Coupled problems require the simultaneous solution of more than one physics module of LS-DYNA to

obtain an accurate result.

Accurate CFD analysis will require structural coupling

The mechanical behavior of structural parts subject to

heat and fluid pressure loads have been neglected.

Underbody
Heat Fluid Pressure
panels

Until the road test

 Fluid Structure Interaction
Roof vibration analysis

• CFD analysis of full

vehicle.
• Couple parts of the
structure to analysis
the response in a
realistic environment.
 Fluid Structure Interaction

Three different options to solve the same problem:

• Solve Full Navier –Stokes with FSI non linear coupling (using weak
or strong coupling ).
• Solve Potential flow with a non-linear step at the end.
• Solve the structural analysis alone using the output from Navier-
Stokes (transient or steady state solver) and the *LOAD_SEGMENT
automatically generated input deck. Use
*ICFD_DATABASE_DRAG to write the files.
Fluid Structure Interaction. Results.

Navier Potential
Stokes Flow

Velocity

Pressure
 Fluid Structure Interaction. Results.

Navier Potential
Stokes Flow

24 hours 20 minutes
46 times faster

2 seconds Using *LOAD_SEGMENT

from Navier-Stockes solution
 Thermal Coupling

Monolithic coupling: big jumps of

physical properties.

Weak coupling: small jumps or very

fine mesh at the interface.
 Thermal coupling

Options to solve the same problem:

• Solve Full Navier –Stokes with thermal non linear coupling (using
monolithic or weak coupling). Shut off N.S after a certain steady
state has been reached and continue with a pure thermal coupling
analysis.
• Solve Navier Stokes using the steady state or potential flow solver
and continue with conjugate heat transfer analysis once steady
state has been reached.
• Solve the thermal analysis alone using the output from Navier-
Stokes (transient or steady state solver) and the
*BOUNDARY_CONVECTION_SET automatically generated input
deck. Use *ICFD_DATABASE_HTC to write the files.
 Conjugate Heat: Radiation

• Run steady state Navier-Stokes or Potential flow. Use

*ICFD_CONTROL_GENERAL to set it up.
• Once steady state reached or Potential flow finishes the
conjugate heat solver will use the steady velocity for the thermal
analysis.
Conjugate Heat: Radiation. Results.

Navier Stokes Potential Flow

Velocity Velocity
Conjugate Heat: Radiation. Results.

Navier Stokes Potential Flow

Temperature Temperature

Time: 1 hour 52 min

Time: 48 hours
26 times faster
 DEM Coupling

• Two-way coupling
• Particles affect fluid volume

Water management: Rain Simulation

 Recent developments introduced in
R10 :
• Added Steady state solver. See ICFD_CONTROL_GENERAL and
ICFD_CONTROL_STEADY.
• Added wave damping capabilities. See ICFD_DEFINE_WAVE_DAMPING
• Added Windkessel boundary conditions for blood flow. See
ICFD_BOUNDARY_WINDKESSEL
• Option to output loads coming from the fluid and applied on the structure
by using ICFD_DATABASE_DRAG keyword option. Similar feature for
thermal and HTC and using ICFD_DATABASE_HTC
• Two way coupling with DEM particles
• Option to shut off Navier Stokes solve after a certain time for conjugate
heat transfer analysis. See ICFD_CONTROL_TIME.
 Currently working on :

• Periodic boundary conditions

• Sliding mesh capabilities
• Immersed FSI capabilities
• Monolithic FSI
• 1D parachute model.
• Boundary layer mesh improvements in complex geometry cases.
 Current main usage:

• The EM solver solves Eddy currents

Using a coupled FEM-BEM method
• This implies that no air mesh
is needed which allows complex shapes
And strong deformations to occur
• The EM solver is therefore the perfect
candidate to solve coupled mechanical
thermal problems where strong
deformations occur such as in
Electromagnetic forming bending welding
and so forth
 Current main usage:

• More features have recently been introduced for such applications :

– Axisymmetric solver (R10)

– Conductivity function of material properties defined by the user with a
DEFINE_FUNCTION
– Option to define a circuit using a circuit equation and a DEFINE_FUNCTION to allow
more complex types of circuits.

• Investigation is under way to add magnetic material capabilities

through the introduction of an alternative monolithic solver.
 Three new applications :

• Resistive Spot Welding (RSW) capabilities

– Extension of the resistive heating solver.
– Introduction of EM_ISOPOTENTIAL to define a potential difference between
electrodes and EM_CONTACT_RESISTANCE to define a contact resistance
– Current capabilities are 3D, currently working on 2D solver.
• Battery short cut modelling
– Extension of the resistive heating solver.
– Introduction of circuit models to model ion transfers in batteries (See EM_RANDLES)
– Extension of EM capabilities to Thick shells
• Cardiac solver for heart modelling
– Extension of the resistive heating solver
– Ten Tusscher & Panfilov cell models
 Introduction - RSW

Electrodes on each sides of 2 sheets to be welded :

• Pressure
• Current flow => Joule heating => formation of a molten weld nugget

Coupled mechanical/EM/thermal simulations 20

 RSW and contact resistance

In RSW, contact resistance plays a very important role in the heating of the nugget

New model in LS-DYNA for local contact resistance (in 3D) depending
on local parameters, using *DEFINE_FUNCTION, e.g. Jonny-Kaars
model :

21
EM model for contact resistance (1)

Local contact resistance =

FUNCTION( Current flow
pressure, depends on
temperature, Bulk and
Electrical conductivity,
contact resistance Joule heating
Contact distance, ….
Using *DEFINE_FUNCTION Is locally added
To thermal solver

22
 EM model for contact resistance (2)
FEM solve:
Contact resistance added in (S0 + D) * ϕ = 0
stiffness matrix Where
• S0 is the Laplacian operator (nodes x nodes)
• D has
• 1/rs at (N1,N1) and (N2,N2)
• -1/rs at (N1,N2) and (N2,N1)
• 0 elsewhere
Row N1 gives:
N1
N2 (S0 * ϕ ) N1 + (D * ϕ ) N1 = 0
i1 + 1/rs (ϕ1-ϕ2) = 0
(ϕ2-ϕ1) = rs i1

And similar at row N 2

On rows not connected to contact
S0 * ϕ = 0 ensures the free divergence of
rs the current in the plates (no charge
accumulation) 23
Contact resistance depends on local
parameters

Local contact resistance =

FUNCTION(
ak, Total contact area,
pressure,
temperature,
Electrical conductivity,
Contact distance, ….
Using *DEFINE_FUNCTION
24
 EM cards to setup contact resistance
*EM_CONTROL_CONTACT
EMCT CCONLY COTYPE EPS1 EPS2 EPS3 D0
1 1

*EM_CONTACT
CONTID COTYPE PSIDM PSIDS EPS1 EPS2 EPS3 D0
18 1 1 2 0.3 0.3 0.3

*EM_CONTACT_RESISTANCE
CRID CONTID CTYPE CIRCID JHRTYPE D0
12 18 1 1

LCID D0
Per contact
14

*DEFINE_FUNCTION
FID
14

time Simulation time

area_glo Global contact area A (see equation (8))
area_loc Local contact area A (see equation (8))
dist_ct Distance between the 2 faces in contact
ctpress_mst Contact pressure on master side
ctpress_slv Contact pressure on slave side
temp_mst Contact temperature on master side
temp_slv Contact temperature on slave side
cond_mst Electrical conductivity on master side
cond_slv Electrical conductivity on slave side
vmstress_mst Von mises stress on master side
25
vmstress_slv Von mises stress on slave side
Typical RSW simulation

Current density Temperature

26
 Application
• The new LS-DYNA EM-Contact enables many
approaches to cover the contact resistance for RSW
• The Jonny-Kaars-Model is an approach based on a
resistance function of temperature and pressure
where its parameter are fitted according experiments.

pressure temperature
27
Battery Abuse Simulations
in LS-DYNA

Pierre L’Eplattenier, Sarah Bateau-Meyer, Iñaki Çaldichoury,

European LS-DYNA Conference, May 2017

Battery - Introduction
Battery – Distributed Randles circuit
model
 Coupling between the solvers
Electrochemical
• Ordinary differential equations
(Randles circuit model)
• Finite element analysis

Thermal Structural
Finite element analysis; Finite element analysis;
3-D Heat diffusion with source Nonlinear continuum mechanics
terms
 3.8

Cell Voltag
3.6

Getting the Randles circuit parameters

3.4

Type B 3.2
4.2 0 20 40 60 80 100
Discharge
Time (s)
4
Charge
Average
100
HPPC tests
3.8 50

Current (A)
Voltage (V)

0 4.1
3.6
-50
4
3.4 -100

-150 3.9
3.2 0 20 40 60 80 100

Voltage (V)
Time (s)
3.8
3
0 5 10 15 20

C/10 Capacity (Ah)

3.7

3.6
capacity tests 3.5

3.4
1.638 1.64 1.642 1.644 1.646 1.648
Time (s) 4
x 10
 External short (1)
External short on a cell module

In collaboration with J. Marcicki et al

Ford Research and Innovation Center,
Dearborn, MI, USA
 external short (1):
Exp. vs Num. temperature elevation at different locations

In collaboration with
J. Marcicki et al
Ford Research and
Innovation Center,
Dearborn, MI, USA
 External short (2)
Conducting cylinder falling on the tabs of a cell creates an external short

Current
density

Temperature

SOC vs time
 Randles circuits using Composite
Tshells

User mesh Layered mesh Layered mesh Randles circuits

Conductors only
a b c d
 Composite Tshells: Definition of the
- Tab
node sets
+ tab
(solid) (solid)

Node set 1
Node set 2

Node set 3 Node set 4

Cell (2 Tshells)

Electrical connections using *EM_ISOPOTENTIAL and *EM_ISOPOTENTIAL_CONNECT

 Composite Tshells: Keyword setup
*PART_COMPOSITE_TSHELL
$# title Layer1:positive current collector
Layered_Solid
$# pid elform shrf unused unused hgid unused tshear
3 5 0.833
$# mid1 thick1 b1 ithid1 mid2 thick2 b2 ithid2 *EM_MAT
21 0.225 0.000 0 12 0.325 0.000 0 $--------1---------2---------3---------4---------5---------6---------7---------8
9 0.100 0.000 0 16 0.325 0.000 0 $ em_mid mtype sigma eosId randletype
5 0.250 0.000 0 16 0.325 0.000 0
21 2 5.e6 1
9 0.100 0.000 0 12 0.325 0.000 0
21 0.225 0.000 0 0 0.000 0.000 0 12 1 2
*PART_COMPOSITE_TSHELL 9 1 3
$# title 16 1 4
Layered_Solid
$# pid elform shrf unused unused hgid unused tshear
5 2 6.e6 5
5 5 0.833 2 2 1.e6
$# mid1 thick1 b1 ithid1 mid2 thick2 b2 ithid2 4 2 2.e6
12 0.325 0.000 0 9 0.100 0.000 0 Layer2:positive electrode
16 0.325 0.000 0 5 0.250 0.000 0
16 0.325 0.000 0 9 0.100 0.000 0
12 0.325 0.000 0 21 0.225 0.000 0
0 0.000 0.000 0 0 0.000 0.000 0
Positive tab
*PART
$# pid secid mid eosid hgid grav adpopt tmid
6 1 2
*PART
$# pid secid mid eosid hgid grav adpopt tmid
7 1 4
 Composite Tshells: Internal short (1)
Module of 10 adjacent cells crushed by a sphere
• Each cell is composed of
• 228 *ELEMENT_TSHELL
• 22 unit cells (89 layers)
• 252 Randles circuit in each unit cell
• 55,440 Randles circuit total
Composite Tshells: Internal short (1)

Potential Current density

State Of Charge vs time

 Composite Tshells: Internal short (2)
Same 10 cells module crushed
by a cylinder

Potential Current density

State Of Charge vs time

 LS-PREPOST Battery Packaging
Application
• Easy design of the layers of a single cell
• Addition of connecting tabs
• Multiplication of cells to create modules
• Electrical connections

a b

c d
 Battery – Plans for the future
• Collaborations with Ford Research and Innovation Center and
Oak Ridge National Labs to improve:
• Mechanical simulations of layered cells
• Criteria for onset of internal short circuits
• Setting of internal short resistance

• Development of more macroscopic models for modules and packs

• Addition of new features in LS-PREPOST battery packaging

application
Electrophysiology modeling

44
 Motivation
• Experimental studies involving the in-vivo human heart are possible
and often available, but they are expensive and very limited.
• Well defined numerical modeling is emerging as a powerful tool
that can help to interpret experimental data.
• Cardiac modeling is a complex problem. The maturity of the models
of electrical propagation in the heart is still not comparable with
the one achieved in other engineering fields mainly due to :

 Non linear anisotropic inhomogenous material properties

 Direct observation of electromechanical potential distribution is not
trivial. Validation experimental results are difficult to obtain.
 The problem not only involves multiphysics but is extremely multi-
disciplinary.
 Electrophysiological models
The bidomain model : well-established description of the electrical
activity of the myocardium on a macroscopic scale, taking into account
the ionic current, the membrane potential and the extracellular
potential.

The monodomain model : The monodomain model is a simplification

of the bidomain equations. It assumes that conductivities are
proportional in the intracellular and extracellular spaces
 I. Verification of cardiac tissue
electrophysiology simulator using LS-DYNA

• GOAL
test the ability of LS-DYNA for cardiac tissue simulations

Benchmark:
Verification of cardiac tissue electrophysiology simulator using a N-version
benchmark, Philosophical Transactions of the Royal Society A: Mathematical,
Physical and Engineering Sciences, vol 369, issue 1954, pp 4331-4351,
November 2011

47
I. Verification of cardiac tissue electrophysiology simulator using
LS-DYNA
• BENCHMARK GOAL
Cuboid heart sample with stimulus on one corner. We observe the
propagation of the potential inside the cell by determining the nodes’
activation time.

Activation time
instant where the potential
becomes positive

48
I. Verification of cardiac tissue electrophysiology simulator using
LS-DYNA
• MODEL DEFINITION
Variable Description
equations monodomain
material transversely isotropic
PDE solver explicit
cell model Ten Tusscher & Panfilov
variant epicardium cell model
numerical integration scheme Qu-Garfindel Operator Split
mesh type hexahedral
solution method finite element
basis function linear Nedelec elements (FEMSTER)
pre-conditioners none
matrix solver hybrid-parallel, multifrontal, sparse direct
solver (MF2)
system architecture Serial or MPP
49
I. Verification of cardiac tissue electrophysiology simulator using
LS-DYNA
• MODEL DEFINITION
𝜕𝑉
β𝐶𝑚 + β𝐼𝑖𝑜𝑛 𝑢, 𝑉, 𝑡 − 𝛻. σ𝛻𝑉 = 𝐼𝑠𝑡𝑖𝑚 𝑥 , 𝑡 monodomain equation
𝜕𝑡
1 𝜕𝑢
= 𝑓(𝑢, 𝑉) cell model : ten Tusscher & Panfilov ionic equations
𝜕𝑡

𝑉 : membrane potential 2
𝑡 : time
σ : conductivity tensor Projection onto the FEM basis functions
𝐶𝑚: membrane capacitance 𝑑𝑉
β : surface area to volume ration β𝐶𝑚 𝑀. + β𝐼𝑖𝑜𝑛 − 𝑆. 𝑉 = 𝐼𝑠𝑡𝑖𝑚
𝐼𝑠𝑡𝑖𝑚 : stimulus current, applied at the position 𝑥 𝑑𝑡
𝐼𝑖𝑜𝑛 : single cell ionic current 𝑉 , 𝐼𝑠𝑡𝑖𝑚 , 𝐼𝑖𝑜𝑛 ⟶ nodal vectors
𝑢 : set of cell-level variables ⟶ 19 for ten Tusscher model
𝑀 : mass matrix 𝑀(𝑖, 𝑗) = Ω
Φ𝑖 Φ𝑗 𝑑Ω
𝑆 : stiffness matrix S(𝑖, 𝑗) = Ω
σ𝛻Φ𝑖 . 𝛻Φ𝑗 𝑑Ω
3 explicit Qu-Garfindel Operator Split
𝑑𝑡
𝑉 = 𝑉𝑡 𝑀. 𝑉𝑡+1/2 = 𝑀. 𝑉𝑡 − 𝑆 . 𝑉𝑡 Integrate diffusion operator for half timestep
2β𝐶𝑚

𝐶𝑚 𝑉 = 𝐼 𝑢, 𝑉
∗
𝑉 = 𝑉𝑡+1/2 𝑉 𝑑𝑢 Integrate ionic operator for full timestep
𝑡+1/2 = 𝑓(𝑢, 𝑉)
𝑑𝑡
𝑑𝑡
𝑉 = 𝑉 ∗ 𝑡+1/2 𝑀. 𝑉𝑡+1 = 𝑀. 𝑉 ∗ 𝑡+1/2 − 𝑆. 𝑉 ∗ 𝑡+1/2 Integrate diffusion PDE for half timestep
2β𝐶𝑚 50
I. Verification of cardiac tissue electrophysiology simulator using
LS-DYNA
• 9 SIMULATIONS
• RESULTS
dx (mm) number of elements dt (ms) number of time steps - 8 successful simulations
0.5 3,360 0.05 1,600 - 1 failed simulation :
0.2 52,500 dx = 0.1 mm with dt = 0.05 ms
0.01 8,000
β𝐶𝑚 𝑑𝑥2
CFL condition 𝑑𝑡 ≤ = 0.046 ms
0.1 420,000 0.005 16,000 2σ𝑙 σ𝑡

The results are very similar to the benchmark paper ones

Activation time along P1-P8 for dt = 0.05 ms
and dx = 0.5 mm, dx = 0.2 mm and dx = 0.1mm
Activation time at P8

51
I. Verification of cardiac tissue electrophysiology simulator using
LS-DYNA
• ELECTRICAL POTENTIAL PROPAGATION

52
 II. Developments

• Introduction of different solvers for the monodomain equations

• Introduction of bidomain model
• Presentation of the cards in LS-DYNA

53
 II. Developments
• TO FACE THE CFL CONDITION
PDE solver implicit

implicit Qu-Garfindel Operator Split This term now in the lhs

𝑑𝑡
𝑉 = 𝑉𝑡 (𝑀 + 𝑆). 𝑉𝑡+1/2 = 𝑀. 𝑉𝑡 Integrate diffusion operator for half timestep
2β𝐶𝑚
𝐶𝑚 𝑉 = 𝐼 𝑢, 𝑉
∗
𝑉 = 𝑉𝑡+1/2 𝑉 𝑑𝑢 Integrate ionic operator for full timestep
𝑡+1/2 = 𝑓(𝑢, 𝑉)
𝑑𝑡
𝑑𝑡
𝑉 = 𝑉∗ 1 (𝑀 + 𝑆). 𝑉𝑡+1 = 𝑀. 𝑉 ∗ 𝑡+1/2 Integrate diffusion PDE for half timestep
𝑡+
2 2β𝐶𝑚

Activation time at P8
 II. Developments
• TO GAIN TIME
numerical integration scheme Dave’s Operator Split

explicit Dave’s Operator Split

At even time step At odd time step

Integrate diffusion operator for one timestep Integrate ionic operator for full timestep
𝑑𝑡
𝑉 = 𝑉𝑡 𝑀. 𝑉𝑡+1 = 𝑀. 𝑉𝑡 − S . 𝑉𝑡 𝐶𝑚 𝑉 = 𝐼 𝑢, 𝑉
2β𝐶𝑚
𝑉 = 𝑉𝑡+1 𝑉∗ 𝑡+1 𝑑𝑢
Integrate ionic operator for full timestep = 𝑓(𝑢, 𝑉)
𝑑𝑡
𝐶𝑚 𝑉 = 𝐼 𝑢, 𝑉 Set 𝑉𝑡+1 = 𝑉 ∗ 𝑡+1
𝑉 = 𝑉𝑡+1 𝑉 ∗ 𝑡+1 𝑑𝑢 Integrate diffusion operator for one timestep
= 𝑓(𝑢, 𝑉)
𝑑𝑡 𝑑𝑡
Set 𝑉𝑡+1 = 𝑉 ∗ 𝑡+1
𝑉 = 𝑉𝑡 𝑀. 𝑉𝑡+1 = 𝑀. 𝑉𝑡 − S . 𝑉𝑡
2β𝐶𝑚

Machine time – simulation time = 80 ms (all the runs were done in serial)
Numerical integration scheme dt1=0.05ms - dx1=0.5mm dt1=0.05ms - dx3=0.1mm dt3=0.005ms - dx3=0.1mm
explicit Qu-Garfindel Operator Split 1min14s X 33h15min30s
explicit Dave’s Operator Split 59s X 24h49min3s
implicit Qu-Garfindel Operator Split 1min12s 3h23min58s 34h40min16s
implicit Dave’s Operator Split 58s 2h32min30s 24h53min12s
 II. Developments
• TO INCREASE THE ACCURACY
equations bidomain
PDE solver implicit
numerical integration scheme Spiteri-Ziaratgahi Operator Split

𝜕𝑉
β𝐶𝑚 + β𝐼𝑖𝑜𝑛 𝑢, 𝑉, 𝑡 − 𝛻. σ𝑖 𝛻𝑉 − 𝛻. σ𝑖 𝛻𝑢𝑒 = 𝐼𝑠𝑡𝑖𝑚 𝑥 , 𝑡
𝜕𝑡 bidomain equations
1 𝛻. σ𝑖 𝛻𝑉 + 𝛻. σ𝑖 + σ𝑒 𝛻𝑢𝑒 = 0
𝑢 𝑒 : extracellular potential
𝜕𝑢 σ𝑖 : intracellular conductivity tensor
= 𝑓(𝑢, 𝑉)
𝜕𝑡 σ𝑖 : extracellular conductivity tensor

Projection onto the FEM basis functions

2 𝑑𝑉 𝑉 , 𝑈𝑒 , 𝐼𝑠𝑡𝑖𝑚 , 𝐼𝑖𝑜𝑛 ⟶ nodal vectors
β𝐶𝑚 𝑀. + β𝐼𝑖𝑜𝑛 − 𝑆𝑖 . 𝑉 − 𝑆𝑖 . 𝑈𝑒 = 𝐼𝑠𝑡𝑖𝑚 2 stiffness matrices 𝑆𝑖 𝑖, 𝑗 = σ𝑖 𝛻Φ𝑖 . 𝛻Φ𝑗 𝑑Ω
𝑑𝑡 Ω
𝑆𝑖 . 𝑉 + 𝑆𝑖𝑒 . 𝑈𝑒 = 0 𝑆𝑖𝑒 (𝑖, 𝑗) = Ω
(σ𝑖 + σ𝑖𝑒 )𝛻Φ𝑖 . 𝛻Φ𝑗 𝑑Ω

3 implicit Spiteri-Ziaratgahi Operator Split

𝑢𝑡+1 = 𝑢𝑡 + 𝑑𝑡 𝑓(𝑢𝑡 , 𝑉𝑡 , 𝑡) solved using a PCG method, where the
β𝐶𝑚 β𝐶𝑚 preconditioner is the diagonal line of
𝑀 + 𝑆𝑖 𝑆𝑖 . 𝑉𝑡+1 = 𝑀. 𝑉𝑡 − β𝑀. 𝐼𝑖𝑜𝑛 (𝑢𝑡+1, 𝑉𝑡 , 𝑡) the matrix, or with the hybrid-parallel,
𝑑𝑡 𝑈𝑒 𝑡+1 𝑑𝑡
𝑆𝑖 𝑆𝑖𝑒 0 multifrontal, sparse direct solver, MF2

cell model Purkinje

III. Creation of a model

57
 III. Creation of a model
• LS-PREPOST
1 Build a mesh 2 Select the nodes where the stimulus is applied

Mode ⟶ CreEntity ⟶ Set Data

⟶ Set Nodes ⟶ Create

Sel. Nodes window : Pick ByNode

Apply

To select nodes inside the solid

- Blank the unwished elements with EleTools ⟶ Blank ⟶ Element
Sel. Elem. window : Pick or Area ByElem
Apply
- Create the nodes set using Sel. Nodes window : Area ByNode 58
 III. Creation of a model
• INPUT DECK – MECHANIC (for now, i.e. for pure EP model without mechanical coupling)

59
 III. Creation of a model
• INPUT DECK – ELECTROMAGNETISM for MONODOMAIN

emsol 11 ⟶ monodomain equations

numerical integration scheme:
numls 1 ⟶ explicit Qu-Garfindel
2 ⟶ implicit Qu-Garfindel
3 ⟶ explicit Dave
4 ⟶ implicit Dave

0.017606 0 0
emsol = 11 ⟶ *EM_MAT_003 Definition of 1 conductivity tensor e.g. σ = 0 0.133418 0 in 𝑆. 𝑚 −1
0 0 0.017606

The conductivity is more important along the direction Y, which represents the fiber length.

60
 III. Creation of a model
• INPUT DECK – ELECTROMAGNETISM for BIDOMAIN

emsol 12 ⟶ bidomain equations

numerical integration scheme:

numls 1 ⟶ implicit Spiteri-

Ziaratgahi
where the preconditioner is the
diagonal line of the matrix

numls 2 ⟶ implicit Spiteri-

Ziaratgahi
with the hybrid-parallel, multifrontal,
sparse direct solver, MF2 (faster)

emsol = 12 ⟶ *EM_MAT_005 Definition of 2 conductivity tensors σ𝑖 and σ𝑒

61
 III. Creation of a model
• INPUT DECK –
CELL MODEL VARIABLES

*EM_EP_TENTUSSCHER

Variables linked to a material id

62
 III. Creation of a model
• INPUT DECK - STIMULUS
*EM_EP_TENTUSSCHER_STIMULUS

Definition of :
- the starting time
- the period
- the duration
- the amplitude

*EM_EP_TENTUSSCHER_STIMULUS2

Stimulus loaded by a load curve representing

Node set id where stimulus is applied stimulus amplitude vs time

63
 III. Creation of a model
• ELECTRICAL POTENTIAL PROPAGATION - 2 STIMULUS

64
 CONCLUSION
• Different EP models in LS-DYNA, for both monodomain and
bidomain equations
• The ten-Tusscher cell model has been introduced
• They give good results on the first benchmark tests
• These models are available to the users through new cards
• More cell models will be added in the future

• What should be the priorities on pure EP?

– Other cell models (Purkinje, …) ?
– Introduce fractal Purkinje network ?
– Try runs with many elements ?
– Try runs with models closer to full heart with different cell models ?
• We are interested in the APD restitution results and whether more
developments are needed to simulate tachycardia and fribrillation
65