Modeling batteries for mobile applications

Modeling batteries for mobile applications
ECV Final Seminar
Ari Hentunen
Aalto University, Espoo, Finland
11.5.2016
Battery Modeling
User need
Objectives
Battery characterization
SOC range of interest: 10–90 %
Dynamic simulations of xEVs
Voltage error less than 2 %
Battery performance assessment
at end-of-life
Temperature error less than 2 ◦ C
Battery emulation at powertrain
testbed
Users
Drive train developers
Computationally lightweight
Automated model extraction
Experiments with a cell,
module, or pack
Offline time-series I–U–T data
Evaluation of capacity and
impedance at end-of-life
Battery system developers
Vehicle software developers
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
2/17
11.5.2016
ECV Final Seminar
Model-Based Performance Assessment
Electrical and thermal behavior can be simulated with different duty
cycles and ambient temperatures
Typical duty cycles, harsh conditions, worst case scenarios, etc.
Performance at the end-of-life conditions can be evaluated
Different battery sizes and configurations can be evaluated rapidly to
find the best solution
Time, cost, and effort for experimental testing is minimized
Battery size and performance can be optimized
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
3/17
11.5.2016
ECV Final Seminar
Approach for Multi-Cell Characterization
Battery module is selected as the basic unit
Includes the effects of packaging and cooling on the thermal performance
Test equipment cost is moderate
One module is characterized experimentally by using intermittent
charging and discharging at several rates and temperatures
Current and voltage data are obtained from a battery cycler
Temperature data are obtained from a battery management system (BMS)
Use of electrical and thermal networks to represent electrical and
thermal behavior
Module-model characterizes the average behavior of a module
Easily scalable for any pack-configuration
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
4/17
11.5.2016
ECV Final Seminar
Battery
Specification: Kokam SLPB 40 Ah (NMC), 26 V module
Table: Specification of the battery.
Property
Unit
Cell∗
Cell
Module
Nominal capacity
Nominal voltage
Max voltage
Cut-off voltage
Charge current
Discharge current
Energy
Nominal temperature
Max temp. (charge)
Max temp. (discharge)
Cycle life @ 80 % DOD
Ah
V
V
V
A
A
kWh
◦C
◦C
◦C
40
3.70
4.20
2.70
120
320
0.15
25
45
60
1 200
40
3.70
4.15
3.00
120
320
0.15
25
45
55
1 200
40
25.90
29.05
21.00
120
320
1.04
25
45
55
1 200
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
5/17
11.5.2016
ECV Final Seminar
Voltage Characteristics
ib Current
ub = uoc (sQ ) − Zi (t, sQ , ib , T ) ib
ub Voltage
uoc OCV
sQ State of charge
Zi Internal impedance
T Temperature
A lumped-parameter electrical equivalent circuit with dynamic
parameters is used to characterize voltage behavior
A systematic and flexible method for the extraction of impedance
parameters was developed1
1 Ari
Hentunen, Teemu Lehmuspelto, and Jussi Suomela. “Time-Domain Parameter
Extraction Method for Thévenin Equivalent Circuit Battery Models”. In: IEEE Transactions on Energy Conversion 29.3 (2014), pp. 558–566.
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
6/17
11.5.2016
ECV Final Seminar
Battery Dynamics
Impedance, electrochemical impedance spectroscopy measurement
Mass transport
Im(Z) [mΩ]
−1
−0.5
Charge transfer and
electric double layer
Ohmic
Re(Z) [mΩ]
0
0.5
f = 1 kHz
1
Inductive
1.5
f = 1 Hz
0.5
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
7/17
11.5.2016
ECV Final Seminar
Electrical Model
Equivalent circuit
Inputs
+ u0 −
R0
Current and temperature
uoc
External
Prediction of usable capacity, SOC, OCV,
self-discharge, and generated heat
+ u1 −
R1
+ un −
Rn
C1
Cn
+
−
ib
+
ub
−
ub Battery voltage
Equivalent circuit
ib Battery current
Prediction of terminal voltage
uoc Open-circuit voltage
Remark
R0 Ohmic resistance
Resistances and capacitances are
functions of the SOC, temperature, rate,
current direction, and SOH
Rn Dynamic resistances
Cn Dynamic capacitances
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
8/17
11.5.2016
ECV Final Seminar
Heat Generation Characteristics
ib Current
¡
¢
Pgh = uoc − ub ib + ib T
|
{z
}
polarization heat
∆S
| {z F }
entropic heat
ub Voltage
uoc OCV
T Temperature
∆S Entropy change
F Faraday constant
A novel method for ∆S characterization was developed, in which
entropy-change is characterized from electrical characterization test
Traditionally, ∆S characteristics are obtained from an entropy-change
characterization test2 or from literature
2 K.
Jalkanen, T. Aho, and K. Vuorilehto. “Entropy change effects on the thermal behavior of a LiFePO4/graphite lithium-ion cell at different states of charge”. In: Journal of
Power Sources 243 (2013), pp. 354–360.
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
9/17
11.5.2016
ECV Final Seminar
Thermal Dynamics
T Temperature
¡
¢
dT
mcp
= Pgh − hA T − Ta
dt
|
{z
}
transferred heat
Ta Ambient temperature
m Mass of the battery
cp Specific heat capacity
Pgh Generated heat
h Heat transfer coefficient
A Surface area of the battery
Mass, specific heat capacity, heat transfer coefficient, and surface area
are unknown
A lumped-parameter thermal equivalent circuit is used to characterize
thermal behavior
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
10/17
11.5.2016
ECV Final Seminar
Thermal Model
Equivalent circuit
Pgh
+ θ1 −
Rth1
+
+ θn −
Rthn
Pd
+
T1
Cth1
Tn
Cthn
Ta
−
Inputs
Generated heat and ambient temperature
−
+
−
Pgh Generated heat
Ta Ambient temperature
Equivalent circuit
T Battery surface temperature
Prediction of surface temperature and
dissipated heat
Pdh Dissipated heat
θn Temperature rises
Rn Thermal resistances
Cn Thermal capacitances
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
11/17
11.5.2016
ECV Final Seminar
Thermal Characterization Test
Identification of Heat Capacity and Thermal Resistivity
37
Est polarization heat
Exp temperature #1
Exp temperature #2 35
Exp temperature #3
Exp temperature #4
Generated heat [W]
15
Current [A]
20
0
−20
14
33
13
31
12
29
11
27
Temperature [◦ C]
16
Current
40
−40
0
2
4
6
8
10
12
Time [min]
14
16
18
20
10
0
2
4
6
8
10
12
Time [h]
14
16
18
25
20
Generated heat can be predicted from the voltage response
Every temperature-sensor is modeled separately
Weighted average value is calculated
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
12/17
11.5.2016
ECV Final Seminar
Electrical Characterization Test
PD & PC with 2 % pulse width, 10 min rest, and 2.5 ◦ C ∆T threshold
Characteristics
20
30
0
20
−20
10
Impedance
−40
0
Entropy change
6
12
Time [h]
18
24
29
28
Voltage [V]
Voltage
Temperature
Ambient
27
27
25
26
23
25
21
0
24
6
12
Time [h]
18
24
Temperature [◦ C]
0
DOD [Ah]
Current 40
DOD
Current [A]
40
Capacity
Open-circuit voltage
Model Extraction
50
50
Resistance [mΩ]
40
Capacity [Ah]
10 °C
15 °C
40
30
20
20 °C
25 °C
30 °C
35 °C
40 °C
30
20
10
10
0
0
0
10 °C
15 °C
20 °C
25 °C
30 °C
35 °C
10
30
40
50
60
SOC [%]
70
80
90
100
(b) Ohmic resistance (R0 ).
(a) Full-charge capacity.
30
400
200
of
of
of
of
of
of
of
10 ◦ C / (C/3)
10 ◦ C / 1C
25 ◦ C / (C/3)
25 ◦ C / 1C
40 ◦ C / (C/3)
40 ◦ C / 1C
all exps
28
Voltage [V]
Avg
Avg
Avg
Avg
Avg
Avg
Avg
300
∆S [J/(mol K)]
20
40 °C
100
26
24
0
22
−100
−200
0
10
20
30
40
50
SOC [%]
60
70
80
(c) Entropy change.
90
100
20
0
RMS 10 ◦ C
RMS 25 ◦◦ C
RMS 40 C
10
20
30
40
50
SOC [%]
60
70
80
(d) Open-circuit voltage.
90
100
Model Validation
Underground mining load-haul-dump (LHD) loader cycle @ 25 ◦ C
45
Exp voltage
Sim voltage
Exp temperature
Sim temperature
Voltage [V]
28
40
26
35
24
30
22
0
Temperature [◦ C]
30
25
1
2
3
4
5
Time [h]
2
Voltage
1
Temperature
1
0
0
−1
−1
−2
0
Error [◦ C]
Error [%]
2
−2
1
2
3
4
5
Time [h]
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
15/17
11.5.2016
ECV Final Seminar
Model Validation
90
60
60
0
30
0
30
0
Sim generated heat
Current
−120
159
160
161
162
163
Time [min]
Sim dissipated heat
164
−120
243
35.25
26
35
34.75
25.75
34.5
25.5
34.25
25.25
244
245
246
247
Time [min]
Sim dissipated heat
248
249
26
−30
250
38.5
Exp voltage
Sim voltage
Exp temperature
Sim temperature
25.5
Voltage [V]
Exp voltage
Sim voltage
Exp temperature
Sim temperature
26.25
0
Sim generated heat
Current
−30
166
165
26.5
Voltage [V]
−60
38
25
37.5
24.5
37
Temperature [◦ C]
−60
Heat [W]
120
60
Current [A]
90
60
Heat [W]
120
Temperature [◦ C]
Current [A]
LHD cycle at around 46 % SOC (left) and 22 % SOC (right) @ 25 ◦ C
34
25
159
160
161
162
163
Time [min]
164
165
33.75
166
24
243
244
245
246
247
Time [min]
248
249
36.5
250
Differences in the heat generation are caused by the entropy change
Model is capable to capture the electrical and thermal behavior very well
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
16/17
11.5.2016
ECV Final Seminar
Summary
Electrical and thermal battery model was presented and parameterized
for a Li-ion battery module with NMC chemistry
Automated parameter extraction from experimental I–U–T time-series data
Cell-, module-, or pack-level experiments
Entropy-change can be characterized from electrical characterization tests
Electrical and thermal performance can be assessed at any load profile
Heat generation and dissipation as well as required cooling power can be
predicted
Performance degradation due to aging can be evaluated
Entropic heat generation is significant for charge depleting cycles
Model can also be used in conjunction with a battery cycler to emulate
a battery in full-scale hardware-in-the-loop testing to accelerate
powertrain development
Modeling batteries for mobile applications
A. Hentunen
Aalto University, Espoo, Finland
17/17
11.5.2016
ECV Final Seminar