Hindawi

Wireless Communications and Mobile Computing

Volume 2021, Article ID 4211010, 18 pages
[Link]

Research Article
The Development of Data Acquisition System of Formula SAE
Race Car Based on CAN Bus Communication Interface and
Closed-Loop Design of Racing Car

Zhonghao Li , Da Wang , and Qiao Kang

Jilin University, College of Automotive Engineering, Jilin University, Changchun, China

Correspondence should be addressed to Da Wang; wangda_gspeed@[Link]

Received 21 June 2021; Revised 13 July 2021; Accepted 23 July 2021; Published 26 August 2021

Academic Editor: Balakrishnan Nagaraj

Copyright © 2021 Zhonghao Li et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

This paper is mainly based on the college formula race car, through the application of CAN bus communication interface settings,
the layout and adjustment of the car sensors, and then carry out the data collection and analysis of the car. CAN bus is the local area
network of controller and is one of the field bus which is put forward by German Bosch company and applied most widely in the
world. The development of the data analysis system is mainly based on the MoTeC i2 data analysis software. MoTeC i2 data analysis
software has been developed for many years under the continuous research of professional teams all over the world. According to
the data analysis requirement of FSAE undergraduate Formula 1 teams, this paper introduces a series of data analysis methods
closely related to FSAE, mainly including data channel research. This paper expounds the reliability, suspension, tire
temperature, steering, and braking data analysis, respectively. Based on the data analysis template created by Motec i2, this
paper conducts analysis and basic application according to the data of the real car, so as to meet the requirements of verifying
design made by simulation model and tuning the real car scientifically. By means of wheel speed sensor, acceleration sensor,
displacement sensor, multichannel temperature sensor, and so on, we conduct a series of tests on 2020 season car, including
endurance test, linear acceleration experiment, steady rotation, etc., evaluate the data collected, and then carry on a reasonable
set-up of racing, ultimately improving lap speed.

1. Introduction excellent performance, can successfully complete all or part

of the race links. China FSC is committed to building a pub-
Formula SAE (FSAE) was first proposed by an American fac- lic platform for the training and selection of outstanding
tory in 1978. This race is held for college students to collect domestic automotive talents, and through general assess-
small formula cars with creative design and scientific ideas. ments to improve students’ comprehensive capabilities in
These cars are stable and durable in all aspects of perfor- five aspects, including design, manufacturing, cost control,
mance. Compared with foreign countries, our domestic col- commercial marketing, communication, and coordination.
lege student formula racing started late. The first Formula In order to improve the comprehensive quality of automo-
Student China (FSC) competition was held at the Shanghai tive students, make long-term talent accumulation for the
International Circuit in 2010. Formula Student of China development of China’s automotive industry, and promote
(“China FSC”) is a car design and manufacturing competi- the strategic direction of China’s automotive industry from
tion organized by students majoring in automotive engineer- a “manufacturing country” to an “industrial power” [1].
ing or automotive related in universities. Each participating Gspeed Electric Racing Team of Jilin University has partici-
team in accordance with the rules of the race and racing car pated in the FSC series for 5 consecutive years. In 2019, it
manufacturing standards, in one year’s time to design and went to Japan to participate in the Formula Student Japan
manufacture a small single-seat recreational racing car in and won the second prize of design project many times. In
the acceleration, braking, handling, and other aspects of 2018, it won the first prize of the FSC. GspeedE-2019 and
2 Wireless Communications and Mobile Computing

GspeedE-2020 race cars are shown in Figures 1 and 2,

respectively. When we design a racing car, we often fail to
make a reasonable design for some parameters of each sys-
tem, and the rough design fails to support and prove the
completed design with data. Therefore, we installed a data
acquisition system on our race car, that is, the sensors we
need and the wiring harnesses which connect them to a
low-voltage electrical system and panel. Then, we used
MoTec i2 to deal with the data we collected. Due to the large
proportion of engineering design events and dynamic events
in Formula Student [2], data analysis of race cars, which plays
an important role in verifying design, guiding tuning, and fur-
ther contributing to future design, do show its important
value. Therefore, we do research on analysis methods of race
cars and put it into practice by using MoTec i2 templates Figure 1: GspeedE-2019 racecar.
and math channels, which quickly analyze the large amount
of data generated by each test. In this way, we obtain the direc-
tion of tuning and improve the performance of our car.
At the same time, the data acquisition and analysis sys-
tem of the whole car is a feedback system for driver training
and making suggestions for tuning [3]. Through scientific
road test and data analysis, the whole process can form a
closed-loop design. In addition, the driving habits and styles
of different drivers can be summed up by recording and ana-
lyzing the same project they conduct. In this way, different
car set-up methods can be matched for different drivers.
Figure 3 shows the flow chart of data collection and analysis
of the race car.
Figure 2: GspeedE-2020 racecar.
The sensor interface setup and connection of the vehicle
are based on CAN bus communication protocol, among
which, the car has three CAN communication networks: troubleshoot the fault. For Formula E racing, the important
reliability data consist of high voltage battery voltage, high
(1) Motor control network, connecting the vehicle con- voltage battery temperature, motor output power, low volt-
troller with the motor controller, to realize the func- age battery voltage, etc. Second, there are some reliability
tion of sending the torque instruction of the drive indicators, such as tire pressure, and brake pressure.
system and obtaining the working state of the motor Figure 5 is the reliability data template.
of the drive system To check the reliability data, the first step is to observe the
maximum and average values of each channel, and the sec-
(2) The conventional information communication net-
ond step is to examine its changing trend. The observation
work realizes the information transmission between
of maximum value can determine whether the data is within
the vehicle controller and the battery management sys-
safe range, while the observation of variation trend can deter-
tem, instrument panel, sensors, and other equipment
mine whether the car is running normally during the whole
(3) In order to charge the communication network, the process and can predict the state of the car when running
experimental information transmission between the continuously. Especially in endurance test, the data of battery
battery management system and the charger temperature and power usually change linearly.
In order to know the condition of reliability parameters
Using the MoTeC C125 racing digital instrument, ECU during each run, we set up an alarm mechanism on the car
uses CAN-B to collect and record training data for improved instrument panel. In the endurance race, it is one of the key
vehicle performance. factors to detect abnormal reliability parameters in advance
Figure 4 shows the CAN bus network of the whole vehicle. and make a right decision.
Similarly, an alarm mechanism can be set in the data
2. Reliability Data Analysis analysis software to find out the problematic channel data
in the first time.
As we know, one of the most important functions of the data If the voltage provided by the power battery is found to be
acquisition system is to find the potential faults of race car by too low, it is considered that the power supply system of the
monitoring the reliability data, which determines whether battery box body has poor contact and the battery module
the car can run normally or not. So we should make obtain- bolts have loosened. At this time, the battery box should be
ing and checking reliability data a top priority. Once any- inspected and the battery management system (BMS) should
thing abnormal is found, stop the car immediately and be used for fault detection.
Wireless Communications and Mobile Computing 3

Tire pressure
High and Battery
Power and brake
low voltage temparature
pressure

Reliability data Judge the safety Provide

analysis condition of the endurance
car racing strategies

Wheel Damper
Roll Roll
dynamic speed and
gradient stiffness Choose
vertical load force Verify Adjust anti-
Evaluate proper tire
stiffness rollbar and
steering pressure and
Suspension data design damper to
characteristics help tire data
analysis objective proper gear
analysis

Tire tempareture

Tire data Evaluate camber Evaluate tire

analysis angle pressure

Understeering angle Attitude angular velocity

Streering data Evaluate steering

analysis characteristics

Front and rear

Maximum Wheel lock
brake pressure
braking force condition analysis
distribution
Evaluate steering
Braking data Avoid
characteristics when
analysis wheel lock
braking in turn

Figure 3: Flow chart of data collection and analysis for racing car.

C125 instument

Steering angel sensor

Hall wheel speed sensor

Motor controller

Brake oil pressure sensor Vehicle controller

CAN-A CAN-B
Suspension displacement sensor

Multi-channel temperature sensor

Accelerator pedal sensor

Motor controller1 AMS/battery pack Charger

Motor controller2 Yaw rate sensor

CAN-A CAN-C
CAN-B Other signals

Figure 4: Vehicle CAN network.

3. Suspension System Data Analysis The purpose of suspension data analysis is to verify the
design and guide the adjustment through the compilation
The suspension system of this season’s Gspeed formula E and processing of the data. Detailed tuning projects consist
racing car is an independent suspension of unequal length, of tire pressure, shock absorber gear, antiroll bar stiffness,
with a third spring system and a T-shaped anti-roll bar. load distribution of front and rear tires, adjusting four-
This part includes the calculation of roll gradient and roll wheel positioning parameters, and carrying out the corre-
stiffness, load transfer, and the analysis of shock absorber. sponding fault troubleshooting.
4 Wireless Communications and Mobile Computing

18 Power (kw) 15.93

16
14
12
10
8
6
4
2
–0
60
Speed (km/h) 51.38
50

40
30

20

10

–0
400 Batvoltage (V) 374.0
350
300
250
200
150
100
50
–0

m:s 4:20 4:40 5:00 5:20 5:40

Figure 5: Reliability data template.

3.1. Roll Gradient (RG) Analysis. In vehicle dynamics, roll displacement of the shock absorber with the traditional linear
gradient refers to the roll angle of a vehicle per unit lateral displacement sensor. And when the suspension arrangement
acceleration. style is not suitable for using displacement sensor (for
Equation (1) is the definition formula of roll gradient RG. instance, a decoupling suspension with single spring
mounted between two rockarms), we can use angular dis-
αroll placement sensor to get suspension displacement indirectly
RG = : ð1Þ
Glat by measuring rockarm rotation angle. The angle and the sus-
pension displacement usually have high linearity, whose lin-
In the above formula, αroll is roll angle (deg); Glat is lateral ear coefficient can be fitted by means of parallel wheel jump
acceleration (g). simulation in Adams-Car [2]. After this work, what we need
Therefore, in order to calculate the roll gradient, the first to do is to set up math channel in MoTec i2. Figure 6 shows
parameter we need to get is the roll angle. The following is the the template of suspension system roll angle data.
mathematical channel formula of the front and rear suspen- It should be noted that the roll angle calculated from the
sion roll angle α F andαR and the total suspension roll angle α. suspension displacement is only the roll angle caused by the
In order to get the roll angle of the vehicle, the left and right lateral transfer of sprung mass, that is, the suspension roll
suspension displacements should be calculated, and the trig- angle. Therefore, the following calculation is also the roll gra-
onometric function formula should be used to solve the dient of the suspension system using the roll angle of the
problem. Table 1 shows the vehicle parameters required by suspension.
formulas (2)–(4). Then, we create a scatter graph with the vertical axis as
roll angle and the horizontal axis as lateral acceleration and
"
 #
xSuspLF − xSuspRF MR F 180 obtain the trend line equation, whose slope is the roll gradi-
α F = ArcTan · , ð2Þ ent of the suspension system. Figure 7 is the roll gradient dia-
TF π
gram of GspeedE-2019 in a test. From the equations in the
"
# lower left corner of each template in the figure, it can be seen

xSuspLR − xSuspRR MRR 180 that the roll gradients of front suspension, rear suspension,
αR = ArcTan · , ð3Þ and total suspension are 0.552°/G, 0.557°/G, and 0.555°/G,
TR π
respectively.
"

 #
xSuspLF − xSuspRF MR F + xSuspLR − xSuspRR MRR 180 3.2. Roll Stiffness. The roll stiffness of suspension refers to the
α = ArcTan · :
T F + TR π total elastic recovery couple moment given by the suspension
ð4Þ system to the carriage under the unit carriage rotation angle
when the vehicle rolls (the wheels remain on the ground).
Among these parameters, suspension displacement is The roll stiffness of the suspension will affect the lateral load
what we should detect. It can be obtained by measuring the transfer of the car during steering and then affect the wheel
Wireless Communications and Mobile Computing 5

Table 1: Parameters in formulas (2)–(4). Roll moment per lateral acceleration M ϕ :

Parameter Description
M ϕ = W S hRM : ð7Þ
xSuspLF Left front suspension displacement
xSuspRF Right front suspension displacement Then, the total suspension roll stiffness K ϕ is
xSuspLR Left rear suspension displacement
Mϕ
xSuspRR Right rear suspension displacement Kϕ = = 1161:379N · m/∘ : ð8Þ
RG
MR F Front suspension motion ratio
MRR Rear suspension motion ratio According to the distribution of front and rear roll stiff-
TF Front track width ness, the front and rear suspension roll stiffnessK ϕF and
TR Rear track width
K ϕR can be calculated as follows:

RGR
1 Front suspension roll angle (°) K ϕF = K ϕ · = 583:308N ⋅ m/∘ , ð9Þ
RG F + RGR

0 K ϕR = K ϕ − K ϕF = 578:072N ⋅ m/∘ : ð10Þ

Because in many cases, the roll stiffness provided by the

–1 spring cannot meet the requirement, so the antiroll bar
1 Rear suspension roll angle (°)
mechanism is added. Therefore, suspension roll stiffness is
the sum of the roll stiffness provided by spring and antiroll
0 bar. Table 3 shows the vehicle parameters required by formu-
las (11) and (12).
–1
1 Total suspension roll angle (°) K ϕF = KSpringF + KARBF , ð11Þ

0 K ϕR = KSpringR + KARBR : ð12Þ

In the above formula, KSpringF (N · m/∘ ) is the roll rate of

–1
the front spring; KSpringR (N · m/∘ ) is the roll rate of the rear
5:50 7:30
spring; KARBF (N · m/∘ ) is the roll rate of the front antiroll
Figure 6: Template of suspension system roll angle data. bar; KARBR (N · m/∘ ) is the roll rate of the rear antiroll bar.
After calculating the roll stiffness of the spring through
the known number of spring pounds, the roll stiffness of
load of the inside and outside wheels as well as the front and the antiroll bar can be further calculated [5].
rear wheels. The difference of wheel load will lead to the dif- By comparing the roll gradient and roll stiffness with
ference of the slip angle of the tires and then affect the steer- the design objective, the gap between the actual race car
ing characteristics [4]. According to the above roll gradient and the simulation design can be obtained. If there is a
analyzed by MoTec i2 software and the vehicle parameters large error, it is necessary to examine the specific cause
provided in Table 2, the roll stiffness of the suspension sys- of the error (for example, processing or assembly problem
tem and its components can be further calculated. Among of parts) and solve it in time. In the test and tuning of real
them, the collection and analysis result of the roll gradient vehicle, after the ideal tuning strategy is obtained, the cor-
is RG F = 0:552°/GRGR = 0:557°/G RG = 0:555°/G . responding roll gradient can be used as the reference aim
We can obtain the following parameters by taking the of the next season.
torque on the ground:
Height of sprung mass center hS : 3.3. Dynamic Vertical Load of Tires. The vertical load
directly affects the grip performance of the tire, while the
load distribution between front and rear affects the steer-
W T h − W UF R F − W UR RR ing characteristics of the car. In addition, the amount of
hS = : ð5Þ
WS tire load transfer can also affect the aerodynamic perfor-
mance of the car. Therefore, knowing the tire dynamic
vertical load is beneficial to evaluating the balance of a
Roll moment lever arm hRM : racing car, as well as the choice of tire pressure and anal-
ysis of tire surface temperature. By optimizing the four-
wheel positioning parameters and other tuning methods,
hRM ≅ hS − ½Z F + ðZ R − Z F Þð1 − aS Þ = 0:252m: ð6Þ the tire load transfer of the racing car can be reduced
6 Wireless Communications and Mobile Computing

1.0 Front suspension roll angle (°) 1.0 Rear suspension roll angle (°)

0.5 0.5

0 0

–1.5 –1.5

y=0.552x+0.00567 y=0.552x+0.00571
–1.0 G Force lat (G) –1.0 G Force lat (G)
–1.5 0 1.5 –1.5 0 1.5
1.0 t l suspension
Total n i rroll
ll angle
gl (°)

0.5

–1.5

yy=0.552x+0.00534
= 5522x+0.00534
=0.
=0 +0 00534
0
–1.0 GFForcee llat (G
(G)
G
–1.5 0 1.5

Figure 7: Roll gradient data template.

Table 2: Basic parameters of racing car.

Parameter Description Value : front Rear Total

WT Total mass (kg) 134.2 170.8 305
WU Unsprung mass (kg) 21 23 44
WS Sprung mass (kg) 113.2 147.8 261
h Height of mass center (m) 0.28
Z Height of roll center (m) 0.03 0.04
R Tire load radius (m) 0.235 0.235
T Track width (m) 1.266 1.241
L Wheel base (m) 1.575
aS Sprung mass distribution on the front suspension (%) 45

Table 3: Parameters in formulas (11) and (12). and balanced as far as possible, and the ultimate perfor-
Parameter Description mance of race car can be further improved.
In order to calculate the dynamic vertical load of each
K SpringF Roll rate of the front spring wheel, it is necessary to understand the various parameters
K SpringR Roll rate of the rear spring and influencing factors involved. And Figure 8 shows each
K ARBF Roll rate of the front antiroll bar
component of wheel load.
The vertical load on each wheel is the sum of the compo-
K ARBR Roll rate of the rear antiroll bar nents listed above. The detailed analysis and calculation are
as follows.
Wireless Communications and Mobile Computing 7

Lateral weight transfer of

Statical wheel load
the sprung mass

Lateral weight transfer of

Lateral load transfer
the unsprung mass

Lateral geometric weight

Wheel vertical load Longitudinal load transfer
transfer

Influence of track
inclination and slope

Aerodynamic effects

Figure 8: Each component of wheel load.

W uR · Glat · RR
ΔW uR = : ð14Þ
TR
It should be noted that in the early design, we used the
tire load radius as an approximate value of unsprung mass
center height. After the design work is completed, the height
of wheel carrier mass center can be calculated in the 3D
modeling software, which can be used as a more accurate
value of unsprung mass center height.

(ii) Lateral geometric load transfer ΔW g

WsF · Glat · ZF · ðS/LÞ

ΔW gF = , ð15Þ
TF
Figure 9: Static wheel load measurement.
WsR · Glat · ZR · ð1 − S/LÞ
ΔW gR = : ð16Þ
TR
3.3.1. Static Wheel Load. We can measure the static load of In the above formula, S is the distance (m) from the geo-
each wheel when the car is in the ready state (that is, the metric center of rear axle to the sprung mass center in the
car with fully equipped driver). Figure 9 shows static wheel side view. The calculation formula is as follows:
load measured by loadometer.
WSF
S=L· : ð17Þ
3.3.2. Lateral Load Transfer. We can measure the force of the WS
suspension push rod through the pull pressure sensor, and
then we can get the curve of tire load change with time. After It should be noted that the unsprung mass and the geo-
that, by using the mechanical formula to calculate the other metric load transfer cannot be measured by pull pressure
force elements of the suspension system, we can get the value sensor and displacement sensor. Except lateral acceleration,
of the downward pressure produced by the aerodynamic other parameters are in Table 1, considered as theoretical
package on the race car, which is conducive to the design of value.
the aerodynamic package on the racing car. Figure 10 shows
the data collected by the pressure sensor. (iii) Lateral load transfer of the sprung mass ΔW s

(i) Lateral load transfer of the unsprung mass ΔW u W s · Glat · hRM

ΔW sF = · q, ð18Þ
TF

W uF · Glat · R F W s · Glat · hRM

ΔW uF = , ð13Þ ΔW sR = · ð1 − qÞ: ð19Þ
TF TR
8 Wireless Communications and Mobile Computing

200 Susp force FL (kg) 73.8

150
100
50
–0
200 Susp force FR (kg) 62.0

150
100
50
–0
Susp force RL (kg) 135.4
200
150
100
50
–0
Susp force RR (kg) 55.1
200
150
100
50
–0
m:s 4:20 4:40 5:00 5:20 5:40

Figure 10: Data acquisition by pressure sensor.

L dFL (N)
LoadFL (N 6
688.69
1200

600

LoadFR
L dFR (N)
(N 655 76
7
655.76
1200

600

39 88
8
839.88
1200 LoadRL (N)
(

600

LoadRR
L dRR (N) 04 68
6
804.68
1200

600

m:s 0:20 0:40

Figure 11: Tire dynamic load template.

The definition of q is transfer on front and rear suspension compared with total
amount and then calculate respective values.

K ∅F RGR 3.3.3. Total Longitudinal Load Transfer ΔW Long .

q= = : ð20Þ
K ∅F + K ∅R RG F + RGR
W · GLong · h
ΔW Long = : ð21Þ
From the above equations, we can see the importance of L
calculating roll gradient. The roll gradient can not only be
used to further calculate the roll stiffness, verify whether the To sum up, we can find that a large part of the calculation
real car reaches the preliminary design expectation, and of load transfer is to calculate the value of mechanical con-
guide the adjustment of spring pounds and antiroll bar stiff- cepts such as the mass center height, the roll center height,
ness but also reflect the proportion of the lateral sprung mass and the roll moment lever arm, while the lateral and
Wireless Communications and Mobile Computing 9

Total
wheelloads

Unsprung lateral
and longitudinal
weight transfer

Geometric lateral
Spring motion ratio Spring oreload Anti-rollbar motion ratio and longitudinal
weight transfer

Aerodynamics
forces

Shok/spring travel Spring force at

potentiometer wheel

Shock force versus shock Shock force at

speed wheel

Anti-rollbar force at
Anti-rollbar torsion
wheel

Figure 12: Tire load collection and calculation process.

Damper vel FL (mm/s)

D m 1.7
150
100
50
0
–50
–100
–150

10 Susp
u pos FL (mm)
(mm) 0 37
0.37
8
6
4
2
0
–2
–4
–6
–8
–10
–12
–14
m:s 1:40 2:30 3:20 4:10 5:00

Figure 13: Displacement and velocity data acquisition diagram of shock absorber.

longitudinal accelerations are obtained by the digital acquisi- Center in Xiangyang, Hubei Province, which is authorized
tion equipment. by the National Automobile Testing Ground. As a result,
road conditions can be guaranteed.
3.3.4. Influence of Track Slope. Since FSAE racing venues are
usually well paved, this effect is ignored here. For example, 3.3.5. Aerodynamic Effect. The aerodynamic effect on car can
the 2020 season was held at the Da’an Automotive Testing be measured with relevant sensors. When all components are
10 Wireless Communications and Mobile Computing

Percent
Damper_speed_FL Damper_speed_FR

30

20

10

Damper_speed_RL Damper_speed_RL

30

20

10

–60 –30 0 30 60 –60 –30 0 30 60

Figure 14: Damping velocity histogram data acquisition graph.

known, the total vertical load on each wheel can be calcu-

lated. Figure 11 is the dynamic load data of the tire in a skid-
pad track event (the skidpad track consists of two pairs of
concentric circles in a figure of eight pattern) [3].
According to the average value in the upper right corner
of the tire load graph above, the load transfer in the skidpad
test is distributed as the front 48% and the rear 52%. Theoret-
ically, neutral steering occurs when the distribution of load
transfer is equal to the distribution of axle load [1]. Since
the axle load distribution of the real car is the front 45%
and the rear 55%, it can be concluded that the steering char-
acteristics of the car are very close to the neutral steering.
According to the tire dynamic load, we can also choose
more appropriate tire pressure according to the tire file pro- Figure 15: Position of tire temperature sensor.
vided by manufacturer. In FSAE, the straight line accelera-
tion, skidpad event, and autocross event are performed in To get the velocity, we can take derivative of the shock
cold tire conditions [3], so in these events, the tire pressure absorber displacement signal.
can be directly selected according to the tire load that
required lateral and longitudinal acceleration determine,
without extra consideration of the tire temperature influence. d
The collection and calculation processes of tire load are vdam = s : ð23Þ
dt dam
shown in Figure 12.
3.4. Damper Analysis Study. For the data acquisition and And Figure 13 shows the displacement and velocity data
analysis system, the analysis of the shock absorber is mainly acquisition diagram of shock absorber.
to analyze its velocity through the displacement sensor. In part 2.1 we have discussed, the shock absorber can be
Equation (22) is the relation between the damper force and measured by linear displacement sensor directly or calculated
its velocity. through rockarm rotation angle.
The velocity interval of the shock absorber is mainly
F dam = C · vdam : ð22Þ divided into three parts:

In the above formula, F dam is the force generated by (i) The First Interval (<5 mm/s). The velocity in this
shock absorber, C is the damper coefficient, and vdam is the area is mainly caused by friction in the suspension
velocity of the shock absorber. system
Wireless Communications and Mobile Computing 11

50.00

LF Tre Temp 01 (2) 21

LF Tre Temp 02 (2) 23

LF Tre Temp 03 (2) 32

LF Tre Temp 04 (2) 33

LF Tre Temp 05 (2) 34

LF Tre Temp 06 (2) 34

LF Tre Temp 07 (2) 34

LF Tre Temp 09 (2) 34

LF Tre Temp 09 (2) 35

LF Tre Temp 10 (2) 35

LF Tre Temp 11 (2) 35

LF Tre Temp 12 (2) 35

LF Tre Temp 13 (2) 34

LF Tre Temp 14 (2) 34

LF Tre Temp 15 (2) 31

LF Tre Temp 16 (2) 24

15.00
m:s 108:20 110:00 111:40 113:20 115:00 116:40 118:20 120:00 121:40 123:20 125:00 126:40 128:20

Figure 16: Tire temperature data acquisition graph.

𝛽 absorber, import the data, and then generate the template [7].
Figure 14 shows the damping velocity histogram of the car
α after an endurance test.
A B In order to improve the tire grip performance and main-
tain a certain contact between tire and ground, the damping
adjustment should make the histogram conform to the nor-
K mal distribution as far as possible, and the histogram shape
of the four wheels should be consistent with each other, as
L

R well as the difference between peak values should be as small

α as possible [2].
𝛽
O 4. Tire Temperature Analysis
Figure 17: Ideal Ackermann angle. As we know, tire surface temperature is an external indica-
tion of how well the tire works during testing. By analyzing
the tire surface temperature and considering various factors
of suspension kinematics and dynamics, the ideal working
(ii) The Second Interval (5-25 mm/s) (Low-Speed Damp- temperature range of the tire can be obtained, and the adjust-
ing Interval). This area is the main area where the ment trend of tire camber can be deduced, as well as the real-
shock absorber responds to chassis motion of the time state of tire pressure in the test. By adding a tire temper-
car and is also an important area for driver to per- ature sensor, real-time tire temperature distribution and tem-
ceive the car perature rise data during running can be obtained
(iii) The Third Interval (25-200 mm/s) (High Speed conveniently, and the relationship between tire temperature
Damping Interval). The velocity of shock absorber and damping ratio can be established to guide the racing
in this area is mainly caused by road excitation. car set-up [8].
The impact of road shoulder usually allows the
velocity to exceed 200 mm/s 4.1. Evaluate Wheel Camber from Tire Temperature. As the
car runs, the camber of wheels determines the position in
In order to make the time ratio of shock absorber velocity which the tires are in contact with the ground, and the sur-
in each interval more intuitive, damping velocity histogram, face temperature of tires provides the desired position. When
an important tool for damper analysis and adjustment, is the car corners, for the outside wheels, by observing the color
presented here. The damping velocity histogram is a graph contrast diagram of tire temperature distribution, we can
that describes the time proportion of damping in each veloc- conclude that if the temperature distribution is uniform, then
ity interval [6]. By observing the shape of the histogram, we the camber angle is appropriate [9]. If the outer temperature
can roughly analyze the “soft” and “hard” of the suspension. of the outside wheel is significantly higher than that of the
The establishment method of this histogram template is as inner, it means that the camber angle is too small, otherwise,
follows: create a new workbench in MoTec i2, select the his- the camber angle is too large. For the inside wheel, if the tem-
togram template, and choose the time and proportion in hor- perature in the inner or middle of the wheel rises too fast, it
izontal and vertical coordinates, respectively. Finally, add the means that the camber angle is too large; otherwise, it is too
digital channel of the linear displacement sensor of the shock small. Since the outside wheels provide most of the lateral
12 Wireless Communications and Mobile Computing

1.8 G force lat (G) –1.26

1.6
1.4
1.2
1.0
0.8
0.6
0.4
0.2
0.0
–0.2
–0.4
–0.6
–0.8
–1.0
–1.2
–1.4
–1.6
–1.8

–120 –100 –80 –60 –40 –20 0 20 40 60 80 100 120

Steering wheel angle (°) 99.90

Figure 18: Data acquisition of steering wheel angle and lateral acceleration data scatter diagram.

Table 4: Parameters in formulas (26)–(28). and outer tire temperature will be higher than that of the
middle part, while when the tire pressure is too high, the sit-
Parameter Description uation will be opposite. Therefore, the tire pressure can be
θsw Actual steering angle evaluated by the difference between the middle and the aver-
i Steering ratio age inner and outer tire temperature [14]. This difference is
denoted as ΔTpi .
L Lateral acceleration
Gla Roll rate of the rear antiroll bar
T ti ðinÞ + T ti ðoutÞ
V Vehicle speed ΔTpi = T t i ðmidÞ − : ð25Þ
2

force when cornering, the camber of the outside wheels is of In the above formula, T ti ðmidÞ is the tire middle temper-
more concern. We can characterize the camber state by the ature (° C), T ti ðinÞ is the tire inside temperature (° C), and
temperature difference between the inside and outside of T ti ðoutÞ is the tire outside temperature (°C). When the result
the tire [10]. Equation (24) is the calculation formula of tem- ΔTpi is positive, it indicates that the tire pressure is too high,
perature difference between inside and outside tire surface: vice versa. The following figure shows the tire temperature
data collected in the first 20 seconds of a linear acceleration
ΔT ti = T ti ðinÞ − T ti ðoutÞ: ð24Þ event. Through observation, it can be seen that the tempera-
ture in the middle part of the tire is higher and the tempera-
ture on both sides is lower. It can be preliminarily concluded
In the above formula, i is LF (left front wheel), RF (right that the tire pressure is too high and should be lowered.
front wheel), LR (left rear wheel), RR (right rear wheel), T ti Figure 16 shows the tire temperature data acquisition graph.
ðinÞ is the temperature of the inside of the tire (°C), and T ti
ðoutÞ is the temperature of the outside of the tire (°C). 5. Steering System Data Analysis
However, it is not in line with scientific principles to
make a final judgment on the data of a few specific corners, The design of the steering system is based on Ackerman
so it is necessary to evaluate the data of the whole lap com- steering geometry and takes the slip angle of tire into
prehensively [11]. By equipping multichannel tire tempera- account. In the analysis of steering data, the steering charac-
ture sensors, we can obtain real-time tire temperature teristic of the race car is mainly studied. In addition, the
distribution and temperature rise data during running con- steering design of Formula One should not only reduce tire
veniently. Therefore, the proper camber angle can be deter- wear and prolong the service life of tires but also have good
mined by observing the average tire surface temperature steering stability [7]. The racing car mostly uses the broken
difference throughout the lap [12]. Figure 15 shows the lay- steering trapezoid, the steering torque on the steering wheel
out of tire temperature sensor on our car. from the transmission mechanism to the gear of the steering
machine, and then driven by the gear and rack movement.
4.2. Examine Tire Pressure from Temperature Analysis. Both ends of the steering rod are, respectively, connected
Except evaluating whether the camber angle is appropriate, with the steering rack and the steering knuckle arm. Under
we also use the tire temperature to analyze the tire pressure the action of the pull rod thrust and tension, the steering
when the car is running [13]. Comparing the temperature knuckle arm drives the tire to rotate, so as to achieve steering.
of the inner and outer part of the tire with the middle, we A reasonable steering trapezoid structure should make the
can find that when the tire pressure is too low, the inner four steering wheels roll as far as possible without
Wireless Communications and Mobile Computing 13

110 Streering angle (°) –3.8

100
90
80
70
60
50
40
30
20
10
0
–10
–20
–30
–40
–50
–60
–70
–80
–90
–100

m:s 2:00 2:20 2:40 3:00 3:20 3:40

Figure 19: Actual steering angle data acquisition chart.

100 Yaw rate (°/s) 0.00

90
80
70
60
50
40
30
20
10
0
–10
–20
–30
–40
–50
–60
–70
–80
–90
–100
m:s 2:00 2:20 2:40 3:00 3:20 3:40

Figure 20: Yaw rate data acquisition graph.

sideslipping, that is, the instant center of the four steering are presented in this section to evaluate the steering charac-
wheels’ speed intersects at the same point. When the rear teristic of race cars [4]. Figure 18 shows the data acquisition
wheels of the race car do not steer, the steering center of of steering wheel angle and lateral acceleration data scatter
the race car should be on the rear wheel axle line. This steer- diagram [15].
ing relationship is the ideal Ackermann turning angle rela-
tionship, as shown in Figure 17. For this reason, the 5.1. Understeering Angle. According to the definition of steer-
concepts of understeer angle and attitude angular velocity ing characteristic in Race Car Vehicle Dynamics [5], when
14 Wireless Communications and Mobile Computing

2800
Brake
B k force
f –63.70
663 70
2600
2400
2200
2000
1800
1600
1400
1200
1000
800
600
400
200
0
–200
–400
–600
–800
–1000
–1200
–1400
–1600
–1800
–2000
–2200
–2400
–2600
–2800
–3000
–3200
–3400
–3600
–3800
–4000

m:s 2:00 2:20 2:40 3:00 3:20 3:40

Figure 21: Braking force data collection in race car running test.

6
The rear wheel hydraulic (MPa)

0
0 1 2 3 4 5 6 7

The front wheel hydraulic (MPa)

Figure 22: Actual braking force distribution obtained from data collection.

the actual wheel steering angle is equal to Ackerman steering

angle, the vehicle will have neutral steering. When the steer-
ing angle is larger than Ackerman steering angle, the vehicle
understeers. Oversteer on the other hand. Therefore, the dif-
ference between the absolute value of the actual wheel steer-
ing angle θ and Ackerman steering angle θAck can be defined
as the understeering angle θu . The formula is as follows

θu = jθj‐jθAck j, ð26Þ

θsw
θ= , ð27Þ Figure 23: Race car running test.
i
Wireless Communications and Mobile Computing 15

90 WheelSpeed_FL
F (km/h)
WheelSpeed FL (km/
(km/h
m//h
m h))

60

30

1.2 G Force Long (G)

(G

0.6

0.0

–0.6

–1.2

60 Brake pressuree distribution (%)

(%)

40

20

2:35 2:40 2:45 2:50

Figure 24: Brake data analysis template.

L · Gla In the above formula, ωyaw is the actual yaw rate of the
θAck = : ð28Þ
V2 vehicle (° /s), and ωt is the theoretical yaw rate of the vehicle
(° /s). Figure 20 shows the yaw rate data collected.
Table 4 shows the vehicle parameters required by formu-
Through the analysis, the following conclusions can be
las (26)–(28).
drawn:
According to formula (26), the following conclusions can
be drawn: 8
>
> <0, oversteer,
8 <
>
> <0, oversteer, ωa = 0, neutral steer, ð32Þ
< >
>
:
θu = 0, neutral steer, ð29Þ >0, understeer:
>
>
:
>0, understeer:
6. Braking System Data Analysis
By calculating the understeering angle, we can under- In race car competition, we know that if you spend less time
stand the steering characteristics at various stages of the cor- slowing down, you spend more time accelerating and you can
ner. In order to study the overall steering characteristics of improve your lap speed. As for the braking system, the
the car, the average understeering angle of the car can be important analysis data include vehicle speed, wheel speed,
obtained through the steady-state rotation test, and the over- longitudinal acceleration, front and rear brake pressure, and
all steering characteristic of the car can be evaluated [13]. front brake ratio.
Figure 19 shows the actual steering angles of the car collected
by MoTec i2 analysis. 6.1. Braking Force. The real braking force data can be col-
lected through the brake oil pressure sensor, and the maxi-
5.2. Attitude Velocity Analysis. In addition to the understeer- mum longitudinal deceleration obtained in the running test
ing angle, the vehicle steering characteristics can be charac- corresponds to the maximum braking force of the car. The
terized by the difference between the absolute value of the maximum braking force is one of the evaluation parameters
actual yaw rate and the theoretical yaw rate, that is, the atti- of car braking performance, and usually, the maximum lon-
tude angular velocity [16]. The following is the definition for- gitudinal deceleration of the car should be about 95% of the
mula of attitude angular velocity ωa . maximum lateral acceleration [5]. Figure 21 shows the brak-
  ing force data collected in a race car running test, and
ωa = ωyaw  − jωt j, ð30Þ Figure 22 shows the actual braking force distribution
obtained from data collection. We performed a dynamic test
Glat on the car. Figure 23 shows the picture taken during the
ω= : ð31Þ
V test [17].
16 Wireless Communications and Mobile Computing

80 Wheel speed FL [km/h] 39.6

60
40
20
–0
80 Wheel speed RL [km/h] 39.6

60
40
20
–0
Wheel speed RL [km/h] 40.5
80
60
40
20
–0
80 Wheel speed RR [km/h] 40.8

60
40
20
–0
m:s 2:00 2:20 2:40 3:00 3:20 3:40

Figure 25: Wheel speed data.

WheelRpm
m FR 37

FRAcc 99
–99.0

ABSFlagFR
ABSFl FR
R 2
A S l
ABSFlag1FR
R 1

Figure 26: ABS simulation result.

Wireless Communications and Mobile Computing 17

290 290

X distance (m)

X distance (m)
280
280
270
270
260
260
250
250

Figure 27: Comparison of simulation under open road conditions (the figure on the left does not include ABS braking, and the figure on the
right includes ABS braking).

6.2. Braking Lock. The braking lock can be directly observed or slip, and at the same time, the braking force distribution
through the curve of wheel speed data. When decelerating, of the rear wheels is improved then the speed of cornering
the wheel speed will drop in a straight line, when the tire is is improved. At the beginning of design, we use Simulink
locked, the wheel speed will drop rapidly; when the driver and Carsim cosimulation to establish the corresponding
senses that the tire is locked, he releases the brake slightly, model, to judge its feasibility, and whether it can reduce the
and the wheel speed returns to its original straight line, form- braking distance. Figure 26 shows the ABS/EBD simulation
ing a small trough, as shown in the wheel speed curve in result, and Figure 27 shows the comparison of simulation
Figure 24. The braking lock will cause the tire to lose grip per- under open road conditions [18].
formance, which should be avoided as far as possible.
6.3. Braking Pressure Distribution. Braking pressure distribu-
7. Conclusion
tion is the ratio of the front braking pressure to the total brak- This paper introduces the important mathematical channels
ing pressure, and equation (30) is the math channel formula for FSAE racecar data collection of the five types: reliability,
of it: suspension, tire, steering, and braking, as well as the corre-
sponding calculation formula and the analysis template used
BrakeP F in data analysis. Based on the MoTec i2 data processing plat-
BF = · 100%: ð33Þ
BrakeP F + BrakePR form, the paper carries out data analysis and post-processing
for formula student. In addition, we use CAN bus communi-
In the equation, BrakeP F is the front brake pressure cation to layout and design the vehicle sensor and wiring har-
(MPa), BrakePR is the rear brake pressure (MPa). Figure 25 ness system.
shows the wheel speed data. In the sensor fault diagnosis, for CAN bus signal, we use
Braking pressure distribution affects the maximum brak- interrupt counter counting method to diagnose the loss of
ing force in the pure braking condition and the steering char- signal frame or signal line disconnection and other faults.
acteristics in the cornering condition with braking. Too As for reliability data, we analyze the safety and reliability
much distribution of brake pressure on one axle will result of race cars by collecting battery voltage, battery temperature
in the corresponding tire lock. In the corner with braking, and motor power, so as to troubleshoot potential faults.
too much braking pressure to the front axle will lead to As for suspension system, we mainly analyzed the data
understeer, and too much to the rear axle will lead to over- collected by linear displacement sensor and studied the roll
steer. Therefore, according to the track characteristics and gradient, roll stiffness, tire load transfer, and shock absorber,
driver’s feedback, and combined with data collection, we so as to verify our design and tune the suspension to an
can improve the performance of our car by tuning braking appropriate state.
pressure distribution properly. As for tires, we studied whether the corresponding four-
wheel positioning parameters and tire pressure were appro-
6.4. Set the ABS. We analyze information obtained by data priate and whether the utmost performance of tires could
acquisition, concluding that in the process of race, the brak- be exerted through the data collected from the multichannel
ing intensity is usually small and medium-sized. In the tradi- tire temperature sensor.
tional brake system design, in order to make the front wheel As for the steering system, we obtained steering angle and
lock first and achieve stable condition, the traditional design lateral acceleration scatter diagram and change rate of steer-
always matches a large synchronous adhesion coefficient to ing angle and yaw velocity, to judge the steering characteris-
achieve this goal, so that will make the rear wheels have larger tics of race car and make corresponding adjustments.
attachment loss. The inability to provide greater longitudinal As for the braking system, we analyzed maximum brak-
adhesion of rear wheels results in the lack of lateral adhesion ing force, braking pressure distribution, and checked wheel
of the front wheels, which is very detrimental in the process speed data to deal with brake lock. Furthermore, we do some
of cornering and cannot guarantee the speed and stability research on the ABS/EBD system.
of the car. Therefore, the ABS system is set to ensure that Due to our limited knowledge, the data analysis we have
the front and rear wheels do not lock, lose steering ability, done is relatively basic, but for FSAE teams, it has been an
18 Wireless Communications and Mobile Computing

effective and systematic approach. We will continue to [16] A. Tokosch, D. Hake, K. Meah, and J. Maier, “Design and
improve our data analysis method to make it more standard- implementation of a drivetrain for an FSAE electric vehicle,”
ized in the future. in 2019 IEEE International Conference on Environment and
Electrical Engineering and 2019 IEEE Industrial and Commer-
cial Power Systems Europe (EEEIC/I&CPS Europe), Genova,
Data Availability Italy, 2019.
Data sharing not applicable to this article as no datasets were [17] J. Zhang, X. Lv, and Y. Lv, “Research on vehicle control strat-
generated or analyzed during the current study. egy and hardware in loop for pure electric FSAE vehicle,” in
The 2nd International Seminar on Computer Science and Engi-
neering Technology (SCSET) 2020, Shanghai, China, 2020.
Conflicts of Interest [18] R. Luca, V. Giovanni, B. Alessandro, F. Lorenzo, and
F. Giovanni, “Optimization of the performance of a formula
The authors declared no potential conflicts of interest with
SAE engine by means of a wastegate valve electronically actu-
respect to the research, authorship, and/or publication of this
ated,” Energy Procedia, vol. 101, pp. 654–661, 2016.
article.

References
[1] S. A. E. Formula2020, [Link]
[2] Organizing Committee of Chinese Formula Student, 2019 Chi-
nese Formula Student Racing Manual, China Association of
Automotive Engineers, Beijing, 2019.
[3] G. Zhang, Design and Adjustment of FSAE Racing Suspension,
Tongji University, Shanghai, 2017.
[4] J. Segers, Analysis Techniques for Racecar Data Acquisition
Second Edition, SAE International, 2014.
[5] W. F. Milliken and D. L. Milliken, Race Car Vehicle Dynamics,
Warrendale SAE, 1995.
[6] Y. Peng and Z. Deng, “Design and development of data acqui-
sition system based on FSAE,” Electronic Technology and Soft-
ware Engineering, vol. 3, no. 3, pp. 163–165, 2019.
[7] C. Huang, Y. Zhou, and X. Song, “Design of steering data
acquisition system for FSAE racing car,” Equipment
Manufacturing Technology, vol. 8, pp. 96–99, 2018.
[8] Q. Yue, Q. Wei, and X. Ren, “Research and design of FSAE rac-
ing car data acquisition system,” Modern Electronic Technol-
ogy, vol. 41, no. 5, pp. 69–73, 2018.
[9] Z. Deng, C. Xu, and B. Wang, “Design of FSAE racing suspen-
sion data acquisition system based on LabVIEW,” Journal of
Hefei University of Technology (Natural Science Edition),
vol. 38, no. 8, pp. 1018–1022, 2015.
[10] D. Harsh and B. Shyrokau, “Tire model with temperature
effects for formula SAE vehicle,” Applied Sciences, vol. 9,
no. 24, p. 5328, 2019.
[11] A. Ayyad and M. Fathizadeh, “Telemetry and data collection
to improve formula SAE car,” in World Congress on Engineer-
ing and Computer Science, WCES 2017, San Francisco, CA,
2017.
[12] SAE International, 2020 Formula SAE Rules, 2020, April 2020,
[Link]
[13] G. Xiang, Automotive Engineering II, Mechanical Industry
Press, Beijing, 2013.
[14] E. M. Kasprzak and D. Gentz, “The formula SAE tire test
consortium-tire testing and data handling,” in Motorsports
Engineering Conference and Exhibition SAE International,
Dearborn, Michigan, 2016.
[15] B. Chen, Y. Liu, and W. Shi, “Optimization design and virtual
prototype modeling and simulation analysis of FSAE car steer-
ing trapezium,” in 2018 International Seminar on Computer
Science and Engineering Technology, Shang Hai, China, 2018.