Battery Simulation - CADFEM€¦ · Multi-Level Physics in Battery simulation ... channel can be modeled as 1D flow. ANSYS matrices ... ANSYS Workbench

Evgeny Rudnyi

Lucas Kostetzer

Sreekanth Nallabolu

Funded by

Battery Simulation CADFEM GmbH

Marktplatz 2

85567 Grafing

Germany

gefördert vom: funded by

BMBF Project

„Zentrales Innovationsprogramm

Mittelstand (ZIM)

A project for 2 years

CADFEM

ZSW

Zentrum für Sonnenenergie- und

Wasserstoff-Forschung (ZSW)

Baden-Württemberg

LionSmart

- 1-

http://www.zsw-bw.de/index.php?id=2


Multi-Level Physics in Battery simulation

1D System Level: Battery pack and power train Analyze individual systems and integrated performance

Tool used: ANSYS Simplorer & MOR

3D Component level: Battery pack Charge/discharge cycle design, life, safety, cost, thermal

management

Tool used: ANSYS Mechanical & CFD

1D Component level: Battery Cell I-V characteristics, Electro-chemistry analysis

Tool used: ANSYS Simplorer

Model

Extraction

Physic solvers

Me

ch

an

ica

l

Component level

System level

Coupled physics +

Control system

- 2-


Content

Vehicle Dynamic Simulation

Electrochemical Modeling of Li-

ion Battery

Thermal Model of Battery (1D

CFD)

Compact Thermal Models via

Model Order Reduction

Battery Pack Electrothermal

Simulation at System Level

- 3-


Complex interdependency of subsystems in HEV

Electric

Machine

(EM)

GearboxEngine

(ICE)

Control

Strategy

Maximum and mean

electric powerPower

Electronics

Peak and mean

power demand

over time

Downsizing due to EM

Power and Torque

Distribution of load to

EM and ICE

Torque capability

Battery

• Power

• Voltage

• Energy

- 4-


Drive cycle

Backward quasistatic simulation

0

10

20

30

40

50

60

70

80

90

0 200 400 600 800 1000 1200 1400 1600 1800 2000

FTP

Time (sec)

Sp

ee

d (

m/s

)

Speed request

- 5-


Hybrid Electric Vehicle Simulation @ CADFEM

Series HEV model

Model to check the integrity of our semi-physical battery model

Model useful for component sizing

Model useful to test energy management startegies

- 6-


Vehicle Dynamics and Wheel

mv·g

vFaero

Ftrac

Fgrad

Frolla

a

Linear acceleration Force

Gravitational Force

Rolling Resistance force

Aerodynamic drag force

v

Rwheel

vreq , Freq

ωreq , Treq

Treq

ωreq

Treq = FTR * Rwheel

ωreq = Vreq / Rwheel

Freq

vreq

- 7-


Software for System Level Simulation: Simplorer

8

+

-

B11A11 C11

A12 A2

B12 B2

C12 C2

ROT 2ROT 1

ASMS

3~M

J

STF

M(t)

GN

D

m

STF

F(t)

GN

D

Magnetics

JA

MMF

Mechanics

L

HQ

Simplorer Simulation Data Bus / Simulator Coupling Technology

Block DiagramsState-space

Models

Digital/

VHDL

JK-Flip flop with Active-low Preset and Clear

CLK

INV

CLK

CLK

J Q

QB

CLR

PST

Flip flop

K

CLK

CLK

INV

0 0 0 0 1 1 1 1 1 1X-Axis

Curve Data

ffjkcpal1.clk:TR

ffjkcpal1.j:TR

ffjkcpal1.k:TR

ffjkcpal1.clr:TR

ffjkcpal1.pst:TR

ffjkcpal1.q:TR

ffjkcpal1.qb:TR

MX1: 0.1000

PROCESS (CLK,PST,CLR)

BEGIN

IF (PST = '0') THEN

state <= '1';

ELSIF (CLR = '0') THEN

state <= '0';

ENDIF;

statetransition

AUS

SET: TSV1:=0SET: TSV2:=1SET: TSV3:=1SET: TSV4:=0

(R_LAST.I <= I_UGR)

(R_LAST.I >= I_OGR)

EIN

SET: TSV1:=1SET: TSV2:=0SET: TSV3:=0SET: TSV4:=1

State Graphs

Cxy

BuAxx

Electrical circuits Hydraulics, Thermal,

...


Battery Model

Three possibilities in Simplorer

Block Diagrams Equivalent circuit

Analog modeling (Physics based)

- 9-


Content



ion Battery


CFD)





- 10-


HEV / EV

Battery requirements

Safety

Performance

AvailabilityLife

Costs

Source: li-Tec

No Overheating

internal shortcut, overcharge

or abuse

Long life time, 10

years

High energy

density

Fast charging

Low self discharge

Functionality over

wide temperature

range

THERMAL

THERMAL

THERMAL

- 11-


Thermal model of a cell

Thermal simulation coupling:

Heat generation phenomena:

Reaction heat

Ohmic heat from contact resistance

Ohmic heat from solid phase

Ohmic heat from electrolyte phase

Reversible heat (usually ignored)

Cell model Battery pack

Heat

Temperature

elejsjcrTotal qqqqq ,,

K. Smith, C. Wang, J. Power Sources 160 (2006) 662 – 673

- 12-


Battery principle

(Rosario, 2007)

- 13-


Ideal vs. Real Battery

Voltage

SOC

Capacity

Current

Constant voltage during

discharge/charge

Constant capacity for all

discharge/charge

currents

Voltage

SOC

i2

i1

i1>i2Capacity

Current

Voltage drops during

discharge

The higher the

current lower the

Capacity decreases for

higher currents

- 14-

gefördert vom: funded bygefördert vom: funded by

Impedance model: This model comes from those

who measure the impedance

spectrum (Z=V/I) of a battery.

The model allows to describe

high frequency effects well

Reference: Jochen Bernhard Gerschler, Julia

Kowal, Magnus Sander, Dirk Uwe Sauer ,High-

spatial impedance-based modeling of electrical and

thermal behavior of lithium-ion batteries

In Simplorer, Resistor, capacitor

etc. can be combined to create

Impedance model

1-D semi-physical

model: The model is based on

electrochemistry equations.

– The Nernst equation,

– The Buttler-Volmer equation,

– The 1D diffusion,

– The 1D migration of ions.

– Some parameters are

necessary to be adjusted

through testing

CADFEM has implemented Semi-

physical model in Simplorer

1-D physical model: The model is based on

electrochemistry equations. The

best well known model is the

Newman model (Software Dualfoil)

ANSYS has implemented

Newman model in Simplorer

Newman Research Group

CADFEM uses Newman model as

reference model(dualfoil)

Different cell physics Simulation models

- 15-


Physical based models

DUAL FOIL, Fuller and Newman

(1993)

Open source Fortran code

Li-ion battery

Detailed model

6 non linear, coupled PDE’s

Over than 50 parameters

Very high accuracy

Model used to evaluate other battery

models

Implemented in Simplorer

K. Smith, C. Wang, J. Power Sources 160 (2006) 662 – 673

- 16-


Newman Model vs. Semi-Physical model

Newman model

Semi-physical model

Note:

The Semi-physical model uses an

equivalent method to treat the

porous media

- 17-


Material parameters by optimization with optiSLang

Optimization: 11 parameters to fit the SOC Curve

Reference signal

Signal of best design after

40 generations

- 18-


Content



ion Battery


CFD)





- 19-


Battery pack Thermal management

Liquid cooled:

more complex

silent

very efficient

isolation difficult

additional cooling system

necessary

Air cooled:

simple

noisy

less efficient

Source: Prof. Dr.-Ing. Andreas Jossen , TUM- 20-


3D thermal model - example

Pouch cells

Cooling by Air

Rectangular channels

Battery cell Fluid channel

Cutting detail

Simplified CFD domain

- 21-


ANSYS Thermal model

FEM mesh

Automatic, one

push button,

HEX mesh

CFD 1D CFD + Thermal simulation

FLUID116

Average film coefficient is calculated in the CFD problem

- 22-


ANSYS Thermal model - transient

Step response of 1W/cell

HOW to use this in the system level?

- 23-


Content



ion Battery


CFD)





ANSYS

matricesReduced

Model

MOR

- 24-


System Thermal Simulation in Simplorer

Current : Heat Flow

Voltage : Temperature

Resistor : Thermal

resistance

Capacitor : Thermal

capacitor

Voltage source :

Temperature source

Current source :

Heat Flow source

Tamb=f(time)

Cooling by natural convection

Tamb=t1

heat=f(time)

Cooling by natural convection

- 25-


Different 3D pack cooling Simulation models

3D Thermal model

•3D thermal model is very

efficient for flat cooling

channel, in which the cooling

channel can be modeled as

1D flow.

ANSYS

matrices

Complete FEM model

Reduced

Model

Reduced dimension

(State Space Model)

MOR

From 3D model to state space model

Note:For efficient system simulation, the model reduction is

necessary, the 3D thermal model can be easily reduced to a

compact model with MOR method

3D CFD model

•3D CFD model can be used as a

general purpose simulate method

for very complicated 3D cooling

channel.

From 3D model to Foster network • The response of the Foster network under step

input is a sum of several exponentially decaying terms.

Note:For efficient system simulation, the model reduction is

necessary, 3D CFD model can be reduced to a compact

model with LTI(Foster network) method

- 26-


MOR for ANSYS: http://ModelReduction.com

FULL files

Linear Dynamic

System, ODEsCxy

BuKxxExM

MOR Algorithm

Small dimensional

matrices

Current version 2.5

ANSYS Model

Simulink,

Simplorer, Spice,

…

- 27-

http://modelreduction.com/


ANSYS Workbench

Steady State model

Extract system

matrices

Checking

Transient model

Run transient case

Extract temperature results for

comparison with reduced model

(optional)

- 28-


Get System matrices

APDL script

Edx

dt Kx f

f

K

E

- 29-


Reducing the order of the model

mor_for_ansys Cell1.full -K fstatic.full -N 40 -l -C output.txt -m

Cell2.full Cell3.full Cell4.full > mor.out

r

T

rr

rrrr

xcy

ubxAx

State Space format

- 30-


Comparison

Accuracy 1% - 10 dofs

per input

Model reduction time

is comparable with

several static solves

0.00 1000.00 2000.00 3000.00 4000.00Time [s]

0.00

0.50

1.00

Tem

pera

ture

[cel]

Curve Info

TCell1.TTR

ansys_Cell1Imported

System level Device level (FEM)

Simulation time [s] 4000 4000 (100 timesteps)

Dimension 40 (10 x Input) 48500 elements

CPU time [s] <1 ~20 min

- 31-


Content



ion Battery


CFD)





- 32-


Multiphysics Coupling in Simplorer

Ohmic loss of a power transistor:

Resistance as a function of temperature

Ohmic loss as a function of current and

temperature

Thermal dissipation modeled as a

thermal reservoir (C) and a thermal

resistance (R) to the environment

oo TTkRR

oo TTkIRIRIP 222

Ohmic loss

of transistor Temperature

of transistor

Conservative nodes

Temperature

Heat flow

Across Current Heat Flow

Through Voltage Temperature

- 33-


ANSYS Simplorer

3D cooling model

3D Thermal model

3D CFD model

Electrochemistry cell model

Impedance model

1D Semi-Physical model

Heat

Temperature

MOR

LTI

Reduced cooling model

A B

C DState - Space

Coupling between cell models and 3D pack cooling model

- 34-


Simplorer integration

Import Matrices of State Space model

4 conservative

thermal pins

Temperature

reference pin

- 35-


Thermal System Level Simulation

0000

Cell1

Cell2

Cell3

Cell4 T_Ref

H

VALUE=0

H

VALUE=0

H

VALUE=0

Q

VALUE=20H

VALUE=10

0.00 1000.00 2000.00 3000.00 4000.00 5000.00Time [s]

293.00

294.25

295.50

296.75

297.50

Y1

[ke

l]

Curve Info

THM1.TTR

THM2.TTR

THM3.TTR

THM4.TTR

Thermometers

Heat flow sources

- 36-


Electro-Thermal System Level Simulation

SOC definition

0

I1

A

+

V

EQU

ICA:

Q

VALUE=20

Cell1

Cell2

Cell3

Cell4 T_Ref

H

tfm1

H

tfm2

H

tfm3

H

tfm4

H

tfm5

th

p m

th

p m

th

p m

th

p m

0.00 2.50 5.00 7.50 10.00 12.50 15.00 17.50 20.00Time [s]

-375.00

25.00

375.00

14.895

14.900

14.904

0.488

0.500

Curve Info

AM1.ITR

VM1.VTR

SOCoutTR

CONST

const1

- 37-


Electro-Thermal System Level Simulation

Current(prescribed)

Voltage

SOC

Cell

Temperature

- 38-


Less Pollution, A better way to live

1

2

3

Ele

ctr

oche

mis

try

Th

erm

al

Po

wer

Te

mp

era

ture

batt

ery

syste

m s

imu

lati

on

Inductive battery charging system

- 39-

http://upload.wikimedia.org/wikipedia/commons/1/1e/James_Clerk_Maxwell_big.jpg

http://blog-imgs-10.fc2.com/j/p/c/jpcarstoday/mi0220070508sb.jpg

Documents

Battery Simulation - CADFEM€¦ · Multi-Level Physics in Battery simulation ... channel can be modeled as 1D flow. ANSYS matrices ... ANSYS Workbench