My name is Pierre L'Eplattenier. I work at LST and

I will talk to you today about the path

towards

including batteries in electric or hybrid car

crash simulations with

LS-DYNA

and I'd like to thank my colleague

Inaki Caldichoury

who's working on this subject with me

and helping prepare this talk

So just a little bit of context about

automotive safety.

In 1966,

the National traffic and Motor Vehicle

Safety Act

was enacted in the United States,

giving new standards for motor vehicle and

road traffic safety.

And as you can see,

on the curve on the right.

Since then,

the number of death on the car crashes

gas decreased by quite a bit.

Uh, thanks largely to seatbelts,

airbags,

and general knowledge came from car crash

tests.

And use of simulation in the past couple of

decades

has been,

uh, has helped a lot

to getting a better comprehension of

the car

crash events and among the various commercial

offers available.

LS-DYNA

has been widely accepted by the

automotive

community.

Now recently a new type of locomotion has

reemerged because

actually it was one of the first at the

beginning

of the 20th century.

It is the electric car and

the range increase that we have seen in the

recent

years will make it more and more difficult

for car

users to protect their batteries in rigid

containers,

and so the simulation will probably help

quite a lot

in the battery safety.

Uh,

and so these batteries they bring new

challenges,

as you've seen on the news,

very often when battery and electric car

hybrid car gets

into a crash.

You can see pretty large fires getting out,

so how confident are we that battery will not

catch

into fire after a crash?

How safe are we as consumers ?

As engineers can we anticipate these problems?

Where is your degree of understanding?

How much should we protect the battery?

And that's where the simulation can help

quite a

bit.

Now simulating a battery crash is quite

challenging.

First of all,

the physics is very complex and it's

a multi physics which couples tightly

mechanical electromagnetics,

electrochemistry and thermal.

There's a wide range of length scale

with the battery being of the size around

1 meter

or so,

while the individual layers in a cell have

thickness around

tens of micrometers.

And this is the level at which the

internal

shorts happen and trigger the big thermal

runaway battery catching

into fire.

There's also a wide range of time scales,

from crash in millisecond that happen in the

millisecond to

thermal runaway fire and explosion that can

take minutes to

hours or even days.

So luckily, withLS-DYNA now we already had

tightly coupling between mechanics electrical

thermal and what we

had to

work was the electrochemistry on the bottom

right here.

So now we have a module that couples the

mechanical

where the displacement can be given to the

electrical part

and taken into account by the electrochemical

part and I

will come back to that.

We use equivalent circuit for that part and

so we

can have the mechanical deformation trigger

the

Internet shorts

The electrical and electrochemical are

coupled with the

thermal,

so we give the Joule heating to the thermal

and

we get back the temperature parameters.

And,

uh,

the thermal and mechanical also coupled with

the

mechanics, and more recently,

we added the ICFD module into that

whole chain where the thermal and the

ICFD

communicate by the conjugate heat transfer

solver and more recently,

the CFD and the electrical they can

communicate directly so

we can have a conducting fluids and we will

see

briefly that a bit later.

Now,

we have four models depending on the scale

or

the detail that the user wants to study.

So as I said, the internal short they happen at

the cell level and they are the root cause

for

internal runaway.

So it is important to understand the effect

of a

mechanical or thermal abuse on a single cell,

which will probably depend on the type of

cell.

So for that we have what we call the solid

element model where all of the layers are

made using

solid elements and the same as

is used for the mechanics,

the thermal and EM.

And we also have the composite Tshell model

where the

mechanical is modeled using composite

Tshells and the EM

and thermal still use an underlying solid

mesh that we

revealed behind the scenes and it allows to

do faster

run with less element and solid element model

.

Both of these models are for typical studies

at the

level of the cell of the module,

maybe not much larger models.

So modeling a full battery at the level of

each

cell layer would be very expensive

computationally.

So we created the macro model that we called

"batmac"

that allows using the information from a

single cell

study,

and in particular information of the

internal shots in

a full battery model in a car crash,

and so that's the macro model.

That's also can do internal and external

shorts,

but that's available now for pack or a battery.

So for this model we are one or a few

solid element through the thickness for the

mechanics EM and

the thermal and two fields each at

the positive

and negative current collectors.

And we also have the meshless

model where we don't care about the what's

happening internally

in the cell,

but we still have all of the equivalent

circuit and

the energy and so forth that can be

be output by the cell.

That's mostly for external shorts

All of these models are based on equivalent

circuit,

so we use distributed random model where

instead of simulating

the electrochemistry happening between the

anode and cathode and separators,

we use equivalent circuits that have been

used a lot

in the battery model.

But we use them in a distributed mode,

meaning that we have one equivalent circuit

for each node.

In the positive and negative current

collector,

so each node of the positive current

collector is connected

to the opposite node of the negative current

collector.

Such as a Randles circuit.

So the advantage of this distributing circuit

model is that

they are not too expensive computationally.

They can accurately reproduce the main

feature of a cell,

electrical and thermal.

They are easy to

get the cell parameters from the voltage and

current measurements.

It's simple to electrically connect them to

external circuit like

connector buses and to model

external shorts and

the main thing was the ability to model

internal

shorts.

Like on the bottom right where if you have

deformations of the current collectors you

can replace some of

the Randles

circuits by just internal short resistance and

you can

have the value of the short

resistant depending

on the local mechanical parameters or

deformations of the run.

To deal with the different timescales

we do,

usually the mechanics EM and the thermal with

the usual

crash time step during the milliseconds or so

crash time

we get the deformations of the short and then

we

continue the run by freezing the mechanics

and we do

only EMand thermal with now time steps much

larger

around the second or so to let the battery

discharge

through the shots and get the temperature

increase and let

it run for the physical minutes or hours

if you need to. For the simulation,

the EM and the thermal are not too sensitive

to

the time step apart from getting the correct

discretization of

battery discharge and temperature increase,

so it's OK to switch from the millisecond to

the

second time frame for this.

And now, I'll show you a few examples.

This is an external short example that was

performed by

Ford and where they connected

uh,

a little module of five cells together by

some

buses and they just came and connected the

plus and

minus of the bus and let the current flow in

that external short and they looked at the

temperature increase

at different location in the buses and in the

cells

themselves and compare that with the

experiment and you can

see under the right graph that matches pretty

good between

the two.

This is another example of an external shot

where

now it's a 20 cell module that were

connected

on some buses and a plate came and made all

of these buses crash together.

Deform and come in contact together.

And so the current now is doesn't flow

through the

cells anymore,

but can flow through the short circuit

between these different

buses there that come in contact and they

looked at

the elevation of the temperature of the

external shot.

This was done with the meshless model.

So here are some other examples.

On the left it's external short where the

cylinder tube

the conducting cylinder tube comes and makes

the two tabs

into contact and that makes some external

short and the temperature

elevation. In the in the middle,

that's sphere falling on

ten cells module and creating some

internal shorts and that's using the feature

model.

And on the right it's a internal short using

the

"batmac" model where we have a big pack of

50 cells that come and gets impacted by one

of

its corner and it creates some internal

shorts and temperature

elevation.

So these are examples of what can be done.

Now some more recent development that we did.

If the.

OK,

it's coming.

So it's the nail penetration so.

We got a.

penetration with simulation.

So this is example for a nail penetrating

30 cell back and creating holes.

Around that whole some shorts

are created and so we get these shorts and

temperature

elevation.

So we use the "batmac" model here for the

first 10 cells because that's about where the

nail stops

during its penetration and use some

the rest so that we still get all of the

energy,

the current and the voltage coming from all

of the

other cells.

But we didn't care too much about the

internal part

of the cell,

so that's available.

Uh,

also something new that we did is thermal

management

with coupling between the batmac model,

the thermal and the CFD model,

so that's an example of

a pack of 96 cylindrical cells that is placed

as

you can see on the top left picture and they

are different way of cooling that pack

because just in

normal use it heats up.

So there is either direct air flow so air

cooling

that flows directly through these cells

and then there is

indirect liquid cooling by the bottom plate.

So on that bottom template you can have a

little

tube where some water can flow through the

tube and

then the whole package cool down by the

bottom.

So we did some simulation with that and you

can

see for example here.

The difference between no cooling and liquid

cooling by the

bottom plate and the you can see the

repetition of

the temperature in the back and also

the

we gained quite a bit because with no cooling

we

module heated up to up to 43 degrees Celsius,

whereas with liquid cooling

heated up to 24 degrees.

Uh.

Now that's the same thing,

but with air cooling on the left and with air

plus liquid cooling on the right,

and you can see again and again that you have

by using the different kind of cooling system

.

So thermal management is is available also in

LS-DYNA

with this coupling

between EM thermal and ICFD.

Um,

And the nice thing also is that you can use

the same mesh here.

Of course you wouldn't do a battery cooling

simulation at

the same time as a crush experiment,

but you can use the same mesh you use for

thermal management later in the crash.

Uh.

Then another example that I would like to

show if

my OK.

So now we also have the ability to have the

coupling between ICFD,

and the EM and battery module,

where we can have conducting water

flowing over cell.

Here that's conducting where that is

flowing over the buses

connecting the cells and it creates some

short circuits.

Because now the current can flow from one tab

to

the next through the water which is conducting

,

and change completely the repetition of the

current and the

temperature in fact and you can see on the

right

that the state of charge diminishes quite

a lot when you

have these,

Uh,

when you have the water connecting the

different tabs so that's some other ideas of

simulation that are

available.

So just a summary of the different models

between the

Solid Composite Tshells Macro model and

Meshless.

We can have different cell geometrys,

some of them are used for the cell,

some of them are used more for the modules

and

some of them you will use for the pack.

All of them you can have connection to

external

circuit for internal shorts all of them can

do that

except for the meshless you can do external

short

for all of them and they are all coupled with

the thermal mechanics.

And I didn't add a line for the coupling with

ICFD,

which I should,

but they are pretty much just

for now

the Macro model, Composite Tshells

and Solid.

And of course,

the Meshless like

on the last slide

if you just have a flow over the conducting

surface.

So in conclusion,

LS-DYNA is a multiphysics code,

so a framework of battery safety modeling

that couple mechanical

thermal,

electrical and electrochemical responses

has been developed.

The development is done in collaboration

with

uh,

industrial partners,

that was then that suggested

all these development,

but it's available to all of the users.

To our knowledge no other code has this

multiphysics capability,

especially not the coupling between the

mechanics and the rest

for battery crash. Efforts are being

done in

order to couple the model with user defined

subroutines to

go beyond equivalent

circuits, So through user map like subroutines.

So it's it's already starting to be available

.

And the simulation solvers themselves rely on

experiment tests to

characterize their equations,

so there's a effort we believe that is needed

across

the industries and academia to bolster our

collective knowledge about

that.

It is a difficult problem.

And on that, thank you very much and have a

good day.

Bye bye.