You are on page 1of 9

Modeling and Simulation of Nickel-Cadmium Batteries

during Discharge
G IULIANO S ALOMÃO S PERANDIO , C AIRO L ÚCIO NASCIMENTO J ÚNIOR , G ERALDO J OSÉ A DABO
Instituto Tecnológico de Aeronáutica – ITA
Divisão de Engenharia Eletrônica
São José dos Campos-SP, Brazil – 12228-900
+55 12 39475878
giulianosperandio@gmail.com, cairo@ita.br, adabo@ita.br

Abstract—A battery is an essential component of any aircraft. during the first development phases (concept and design) in-
It has several functions such as to start the aircraft propulsion crease exponentially in the latter stages (production and op-
engines and to provide power during electrical emergencies. eration) [1], [2].
The aircraft designer must know the electrical behaviour of
the batteries that have been preselected for possible use in the Batteries are essential components of an aircraft. They are
aircraft at several operating conditions. Typically, the man- mainly used to start the aircraft propulsion engines and as
ufacturers of batteries for aircrafts provide proprietary soft- a backup for the main electrical system during emergencies.
ware that can be used to generate the battery voltage curves Therefore aircraft designers need to simulate the behaviour of
over time for constant and user-specified room temperature the selected batteries in these situations [3]. Typically, battery
and discharge current, assuming that the battery is initially manucfacturers do not provide simulation models. However,
fully charged. This paper shows how these battery voltage they provide the battery voltage curves over time obtained
curves can be used to generate and validate a battery dis- during their laboratory tests considering: a) constant room
charge model. By using such a model, the aircraft design- temperature, b) constant discharge current, and c) battery ini-
ers can simulate the battery behaviour at several operating tially fully charged. Nowadays these voltage curves can be
conditions such as a variable room temperature, a variable obtained by the battery user using the manufacturer’s propri-
discharge current and a partially charged battery. As a case etary software.
study, it is shown how to generate and validate the battery
discharge model for a specified nickel-cadmium aeronautical This article shows: a) how the standard battery voltage curves
battery. The model is then used to simulate the battery during (generated by the manufacturer’s proprietary software) can
an electrical emergency situation considering three types of be used to produce mathematical models for the battery dis-
loads: linear (constant resistance) load, constant power load charge phase, b) how these mathematical models are vali-
and constant current load. In all three cases, the simulation dated and used to simulate the battery discharge dynamics
results were as expected. in the practical situations of concern to the aircraft designers,
for instance, variable discharge current and battery initially
not fully charged. Data from a specific aeronautical nickel-
TABLE OF C ONTENTS cadmium battery is used to illustrate the proposed procedure.
1 I NTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
2 N ICKEL -C ADMIUM BATTERIES . . . . . . . . . . . . . . . . . . 1 This article is organized as follows: Section 2 shows how the
raw data provided by the manufacturer of a certain nickel-
3 I MPROVING THE PAATERO M ODEL U SING THE
cadmium battery can be used to generate and validate the
BATTERY C APACITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
so-called Paatero model. In Section 3 a modification of the
4 M ODEL S IMULATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
Paatero model is proposed in order to increase its accuracy.
5 C ONCLUSION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Section 4 shows how the modified Paatero model was im-
R EFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 plemented using Matlab & Simulink and used to simulate an
B IOGRAPHY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 electrical emergency situation with three types of loads: lin-
ear (constant resistance) load, constant power load and con-
1. I NTRODUCTION stant current load. Finally, conclusions are drawn in Section
5.
Computer simulation is used as much as possible during the
development phases of an aircraft to investigate the aircraft
behaviour in several operational conditions. One of the main 2. N ICKEL -C ADMIUM BATTERIES
reasons for doing so is that the cost to correct errors made Nickel-Cadmium Battery Model
The desired nickel-cadmium battery model should have as
1 978-1-4244-7351-9/11/$26.00 c 2011 IEEE.
input the discharge current, I(t), and the room temperature,
2 IEEEAC Paper #1823, Version 1.3, Updated 13/Jan/2011.
T (t). The battery voltage, U (t), is the model output. 26
25
Battery Data Extraction—The battery used in this paper can
24
have an aeronautical use. The battery data is shown in Table

Voltage [V]
1. The nickel-cadmium battery supplier makes available a 23
22 T = 40 ◦ C
Table 1. Battery data. T = 20 ◦ C
21 T = 0 ◦C
20 T = −20 ◦ C
Parameters Value
0 10 20 30 40 50 60
Capacity 50 Ah
Total cell number 20 Time [min]
Nominal cell voltage 1.2 V
Total voltage 24 V Figure 2. Battery voltage curve for I = 50 A – each mark
represents a value captured manually.
software that produces the discharge battery voltage curve for
a fixed discharge current and room temperature. This battery
voltage curve can be sampled using a cursor that shows the Figure 3 shows the curves used to fit the captured data for
voltage and time for each selected position. For each curve to temperature T = 20 ◦ C and discharge current I = 50 A, I =
be processed in the model, 50 points were manually captured 100 A and I = 200 A.
since the manufacturer’s proprietary software does not allow
the curve data points to be exported to a file. I = 50 A I = 100 A I = 200 A

Figure 1 shows the points captured for T = 20 ◦ C and con- 25


stant discharge current I = 50 A, I = 100 A and I = 200 A.
20
Voltage [V]

26 15
24 10
Voltage [V]

22
5
20
0
18 I = 50 A 0 10 20 30 40 50 60
16 I = 100 A Time [min]
I = 200 A
14
0 10 20 30 40 50 60 Figure 3. Battery voltage curve using a 10th -order
Time [min] polynomial for T = 20 ◦ C.

Figure 1. Battery voltage curve for T = 20 ◦ C – each mark


represents a value captured manually. Figure 4 shows the curves used to fit the captured data for
discharge current I = 50 A and temperature T = −20 ◦ C,
T = 0 ◦ C, T = 20 ◦ C and T = 40 ◦ C.
Figure 2 shows the points captured for a battery discharge
current I = 50 A and room temperatures T = −20 ◦ C, T = The Paatero Battery Model
0 ◦ C, T = 20 ◦ C and T = 40 ◦ C.
Paatero [4] proposed a nickel-cadmium battery model. The
The voltage data was captured for: battery voltage is divided into two parts: the open-circuit volt-
age Uoc and the overpotential Uop . The open circuit voltage
• discharge current: 10 A, 50 A, 100 A, 200 A and 500 A;
Uoc is defined as
• temperature: −20 C, 0 C, 20 C and 40 C.
◦ ◦ ◦ ◦

Uoc = a + b · DOD + (c + d · DOD) · T (1)


Each pair constant discharge current I and constant tempera-
ture T results in a voltage curve U (t). Therefore 20 voltage where T is the battery temperature, DOD means Depth of
curves with 50 points in each curve were manually captured. Discharge (DOD = 1 − SOC, where SOC is the state of
It was decided that each voltage curve U (t) would be approx- charge) and a, b, c and d are the model parameters to be fitted.
imated by a Nth -order polynomial function with time (in min- The DOD value is calculated as
utes) as its argument. After plotting the sum of the mean
Z t
squared fitting error for each curve against the order of the 1
DOD = I(τ )dτ (2)
polynomials, the latter was selected as 10 [3]. Cn 0
Table 2. Parameters for the original Paatero battery model.
25

20
Parameters Value
Voltage [V]

15
a 1.8697 × 101
10 T = 40 ◦ C b −1.5928 × 100
T = 20 ◦ C c −5.8751 × 100
5 T = 0 ◦C d 1.8267 × 10−2
0 T = −20 ◦ C x1 −2.2324 × 10−1
0 10 20 30 40 50 60 70 x2 −5.8855 × 100
Time [min]
x3 1.3887 × 100
x4 1.0041 × 10−2
Figure 4. Battery voltage curve using a 10th -order x5 1.6818 × 101
polynomial for I = 50 A. x6 2.3401 × 10−7
x7 1.1886 × 101
x8 6.7330 × 10−4
x9 −1.0902 × 10−1
where Cn is the nominal battery capacity – a parameter spec- x10 2.2188 × 100
ified by its manufacturer – and I(t) is the discharge current x11 7.5574 × 100
at instant t. It is assumed that the battery is fully charged x12 −1.0355 × 10−4
(DOD = 0) at t = 0. The overpotential Uop is defined as x13 −2.5977 × 100

x5
Dependent variable

Dependent variable
Uop = x1 + x2 · T + x3 · DOD + x4 · |I| +
|I|
+ x6 · ex7 ·DOD + x8 · ex9 ·T + x10 · |I|
 

+ x11 · tanh (x12 · DOD + x13 ) , (3)

where I is the battery discharge current and xi (i = 1 to 10) Independent variable Independent variable
are the model parameters that have to be calculated.
Figure 6. The R2 index.
The total battery voltage is defined as

U = Uoc − Uop . (4)


for all points. The result is defined as the TOTAL SUM OF
Table 2 shows the parameters for the Paatero model that were SQUARES .
calculated using the Matlab function fminsearch, which
implements the least-squares algorithm, and the 20 polyno- The right graphic in Figure 6 shows the distance between each
mial functions that approximate the original data. original point and the proposed fitting curve. The sum of the
squared values of all distances is defined as the RESIDUAL
Figure 5 shows the polynomial functions that are used to ap- SUM OF SQUARES . The equation
proximate the original data and the output Paatero model for
RESIDUAL SUM OF SQUARES
each pair (I, T ): constant discharge current, constant room R2 ≡ 1 − (5)
TOTAL SUM OF SQUARES
temperature. For clarity reasons, Figure 5 shows only 16 out
of 20 cases. shows how the R2 index is defined.

The R2 index was used to measure the quality of the results. Table 3 shows the R2 index for each pair constant discharge
The R2 index measures the goodness of fit of a model. If the current and constant room temperature.
R2 index is 1, the model fits the data in a perfect manner. If
it is 0, the model is as good as the mean value of the data. Since there are some values in Table 3 that are less than 0.7, it
Typically, values greater than 0.7 indicates a good model and is important to find an alternative model that fits the data con-
values lower than 0.3 indicates that the model is not represen- sidering the full range of temperature and discharge current.
tative.
3. I MPROVING THE PAATERO M ODEL U SING
Figure 6 shows two graphics. The left one shows: a) the origi-
THE BATTERY C APACITY
nal points, b) the proposed curve to fit the original points and,
c) a horizontal line at the mean value of the original points. The accuracy of the Paatero model can be improved by con-
The distance between each original point and the horizontal sidering that the battery capacity is not constant (see Equation
line is calculated. The distances are squared and summed 2) but varies with its discharge current and its temperature.
I = 10 A and T = 40 ◦ C I = 50 A and T = 40 ◦ C I = 100 A and T = 40 ◦ C I = 500 A and T = 40 ◦ C
20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 200 400 0 20 40 60 0 10 20 30 0 2 4 6
Time [min] Time [min] Time [min] Time [min]

I = 10 A and T = 20 ◦ C I = 50 A and T = 20 ◦ C I = 100 A and T = 20 ◦ C I = 500 A and T = 20 ◦ C


20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 200 400 0 20 40 60 0 10 20 30 0 2 4 6
Time [min] Time [min] Time [min] Time [min]

I = 10 A and T = 0 ◦ C I = 50 A and T = 0 ◦ C I = 100 A and T = 0 ◦ C I = 500 A and T = 0 ◦ C


20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 100 200 300 400 0 20 40 60 0 10 20 30 0 2 4
Time [min] Time [min] Time [min] Time [min]

I = 10 A and T = −20 ◦ C I = 50 A and T = −20 ◦ C I = 100 A and T = −20 ◦ C I = 500 A and T = −20 ◦ C
20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 100 200 300 0 20 40 60 0 10 20 30 0 2 4
Time [min] Time [min] Time [min] Time [min]

Figure 5. The battery voltage curves and the Paatero model output using the parameters shown in Table 2 – ◦: data and ×:
model.

Table 3. R2 index for each pair constant discharge current temperature can be well modeled by an hyperbola. The hy-
and constant room temperature. perbola equation used in this model is
b1 + T
−20 ◦ C 0 ◦C 20 ◦ C 40 ◦ C C(T ) = a1 , (6)
c1 + T
10 A −0.080 −0.275 −0.565 −0.473 where a1 , b1 and c1 are the function parameters. These pa-
50 A 0.951 0.970 0.896 0.883 rameters shall be solved in order to fit the function to the data.
100 A 0.867 0.866 0.896 0.924 Figure 7 shows how the function fits the data. The parameters
200 A 0.804 0.881 0.895 0.975 were calculated using the least squares method implemented
500 A 0.239 0.762 0.817 0.993 by the Matlab function fminsearch.

The battery capacity also varies with the discharge current as


shown in Figure 8.
Battery Capacity Model
In Section 2 each battery voltage curve was approximated In this case, the changes in the battery capacity can be well
by a 10th -order polynomial function. Each polynomial func- modeled by the equation
tion is then used to perform extrapolation, that is, the voltage C(I) = d1 + e1 I + f1 arctan(g1 + h1 I), (7)
curve is extended until the time it reaches 0 V is found. This
time multiplied by the constant discharge current is defined where d1 , e1 , f1 , g1 and h1 are the parameters of the function.
as the battery capacity at that particular discharge current and These parameters shall be solved in order to fit the function
room temperature. to the data. Figure 8 also shows how the function fits the data.
Equation 7 describes the linear part of the curve as well the
The changes in the battery capacity in relation to the room step in its middle.
ters, the parameters for the Paatero model were recalculated
55
and are shown in Table 5. Figure 9 shows that the modified
50
Capacity [Ah]

45 Table 5. Parameters for the modified Paatero model.


40
Parameters Value
35 data I = 50 A
30 data I = 200 A a 2.6058 × 101
data I = 500 A b 1.0606 × 101
25
c 6.7958 × 10−3
−30 −20 −10 0 10 20 30 40
d −7.8309 × 10−3
Temperature [◦ C] x1 1.8143 × 101
x2 −1.5704 × 10−2
Figure 7. Battery capacity as function of room temperature. x3 1.2528 × 101
x4 −1.5316 × 100
x5 1.3647 × 100
x6 −4.5473 × 100
55
x7 1.5962 × 10−5
x8 4.9049 × 100
Capacity [Ah]

50
x9 −1.4581 × 10−4
45 x10 3.3144 × 100
data T = 40 ◦ C
x11 −1.7183 × 101
40
data T = 0 ◦ C x12 −1.1431 × 101
35 data T = −20 ◦ C x13 1.1311 × 101

0 50 100 150 200 250 300 350 400 450 500


Paatero battery model provides a good fit for the polynomial
Current [A]
functions which are used to approximate the original battery
data.
Figure 8. Battery capacity as function of discharge current.
Table 6 shows the R2 values considering the modified Paatero
model. All the values are above 0.7 indicating that the pro-
Equations 6 and 7 can be joined in a single equation. The posed modified model correctly describes the battery be-
complete battery capacity model haviour over a large range of discharge current and temper-
ature.
b1 + T
C(I, T ) = a1 (d1 + e1 I) Table 6. R2 value for each discharge current and room
c1 + T temperature.
+ (f1 + g1 arctan(h1 + i1 I)) (8)

is the equation used to describe the battery capacity as func- −20 ◦ C 0 ◦C 20 ◦ C 40 ◦ C


tion of its discharge current and its room temperature. 10 A 0.970 0.995 0.975 0.954
50 A 0.991 0.988 0.987 0.961
Table 4 shows the parameters calculated for the battery ca- 100 A 0.988 0.989 0.991 0.975
pacity model expressed by Equation 8. Using these parame- 200 A 0.985 0.986 0.984 0.989
Table 4. Battery capacity equation parameters. 500 A 0.917 0.956 0.966 0.938

Parameters Value 4. M ODEL S IMULATION


a1 2.2385 × 101 The modified Paatero model was implemented in Matlab &
b1 5.7594 × 101 Simulink R2008b in order to simulate an aircraft electrical
c1 3.5629 × 101 emergency. The battery capacity was evaluated by the block
d1 −2.4781 × 10−1 shown in Figure 10.
e1 −4.9694 × 10−4
f1 6.2292 × 101 The block function f (u) shown in Figure 10 implements the
g1 −1.1538 × 100 following function
h1 −9.4743 × 101
i1 7.9808 × 10−1 a1 ∗ (b1 + u(2))/(c1 + u(2)) ∗ (d1 + e1 ∗ u(1))
+(f1 + g1 ∗ atan(h1 + i1 ∗ u(1))) (9)
I = 10 A and T = 40 ◦ C I = 50 A and T = 40 ◦ C I = 100 A and T = 40 ◦ C I = 500 A and T = 40 ◦ C
20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 100 200 300 0 20 40 60 0 10 20 30 0 2 4 6
Time [min] Time [min] Time [min] Time [min]

I = 10 A and T = 20 ◦ C I = 50 A and T = 20 ◦ C I = 100 A and T = 20 ◦ C I = 500 A and T = 20 ◦ C


20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 100 200 300 0 20 40 60 0 10 20 30 0 2 4
Time [min] Time [min] Time [min] Time [min]

I = 10 A and T = 0 ◦ C I = 50 A and T = 0 ◦ C I = 100 A and T = 0 ◦ C I = 500 A and T = 0 ◦ C


20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 100 200 300 0 20 40 60 0 10 20 30 0 2 4
Time [min] Time [min] Time [min] Time [min]

I = 10 A and T = −20 ◦ C I = 50 A and T = −20 ◦ C I = 100 A and T = −20 ◦ C I = 500 A and T = −20 ◦ C
20
Voltage [V]

Voltage [V]

Voltage [V]

Voltage [V]
20 20 20

10
10 10 10

0 0 0 0
0 100 200 300 0 20 40 60 0 10 20 30 0 2 4
Time [min] Time [min] Time [min] Time [min]

Figure 9. The battery curve and the modified model output – ◦: data and ×: model.

1
1 60
I
f(u) 1 C
C Gain
Fcn 1
2
1 DOD
T 2 Divide
s
I
Figure 10. Simulink diagram used to evaluate the battery Integrator
capacity.
Figure 11. Simulink diagram used to evaluate the DOD.

with the parameters shown in Table 4.


utes. The discharge current is integrated during the time and
The DOD value is evaluated by the block shown in Figure the result is divided by the battery capacity.
11. This block uses as input the battery capacity and the dis-
charge current. This block implements the equation Figure 12 shows how the battery voltage is evaluated.
The open circuit voltage, Uoc , is evaluated by the block
calc_Uoc and the overpotential, Uop , is evaluated by the
Z t
1
DOD(t) = I(τ )dτ (10) block calc_Uop. The battery voltage is evaluated subtract-
C(I, T ) 0
ing the overpotential from the open circuit voltage.
where C(I, T ) is the battery capacity for the discharge cur-
rent I and the temperature T . The capacity value is multiplied The block calc_Uoc implements the function
by a gain in order to change the capacity unit from ampère-
hour to ampère-minute. The simulation is evaluated in min- a + b ∗ u(1) + (c + d ∗ u(1)) ∗ u(3) (11)
1680 W. The battery discharge current is evaluated using the
f(u) equation
1
P
DOD calc _Uoc I= , (13)
V
2 1
where P is the power reference adopted in the simulation.
I U
Add The load power divided by the battery voltage results in the
3 f(u)
current used as one of inputs in the main blocks. The Unit De-
T lay block was used to avoid an algebraic loop in the simula-
calc _Uop tion. Its output is equal to its input delayed by one simulation
Figure 12. Simulink diagram used to evaluate the battery step (sample time = −1). The simulation uses a variable-step
voltage. solver (ode45) with 0.1 s as its maximum step size.

Figure 14 shows the battery voltage curve for this case. At

25
while the block calc_Uop implements the function
20
x1 + x2 ∗ u(3) + x3 ∗ u(1) + x4 ∗ abs(u(2))

Voltage [V]
15
+x5 /abs(u(2)) + (x6 ∗ exp(x7 ∗ u(1)) + x8 )
10
∗(exp(x9 ∗ u(3)) + x10 ) ∗ abs(u(2))
+x11 ∗ tanh(x12 ∗ u(1) + x13 ). (12) 5

Both blocks are implemented with the parameters shown in 0 5 10 15 20 25 30 35


Table 5. Time [min]

The simulation considers an electrical emergency situation Figure 14. Simulation results for the constant power load
where the battery feeds the emergency load of an aircraft. case.
For the simulation, the temperature was defined to be T =
−20 ◦ C. This is the worst case regarding temperature.
t = 30 min the battery voltage is 22.3 V.
Table 7 shows the nominal load data for a reference aircraft.
Figure 15 shows a simulation for constant resistance load.
Table 7. Load nominal values.
I
C
Parameter Value T
C
C = f(I,T) DOD
Voltage 24 V −20 I Modelo de Paatero
DOD
T DOD = INT(I)/C
Current 70 A I U

Power 1680 W
T
Scope
U = f(DOD,I,T)
1
70/24
z
Unit Delay
Figure 13 shows the diagram used to simulate an electrical Gain

emergency when the battery feeds a constant power load. In Figure 15. Simulation diagram for the constant resistance
load case.
I
C
T
C
C = f(I,T) DOD
−20 I Modelo de Paatero

T DOD = INT(I)/C DOD


I U
The discharge current is calculated using the Ohm’s law
T
Scope
U = f(DOD,I,T)
V
1 I= , (14)
z 1680 R
Unit Delay Divide
P

Figure 13. Simulation diagram for the constant power load where R is the resistance value. The Gain block contains the
case. inverse of the resistance load.

In this case the value of the resistance is the nominal resis-


this case, the load and the battery have the same power – tance which is calculated by dividing the nominal voltage by
the nominal current – 24/70 V/A = 0.3428 Ω. Figure 16 5. C ONCLUSION
shows the battery voltage for this simulation. At t = 30 min,
This article shows a procedure to generate a nickel-cadmium
the battery voltage is 22.5 V.
battery discharge model using only the data (voltage curves
24 over time) supplied by the battery manufacturer. The results
shows that the battery model fits well the data over a large
23
range of discharge currents and room temperatures. The ar-
22 ticle shows how the baterry model can be used in Matlab &
Voltage [V]

21 Simulink to simulate an electrical emergency with three types


20 of loads: constant power, constant resistance and constant
19
current.
18
Some possible lines for future research are:
0 5 10 15 20 25 30 35 40
Time [min]
• comparison of the simulation results for emergency situa-
tions with the data collected by bench testing the battery in
Figure 16. Simulation results for the constant resistance
the same situations; and
load case.
• generation and validation of a nickel-cadmium charging
model.
Figure 17 shows the simulation diagram for a constant current
load. The nominal current load – 70 A – was used as the
current load. R EFERENCES
[1] E. Safford, “Test automation framework, state-based and
70 I
C
signal flow examples,” in Twelfth Annual Software Tech-
I T
C nology Conference, Salt Lake City, UT, 30 Apr.-5 May
C = f(I,T) DOD
I Modelo de Paatero 2000.
−20 DOD = INT(I)/C DOD
U
[2] B. W. Boehm, Software Engineering Economics. Pren-
I
T
T
Scope
U = f(DOD,I,T)
tice Hall, November 1981.
Figure 17. Simulation diagram for the constant current load
[3] G. S. Sperandio, “Nickel-cadmium batteries model for
case.
discharge simulation,” (in Portuguese) Professional Mas-
ter Dissertation, Instituto Tecnológico de Aeronáutica,
São José dos Campos, Brazil, 2010.
Figure 18 shows the battery voltage for this simulation. At
t = 30 min, the battery voltage is 22.5 V. [4] J. Paatero, “A mathematical model for flooded nickel cad-
mium battery,” Helsinki University of Technology, De-
25
partment of Physics, Tech. Rep., 1997.
20
Voltage [V]

15 B IOGRAPHY [
10

0
0 5 10 15 20 25 30 35 40 45 Giuliano S. Sperandio received his
Time [min] B.Sc. degree in Electrical Engineer-
ing from Universidade Federal do Es-
Figure 18. Simulation results for the constant current load pírito Santo (UFES), Vitória, Brazil, in
case. 2005, the M.Sc. degree in Electrical
Engineering from Universidade Estadual
de Campinas (UNICAMP), Campinas,
The simulation results shows that battery could supply the 3 Brazil, in 2007. He is currently a grad-
types of loads for a time longer than 30 minutes, which means uate student in Aeronautical Engineering from Instituto Tec-
that it satisfies the emergency requirement for the reference nológico de Aeronáutica (ITA), Brazil. His current research
aircraft. Given information about the load, other situations interests are power electronics, switching mode power supply
can also be easily simulated. and batteries.
Cairo L. Nascimento Jr. received the
B.Sc. degree in Electrical Engineer-
ing from the Universidade Federal de
Uberlândia (UFU), Uberlândia, Brazil,
in 1984, the M.Sc. degree in Electron-
ics Engineering from the Instituto Tec-
nológico de Aeronáutica (ITA), São José
dos Campos, Brazil, in 1988, and the
PhD degree in Electrical Engineering from the University of
Manchester Institute of Science and Technology (UMIST),
Manchester, UK, in 1994. Since 1986, he has been a lec-
turer with the Division of Electronics Engineering, ITA, and
has supervised 4 Ph.D. and 14 M.Sc. students. He is the
co-author of a book on intelligent systems and was the chair-
man for the 1999 Fourth Brazilian Conference on Neural Net-
works. His current research interests are the application of
artificial intelligence techniques to Prognostics and Health
management (PHM), intelligent mobile robotics, and the de-
velopment of internet-based remote-access laboratories for
engineering education. Dr. Nascimento Jr. is a member of
the Brazilian Society for Automation (SBA).

Geraldo J. Adabo received the B.Sc.


degree in Electrical Engineering from
the Universidade de São Paulo (USP),
São Carlos, Brazil, in 1981, and the
M.Sc. degree in Electronics Engineering
from the Instituto de Pesquisas Espaciais
(INPE), São José dos Campos, Brazil, in
1985. Since 1987, he has been a lecturer
with the Division of Electronics Engineering, ITA. His cur-
rent research interests are power electronics, satellite power
supply, aircraft electrical systems, communication circuits,
RF/Photonics and unmanned aerial vehicles.

You might also like