Modelling Vehicle Performance (MVP)

Coursework-Straight line performance

testing in MIRA

Department of Aeronautical and Automotive Engineering,

Loughborough University, Loughborough, Leicestershire LE11 3TU,
UK

Student ID: B835763

 Executive summary
Powertrain performance is an essential factor affecting vehicle dynamics, and the dynamic model
is a significant basis for the simulation of vehicle performance. However, the model needs to be
parameterised and verified according to the actual test results. In this paper, a Simulink powertrain
simulation model is used to predict the straight-line performance of a specific vehicle. The actual
measured data were obtained based on the measurements of the coast-down test and acceleration
test on the real track on MIRA. And the critical parameters of the optimised vehicle model are
obtained by fitting the simulated time velocity curves and the measured time velocity curves.
Finally, the powertrain model simulation results and vehicle test results prove that the powertrain
model design becomes accurate, and the simulation calculation is more effective and reliable after
tuning the critical parameters of the test vehicle. In the first section, the test and measurement in
MIRA will be introduced. In the second section, the vehicle simulation model will be introduced.
The optimisation will be discussed in the third section. Finally, in the fourth section, the results
of the optimisation will be discussed and studied.

1
List of Contents
1 Test and measurement ........................................................................................................................3
1.1 Test procedure ...............................................................................................................................3
1.1.1 Coast-down test ...................................................................................................................3
1.1.2 Acceleration test ...................................................................................................................3
1.1.3 Data collection.......................................................................................................................3
1.2 Analysis and discussion ...............................................................................................................4
1.2.1 The process of using data in MATLAB/SIMULINK .....................................................4
1.2.2 Analysis for measured data ...............................................................................................4
2 Vehicle model .........................................................................................................................................5
2.1 Development of the model........................................................................................................5
2.2 Model parameterization .............................................................................................................6
3 Optimisation ............................................................................................................................................8
3.1 Procedure to tune the model parameters ............................................................................8
3.2 The reason of tuning Ad, Bd and Cd ..........................................................................................8
4 Results, analysis and discussion ........................................................................................................9
4.1 Analysis for first optimization ....................................................................................................9
4.2 Analysis for the improved optimisation .............................................................................. 10
4.3 Analysis for the final results .................................................................................................... 11
4.4 Discussion with optimisation method ................................................................................. 12
5 Conclusion ............................................................................................................................................ 12
6 Reference .............................................................................................................................................. 13
7 Appendix ............................................................................................................................................... 14

2
1 Test and measurement
The straight-line performance test was divided into the coast-down test and acceleration test. The
test vehicle chosen for the test was a Ford Focus ST. The coast-down and acceleration tests were
carried out on two parallel straight tracks. And the straight track is one mile long and connected
with two inclined turns. The slope of the two straight tracks is zero.

1.1 Test procedure

1.1.1 Coast-down test

The car was continued to acceleration as it bent. Once it reached the straight section, the speed of
the car was accelerated quickly to 70mph. Meanwhile, the gearbox was placed in neutral, the car
began to coast, and the data was started to collect immediately. The above process was repeated
in the opposite direction on the other straight track. The process of reducing the speed from 70mph
to 40mph and 40mph to 0 was recorded separately.

1.1.2 Acceleration test

The test car was parked at the start of the straight track, and the transmission was placed in the
first gear, then the car was started, and the accelerator pedal was quickly stepped on by the driver.
Then, the gear is continuously lifted until the speed reached 80mph.

1.1.3 Data collection

CANBUS and GPS data was used to give the data of the test vehicle during the tests. The testers
determine the time to start recording data in the car. The measured data consists of information,
including vehicle speed, engine speed, test time. These data would be viewed on the analysis
laptop. After the test on tracks, the weight of the test car was measured. The air density on the test
field during the tests could be collected on the internet. Besides, the direction of the wind, which
included tailwind and headwind was also recorded by testers in the test vehicle, the test records
from the testers are shown below.

1) Coast-down test, the range of vehicle speed is from 70 mph to 40 mph, head wind, the road
surface of last 500 meters is rough;

2) Coast-down test, the range of vehicle speed is from 70 mph to 40 mph, tail wind, the road
surface is smooth;

3) Coast-down test, the range of vehicle speed is from 40 mph to 0 mph, head wind, the road
surface of last 500 meters is rough;

3
4) Coast-down test, the range of vehicle speed is from 40 mph to 0 mph, tail wind, the road
surface is smooth;

5) Acceleration test, upshift from first gear to fifth gear, the range of vehicle speed is from 0
mph to 80 mph

6) Acceleration test, upshift from first gear to fifth gear, the range of vehicle speed is from 0
mph to 60 mph

7) The vehicle weight (including passengers and driver) is 1860 kg, the front axle weight is
1080 kg and the rear axle weight is 780 kg.

1.2 Analysis and discussion

1.2.1 The process of using data in MATLAB/SIMULINK

The usage of the measured data is shown below:

1) Load the measurement data files for different tests in MATLAB;

2) Change the simulation time, vehicle mass, initial vehicle speed, initial engine speed and air
density to the actual values;

3) Run the vehicle model in Simulink;

4) Tune some key parameters to get the best match between simulated vehicle speed and
measured vehicle speed.

1.2.2 Analysis for measured data

There are only some sorts of the measured data are needed to analyse form all test results, which
are shown in Figure 1. Accelerator pedal position, clutch position, measured vehicle speed,
measured engine speed, gear number and test time are employed as inputs to the vehicle
simulation model in section 2. Moreover, because the driver cannot get a perfect initial speed of
70mph or 40mph on each test, redundant test data of test time and vehicle speed needs to be
removed.

4
 Figure 1 Measured data
By importing the values of these parameters shown in Figure 1 to the vehicle simulation model,
a simulated speed is obtained as the output signal from the model. Then, after running the vehicle
model, a figure consists of a time-simulated vehicle speed curve, and a time-measured vehicle
speed curve is generated. For fitting these two curves, some parameters in the vehicle model
should be optimised. In this paper, tyre rolling resistance coefficient (Ad, Bd) and drag coefficient
(Cd) is considered to tune to get the best match for the coast-down test. Then, these tuned
parameters are used to give the simulated and measured vehicle speed curves for the acceleration
test. Finally, the vehicle simulation model becomes more accurate and closer to the real vehicle.

2 Vehicle model
In this section, the development of the vehicle simulation model and the accuracy of the initial
parameters of the vehicle model are discussed.

2.1 Development of the model

The Simulink model of the test vehicle is shown in Figure 2. There are four inputs in the model,
which are throttle position, clutch position, engine speed and gear number. And the output of the
model is measured vehicle speed. The model consists of five systems, which are engine system,
clutch system, transmission system, final drive and driveshaft system, wheel and tyres system and
vehicle chassis system.

5
 Figure 2 Simulink model of Ford Focus ST
As shown in equation 1, the throttle position is the input signal of the engine system. Then, it is
transferred to torque to other systems. The clutch position is the input signal of the clutch system.
Gear number is the input signal of the transmission system, then the gear ratio is found. The input
signal of the final drive and driveshaft system is the torque from the transmission, which is
transferred to output torque to wheel and tyres system.

𝑇𝑤ℎ𝑒𝑒𝑙 = 𝑇𝑒𝑛𝑔𝑖𝑛𝑒 × 𝐺𝑐𝑙𝑢𝑡𝑐ℎ × 𝐺𝑔𝑒𝑎𝑟 × 𝐺𝑓𝑖𝑛𝑎𝑙 𝑑𝑟𝑖𝑣𝑒 × 𝜂 (1)

As shown in equation 2, the force of wheel is calculated in wheel and tyres system. Then it is the
input signal of the vehicle chassis system. In vehicle chassis system, the simulated vehicle speed
is solved.

𝑇𝑤ℎ𝑒𝑒𝑙
𝐹𝑡𝑟𝑎𝑐𝑡𝑖𝑣𝑒 = (2)
𝑅𝑟

2.2 Model parameterization

After importing the measured data to the vehicle model, some parameters in the vehicle model
should be corrected. The parameters are shown below:

1) Initial vehicle speed;

2) Initial engine speed;

3) The weight of the test vehicle;

4) Air density, the data which is needed for calculation are shown in Table 1;

5) Drag coefficient (Cd)

6) tyre rolling resistance coefficient (Ad, Bd)

6
In the vehicle model, the initial vehicle speed should be shifted to the measured initial vehicle
speed. The initial engine speed should be changed as well. The weight of the test vehicle is 1860kg.
The calculated air density is 1.2244kg/m3. The drag coefficient Cd is not constant but affected by
wind speed, wind direction and size of the vehicle (Clancy, 1975). The tyre rolling resistance
coefficient Ad and Bd and drag coefficient Cd are tuned in section three.

Table 1 weather condition of the tests (Skyview, 2019)

Air pressure(mmHg) Air temperature (℃) Relative humidity (%) Air density(kg/m3)

756.8 13.1 71 1.2244

The initial Ad, Bd and Cd are 0.019, 0.00004 and 0.3. The Figure 3 and Figure 4 show the output
signals of the model after running the model for four coast-down tests. The gap between the
simulated vehicle speed and measured speed are huge. Therefore, the model should be optimised.

Figure 3 The output signal under initial parameters for the first coast-down test and third coast-
down test

Figure 4 The output signal under initial parameters for the second coast-down test and fourth
coast-down test

7
3 Optimisation
In this section, the procedure of optimising the vehicle simulation model and the reason for tuning
Ad, Bd and Cd will be discussed.

3.1 Procedure to tune the model parameters

By using the parameter estimation function in Simulink, parameters of the vehicle model can be
tuned. And the steps of the approach are shown below.

1) Select the simulated vehicle speed for the Coast-down test;

2) Select the parameters, which are type rolling resistance (Ad, Bd) and drag coefficient (Cd);
and the reason for tuning these parameters is discuss in section 3.1

3) Define the range of parameters;

4) Plot the measured experiment data and compare with simulated data;

5) Display the parameter values as they change during estimation, the result of the
estimation for the first coast-down test is shown in Figure 5;

6) Updata the model with the estimated values and compare the output signal of the model
for other tests.

Figure 5 the result of estimation in Simulink

3.2 The reason of tuning Ad, Bd and Cd

As shown in equation 3, because the coefficient Ad and Bd are variables, tyre rolling resistance
is difficult to calculate accurately. Moreover, there are losses caused by the wind for a free-rolling
tyre, and it is difficult to measure in wind tunnel tests (Passmore & Le Good, 1994).

8
 𝐹𝑅 = 𝑀𝑔(𝐴𝑑 + 𝐵𝑑 𝑣) (3)

Therefore, Ad and Bd should be obtained by the optimising the vehicle simulation model. Drag
coefficient in this paper is also tuned because the flow direction is changed during the tests.

4 Results, analysis and discussion

As shown in Figure 5, after tuning, the values of Ad, Bd and Cd is changed to 0.0134, 0.000517
and 0.282. Then, the model is updated with these new values and used to run data of other coast-
down tests. The results are shown in Figure 6 and Figure 7.

Figure 6 The output signal under optimised parameters for the first coast-down test and third
coast-down test

Figure 7 The output signal under optimised parameters for the second coast-down test and
fourth coast-down test

4.1 Analysis for first optimization

9
For the second and fourth coast-down tests, because of the effect of a tailwind, the simulated
speed drops faster than the measured speed. So, the first finding is that the parameters Ad, Bd and
Cd for the headwind groups are different from those for tailwind groups. Therefore, the second
coast-down test is used to tune the Ad, Bd and Cd. The result is shown in Figure 8. For this new
optimization, the vehicle speed curves for the second and fourth test are expected to match..

Figure 8 tuning Ad, Bd, Cd in Simulink for the second coast-down test

4.2 Analysis for the improved optimisation

The values of Ad, Bd and Cd are changed to 0.0125, 0.000199 and 0.3003. Then the model has
operated again; the output signals of four different coast-down tests are shown in Figure 9, Figure
10.

Figure 9 The output signal under second optimised parameters for the first coast-down test and
third coast-down test

10
Figure 10 The output signal under second optimised parameters for the second coast-down test
and fourth coast-down test

4.3 Analysis for the final results

As seen in Figure 9 and Figure 10, the simulated and measured vehicle speed curves for the second
test are matched, but the curves for the fourth test do not match well. One of the reasons for that
is the effect of road condition. The last 500 meters of the track for the fourth coast-down test is
rough, so they do not match well. Now, there are two groups of the optimisation results of
optimising Ad, Bd and Cd, which are shown below:

1) Ad=0.0134, Bd=0.000517 and Cd=0.281, first test, head wind, last 500 meters of the test track
is rough;

2) Ad=0.0124, Bd=0.000199 and Cd=0.3002, second test, tail wind, whole test track is smooth;

Therefore, the first group of Ad, Bd and Cd is used to update the vehicle simulation model for
predicting the vehicle performance on the rough road surface and headwind conditions. The
second group of Ad, Bd and Cd is suitable to simulate the vehicle performance on the smooth
road surface and tailwind conditions.

11
4.4 Discussion with optimisation method
Using the estimation of the parameters in Simulink is a very convenient and reliable way to tune
the critical parameters of the vehicle model and discuss the results. In this paper, Ad, Bd -and Cd
are tuned based on the first coast-down test and second coast-down test, because the range of
vehicle speed of these two tests is the same. By comparing the parameters after tuning, the effect
of wind direction and road condition can be discussed. However, this optimisation only considers
the effect of road condition and wind direction, the effect of other weather conditions is not
considered because of the lack of measured data, like yaw angle and wind speed. In order to
validate the model, these two groups of Ad, Bd and Cd are imported to model to plot the results
of the acceleration test, and the figures of acceleration test are shown in the appendix.

5 Conclusion
A Simulink vehicle model is used to predict the straight-line performance of a specific vehicle.
Based on the measured data obtained in the MIRA, tyres rolling resistance Ad, Bd and drag
coefficient Cd of the vehicle model are tuned two times for different direction of the wind. So,
there are two findings in this paper. Wind direction will affect Ad and Bd of the vehicle, and then
the roading condition will also affect the Ad and Bd. For different conditions, different groups of
Ad, Bd and Cd should be used to give a more accurate prediction of the straight-line performance
of the test vehicle. Finally, the vehicle simulation model is proved useful for predicting the
straight-line performance of the test vehicle.

12
6 Reference
Clancy, L. J., 1975. Aerodynamics. s.l.:Halsted Press.

Passmore, M. A. & Le Good, G. M., 1994. A detailed drag study using the coastdown method
(No. 940420). SAE Technical Paper.

Skyview, 2019. Skyview Weather. [Online] Available at: https://www.skyview.co.uk [Accessed

18 Nov 2019].

13
7 Appendix

Figure 11 first acceleration test, Ad=0.0134, Bd=0.000517 and Cd=0.281

Figure 12 first acceleration test, Ad=0.0124, Bd=0.000199 and Cd=0.3002

14