 CITATIONS
SEE PROFILE
All content following this page was uploaded by André Haufe on 07 June 2019.
The user has requested enhancement of the downloaded file.

An overview on airbag simulation with LS-DYNA:
Uniform pressure and ALE-modeling
Überblick über Airbagsimulationen mit LS-DYNA:

Uniform Pressure und ALE-Modellierung
André Haufe & Thomas Münz
DYNAmore GmbH
Dynamore GmbH
Industriestraße 2
70565 Stuttgart
http://www.dynamore.de
CRASHmat / Freiburg / André Haufe 1
Overview
 Motivation of FSI and ALE technology in airbag simulation

 State of the art in Uniform-Pressure-modelling
 UP-airbag models in LS-DYNA
 Wang’s hybrid inflation model
 Constitutive models / Contact definitions / Reference geometry
 Motivation of Euler/ALE
 A quick step into continuum theory
 Balance Equation
 Advection
 Interface reconstruction
 Multi-Material Euler/ALE for airbag simulation
 Interaction with Lagrangian structures
 Coupling and leakage
 Porosity and blockage
 Examples
 Folded airbag with impactor
 OOP-setup of hybrid III (3 year old)
 Summary

Motivation – why FSI and ALE?
Classical airbag modeling

 Major assumption: Uniformly distributed pressure in
airbag during inflation process. Thus no discretization
of the fluid necessary.
 Calculation of internal pressure from scalar gas

equations (EOS) and controlled volume of bag.
 For standard applications this is justified due to the

fact, that the impact of the dummy is only after full
expansion of airbag.
Courtesy of SAAB
Advantages:
Robust, cheap, mostly exact enough,
well tested and documented.
Disadvantages:
No exact flow simulation (first milliseconds not
physically correct).
Validation of the complete airbag model
Courtesy of DC (bag + inflator) necessary.

Motivation – why FSI and ALE?
“Out of Position”-simulation  Due to the small distance between dummy

(passenger) and the inflator, severe injuries and
even fatalities may result from the airbag inflation
process.
 Thus, for OoP-simulations the interaction of fluid-
flow, airbag and dummy is of uttermost
importance (FSI)!
 The fluid has to be discretized in order to
determine the flow effects (CFD)!
 In order to move the fluid mesh arbitrarily with the
car, a suitable mathematical description has to be
used.
Advantages:
Exact simulation of flow, thus realistic behavior
also in the first milliseconds.
Exact pressure distribution inside bag.
Validation of inflator separated from actual bag.
Disadvantages:
Extreme increase in computing time.
Algorithms are still under development.

Introduction to UP-models
Available models in LS-DYNA
 Simple pressure models

 Simple airbag model limited to one
 Adiabatic gas model gas fraction
 Wang-Nefske model
 Hybrid model
 Gas parameters of up to 8 gas fractions
 Temperature vs. time data for the mixture is needed as input
(usually measured through tank test)
 Detailed input for gas outflow (vents, leakage, fabric porosity)
 Jetting parameters can be defined
 A detailed theoretical reflection follows next ….
individual gas
fractions

Wang’s hybrid model
Energy balance is established for the airbag control volume
d
 mu cv   mi hi  mo ho  Wcv  Qcv
Energy out by
heat transfer through
dt airbag surface
h= specific enthalpy
Rate of change
of internal energy Work done by
airbag expansion
Energy insertion into Energy discharge from

airbag by mass flow airbag by mass flow
(inflator) (vents, leakage)
 biTi 
2   biTcv2 
 m i hi   m i  aiTi  2   m o ho   m o  gases
 fi  aiTcv  2 
    

Wang’s hybrid model
p3,T3
Conservation of mass
inflator control
volume
mcv  mi  mo  m12  m23 p1,T1, m12
p2 ,T2 ,
m2 ,V2
mcv   m cv dt m '23
bag leakage
m23 vent
Pressure is obtained via ideal gas law
c p (i ) = constant pressure specific heat of fraction i

mcv  r  Tcv
pcv  cv (i ) = constant volume specific heat of fraction i
Vcv r = gas constant = 8.314J/(K mol)
Constant pressure/volume specific heats are obtained from
c p   f i c p (i ) cv   f i cv (i ) fi = fraction of gas i

Parameters for UP-models
Parameter identification and validation
Gaining parameters for the inflator is not that easy, since

during the short pyrotechnic event the state variables
(temperature, pressure and volume) can not be
measured correctly. Thus so called tank tests are used
to measure the pressure curve and to have a simple yet
effective validation setup.
p3,T3
inflator
p2 ,T2 ,
p1,T1, m12 control
m2 ,V2 volume
m '23
bag leakage
m23 vent

Constitutive model for airbag material
 1  21  31 
   0 0 0 
 Nonlinear loading/unloading behavior of fibers 
E1 E2 E3

 Arbitrary orthotropic definition of material axes  1  
  32 0 0 0 
E2 E3
 Elastic liner, compressive stress-handling, etc.  
 1 
 0 0 0 
E3
C fibres
 
mn
 1 
 0 0 
 G12 
hliner  1 
h  sym. 0 
 G13 
 1 
 Orthotropic 
 (9 unknowns) G23 
1   0 0 0 
 1  0 0 0 

E, nue E  1 0 0 0 
liner
Cmn   
(1   )(1  2 )  1
2
 0 0 
 sym. 1
 0 
 
2
 Isotropic (2 unknowns) 1
2
  
Ea, Eb, Ec, nueab, nuebc, nueca,
Gab, Gbc and Gca
 h  fibres h  liner
(9 parameters LRATIO 
hliner
Cmn  1  liner  Cmn   liner  Cmn
for orthotropic material) h
 h   h 

Contact, reference geometry and jetting
Contact algorithm
 Stable and fast airbag-contact available segment

 Segment orientation arbitrary extension
 Exact modelling important
 Contact surfaces may be discontinuous
Reference geometry
 Various options for deformed elements due to folding

(set stressless, shrink option etc.)
Airbag fabric
Gaussian profile
Origin of wedge
Jetting option
 The jetting option allows to define a direction in the

spatial domain that is subjected to a higher pressure.
Pressure applied

One step back: The ALE methodology
Limits in classical structural mechanics:

Mesh distortion of Taylor bar
Problem: Solution
Large deformations/distortions Adaptivity (re-meshing/re-zoning)
Performance of elements degrades ALE-mesh smoothing

Standard continuum potatoes
Material domain
(Langrangian description)
Spatial domain
 (Eulerian description)
x
x
e2
Displacement field d  xx
e1
e3
More about continuum potatoes
Material domain
(Langrangian description)
  Spacial domain

(Eulerian description)
x
x 

e2 G
xG
e3 e1 Arbitrary reference domain
(ALE description)

More about continuum potatoes
Material domain
(Langrangian description) B
1
  Spacial domain

A
x
1
x 
1

e2 G C
xG
e3 e1 Arbitrary reference domain
(ALE description)

The moving potato: Lagrangian
… introduced by Euler and named after Lagrange!

A

The moving potato: Eulerian
… introduced by Bernoulli/D’Alambert and named after Euler!

B

The moving potato: Arbitrary…
… Lagrangian-Eulerian
C

More than one material…
Material domain
Spatial domain
x
x
Eulerian/ALE domain:
e2 Can be void, gas, liquid or solid material !
e1
e3
A mind experiment:
Translation Deformation
Lagrange
Euler Euler-mesh fixed in space
ALE Moving ALE-mesh
= Advection
t=t1 t=t2

Balance equations
Momentum balance:
ε σ
 x   x( v  x)   b  div σ
advection nodal forces
additional eqn. due to ALE concept mesh smoothing

accelerations
Mass balance:
incremental
   ( v  x)   div v  0 displacements velocities
Energy equation: (explicit cycle)
u   u( v  x)  σ : D  r  q
material velocity mesh velocity

Advection
Material Smoothing
void Simple average method
Equipotential smoothing
Advection
“Lagrangian step” Donor cell scheme (1st order accurate)
Van Leer scheme (2nd order accurate)
Comment on time step and accuracy

The mass flow velocity influences the
Mesh smoothing time step!
Mapping of history A material particle is not allowed to
variables and velocities! move more than through half an
element during one time step:
 x e 2x e 
tcr  min  , e 
nel
 c v 

Interface (re-)construction
Keep material together

Move material fractions
according to velocities (also
for post-processing)
Linear interface
Composite stress tensor

The internal forces are based on the nmat nmat
composite stress tensor
A pressure iteration algorithm assumes
 
*
 
k 1
k k   k 1
k 1
the same hydrostatic pressure in all
materials

Short summary on ALE in LS-DYNA
Example by Lars Olovsson

 Any material can flow through a defined domain
 Currently the maximum number of materials is 8
 The domain may be fixed in space (Eulerian) or
may move arbitrarily (ALE)
 Interfaces between different materials will be traced
and reconstructed
 Stresses will be iterated on element level
Multi-Material Euler Multi-Material ALE

What is this good for?
 It is advantageous to model
- gases
- fluids
- massive/bulky solid materials
(w large deformations)
by Eulerian/ALE-methods
 Often these parts are contained in or are

constrained by other parts. In many cases it might
also be advantageous to model these structures
Lagrangian.
 Interaction between Eulerian/ALE-

and Lagrangian parts (FSI)

Some examples
- gases
- fluids

Lagrangian.
Bulk forming

Some examples
- gases
- fluids

Lagrangian.
- Tanks -

Some examples
- gases
- fluids

Lagrangian.
- Airbags -

Multi-Material in Euler/ALE airbag-simulation
Inflow of gas fraction (multi-material) through point sources
Euler/ALE definition:
Ambient gas (air): individual gas *INITIAL_GAS_MIXTURE
Temperature = const. fractions  Initial temperature of gas
Pressure = const. mixture
Density = const.  Initial density of each
individual gas component
*MAT_GAS_MIXTURE
Hybrid definition:  cp0 und cv0 for each gas
 Temperature = f(t) component and ambient air
 Mass flow = f(t) (currently const. in future
 Mass fractions for temperature dependent)
each gas component vents
 Molekular weight airbag *SECTION_POINT_SOURCE
 cp= f(T) and cv = f(T)  Inflow temperature = f(t) and
for each gas velocity of gas mixture v(t)
component  Inflow mass fraction for each
“straight forward” gas component
 Quick change from a UP hybrid inflator definition to a multi-material ALE definition.

 In LS-DYNA 971 even faster transition: Just add one keyword (Eulerian mesh is
generated automatically, so are constraints, domains and parts)!
Coupling of Lagrangian/ALE (FSI)
 Choice between constraint and penalty methods
Constraint based method: Penalty based method:

Preserves momentum but does not Conserves energy but may be unstable
conserves energy
v4
Lagrangian finite element vL v4 vL
v1 v1
Eulerian finite element
v2 v3 v2 v3
N v i i  vL
N v  v
i i L
I   mi vi  mL vL I  m v  m v
i i L L
I I
 Nodal velocities are forced to follow each other

Coupling of Lagrangian/ALE (FSI)
 Choice between constraint and penalty methods
Constraint based method: Penalty based method:

Preserves momentum but does not Conserves energy but may be unstable
conserves energy
tn t n1
Fcoup  k dcoup  nˆ  nˆ

*
x(XCn , t n1 )
m
k  xi
t 2 xi
n̂
 8  dˆ coup
m*  min mcoup ,  N E , j mE , j 
n 1
 j 1 
Eulerian finite element d=penetration vector
Lagrangian finite element
 Coupling force is proportional to d , pressure vs. penetration curves can be defined

Leakage, blockage and porosity
 Leakage
In order to prevent leakage
(unwanted/undefined loss of
material through Lagrange
boundaries) a number of
sophisticated algorithms are
available.
 Increase number of coupling points
if leakage is a problem
 Porosity
Porosity can be defined, based on a
relative pressure vs. relative porous
fluid velocity relation.  Defined flow (loss
of material)
through fabric
 Blockage
Blockage for airbags – as already
available for CV airbags – is now also  Defined blocking,
available for ALE-formulation thus no flow
through fabric

Example -1-
Folded airbag with rigid tube impactor
UP simulation
0 ms 5 ms 10 ms 15 ms 25 ms
ALE simulation
0 ms 5 ms 10 ms 15 ms 25 ms
ALE simulation – gas visualized
0 ms 5 ms 10 ms 15 ms 25 ms

Example -1-
Folded airbag with rigid tube impactor
Acceleration of tube:
1,4
Eulerian
1,2 UP w/o jetting

UP w jetting
1,0
acceleration [m/s2]
0,8
0,6
0,4
0,2
0,0
0 5 10 15 20 25 30 35
time [ms]

Example -2-
Out-of-Position-simulation (hybrid-III, 3 year old)
0 ms 5 ms 10 ms
15 ms 20 ms 25 ms
Example -2-
Out-of-Position-simulation (hybrid-III, 3 year old): Eulerian simulation

Example -2-
Out-of-Position-simulation (hybrid-III, 3 year old)
Acceleration of head: Acceleration of thorax:
5,0 10,0
Eulerian Eulerian
9,0
UP w/o jetting UP w/o jetting
4,0 UP w jetting 8,0 UP w jetting
7,0
acceleration [m/s2]
acceleration [m/s2]
3,0 6,0
5,0
2,0 4,0
3,0
1,0 2,0
1,0
0,0 0,0
0 5 10 15 20 25 0 5 10 15 20 25
time [ms] time [ms]

Summary
 Uniform pressure models:

 Stable, robust, versatile and widely used simulation technology
 Suitable for impact in standard loadcases
 ALE-technology
 Very attractive technology for the simulation of gases, liquids and large
deformations in solids
 Stable and reliable implementation
 In combination with Lagrangian coupling new fields of
application: Special bulk forming problems, tank sloshing, bird strike etc.
 Coupled fluid and structure simulation for airbags

 Physical correct simulation of early stages in simulation
 Ability to predict impact of flow on airbag & dummy
 Exact pressure distribution in airbag
 Validation of gas generator separated from airbag
 Method of choice in OOP-simulations
 Expensive in terms of simulation time (MPP versions available)
 Current development focuses on more features and user friendly input

Thank you for your attention!
CRASHmat
View publication stats
/ Freiburg / André Haufe 38

Presentation Ale UP Crashmat Final Long

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Presentation Ale UP Crashmat Final Long

Uploaded by

Copyright:

Available Formats

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

An overview on airbag simulation with LS-DYNA: Uniform pressure

Conference Paper · April 2004

The user has requested enhancement of the downloaded file.

Überblick über Airbagsimulationen mit LS-DYNA:

 Motivation of FSI and ALE technology in airbag simulation

CRASHmat / Freiburg / André Haufe 2

Classical airbag modeling

 Calculation of internal pressure from scalar gas

 For standard applications this is justified due to the

CRASHmat / Freiburg / André Haufe 3

“Out of Position”-simulation  Due to the small distance between dummy

CRASHmat / Freiburg / André Haufe 4

Available models in LS-DYNA

 Simple pressure models

CRASHmat / Freiburg / André Haufe 5

Energy balance is established for the airbag control volume

Energy insertion into Energy discharge from

CRASHmat / Freiburg / André Haufe 6

Pressure is obtained via ideal gas law

c p (i ) = constant pressure specific heat of fraction i

Constant pressure/volume specific heats are obtained from

CRASHmat / Freiburg / André Haufe 7

Parameter identification and validation

Gaining parameters for the inflator is not that easy, since

CRASHmat / Freiburg / André Haufe 8

CRASHmat / Freiburg / André Haufe 9

 Stable and fast airbag-contact available segment

 Various options for deformed elements due to folding

 The jetting option allows to define a direction in the

CRASHmat / Freiburg / André Haufe 10

Limits in classical structural mechanics:

CRASHmat / Freiburg / André Haufe 11

CRASHmat / Freiburg / André Haufe 13

CRASHmat / Freiburg / André Haufe 14

… introduced by Euler and named after Lagrange!

CRASHmat / Freiburg / André Haufe 15

… introduced by Bernoulli/D’Alambert and named after Euler!

CRASHmat / Freiburg / André Haufe 16

CRASHmat / Freiburg / André Haufe 17

Euler Euler-mesh fixed in space

ALE Moving ALE-mesh

CRASHmat / Freiburg / André Haufe 19

additional eqn. due to ALE concept mesh smoothing

Energy equation: (explicit cycle)

material velocity mesh velocity

CRASHmat / Freiburg / André Haufe 20

Comment on time step and accuracy

CRASHmat / Freiburg / André Haufe 21

Keep material together

Composite stress tensor

CRASHmat / Freiburg / André Haufe 22

Example by Lars Olovsson

Multi-Material Euler Multi-Material ALE

CRASHmat / Freiburg / André Haufe 23

 Often these parts are contained in or are

 Interaction between Eulerian/ALE-

CRASHmat / Freiburg / André Haufe 24

 Often these parts are contained in or are

CRASHmat / Freiburg / André Haufe 25

 Often these parts are contained in or are

CRASHmat / Freiburg / André Haufe 26

 Often these parts are contained in or are