F2000H252

Seoul 2000 FISITA World Automotive Congress

June 12-15, 2000, Seoul, Korea

CAD of Engine Dynamic Test Bed

Based on Hybrid Simulation
Zhang Junzhi

Lu Qingchun

State Key Laboratory of Automobile Safety and Energy Conservation

Tsinghua University
Beijing 100084, CHINA

Tel86-010-62771839
Fax86-010-62784655
[email protected]
Abstract
The hybrid simulation method is adopted to develop engine dynamic test bed based on eddycurrent dynamometer. The hybrid simulation scheme of engine dynamic test bed is designed. The
principle is discussed. Finally, the CAD method is used to design the main parameters of engine
dynamic test-bed based on simulation ECE15 and US LA4-CH Driving Schedules by Shanghai
Santana 2000 car. The results are compared to the actual test results on the chassis dynamometer.
The hybrid simulation method is proved to be an efficient way by simulation and compare.
Keywords: Engine dynamic test bed,

Hybrid simulation,

CAD

. INTRODUCTION
have been made in British [1], Japan and China [2]
etc., with eddy-current dynamometer and engine
coupled directly. But the simulation accuracy and its
actual use are limited because of the unperfect
performance of eddy-current dynamometer.

Engine dynamic test-bed can simulate various

loaded states of engine during its actual operation
with loading equipment. In the beginning of engine
development, the engine performance can be tested
on engine dynamic test-bed comprehensively without
an engine equipped in the vehicle. Therefore, engine
dynamic test-bed is an economical and efficient
method to test, study and develop engine and vehicle.
Engine dynamic test-bed usually uses AC
dynamometer, DC dynamometer and static hydraulic
dynamometer as loading equipment. They have
preferable composite performance, but they are
expensive and complicated. For the sake of
developing low cost engine dynamic test-bed, some
researches have been done and some achievements
Control of water,
oil temperature
and the state of
intake gas

Engine Substitutive
clutch and
transmission
Transient
parameters
measure and gather

Control of
driver
simulator

Funded by Chinese Shanghai Automobile

Industry Group, State Key Laboratory of Automobile
Safety and Energy Conservation in Tsinghua
University has advanced the hybrid simulation
scheme to develop engine dynamic test-bed based on
domestic eddy-current dynamometer. The hybrid
simulation scheme is introduced to compensate for
the unperfect performance of eddy-current
dynamometer in the paper. The principle of hybrid
simulation is discussed in detail. The control system
is designed based on the simulation of
Control of
Santana 2000GLi car under the ECE
dynamomete
15 and EPA LA4-CH emission test
r
modes. The results are compared with
that of actual tests on chassis
dynamometer.

EddyTorque current
meter dynamometer

Assistant
transmissio

Inertial
flywheels

Emission
analyzing
system

Fig.1. Hybrid simulation scheme of engine dynamic test bed

1

. HYBRID SIMULATION SCHEME OF

ENGINE DYNAMIC TEST BED

To the driving end of the clutch,

Tc = Ted I f

The hybrid simulation scheme of engine dynamic

test bed is illustrated in Fig.1.[3] It consists of the
hybrid simulations of driver, vehicle transmission
system and vehicle resistance.

F f rolling
resistance Fi uphill resistance Fw
aerodynamic drag F braking force (supposing
the brake is not locked during braking.) r wheel
radius i g gear speed of transmission i o gear
speed of differential T1 transmitting efficiency
of transmission T2 transmitting efficiency from
transmission output shaft to wheel I EC
equivalent rotating inertia at the input shaft of
transmission v a velocity

Ted = I f

+ (I EC + I f )

de
dt

(4)

z When clutch slides,

Ted = Tc max I f

d e
dt

(5)

In the test bed,

TD + TDf ( I D + I S + I j ) d c
Tc =
+
2

dt

i g T1
i g T1

(1)

i g io T1T2

(3)

Ted = ( Ff + Fi + Fw + F )r /(ig io T1 T2 )

Tc = (Ff + Fi + Fw + F )r /(igioT1 T2 )

d e
dt

z When transmission is in gear and clutch does

not slidethen

In course of the actual running of vehicle,

supposing Tc is the actual transmitting torque of
clutch, then at the driven end of clutch,

In which I EC

rotating

z When clutch is disengaged and transmission is

in neutral Tc = 0, then

The acceleration resistance is mainly simulated by

flywheel. Because of the variety of vehicles, the
simulated acceleration resistance by flywheel is not
the same as the actual value strictly. On the other
hand, the difference between the actual gear speed
and that of substitute transmission results in the
simulation difference of acceleration resistance.
Finally, the inertia of the substitute clutch and other
rotating parts of the test bed is different from that of
an actual vehicle. All the difference needs to be
compensated by the appropriate control of eddy
current dynamometer. Finally, the acceleration
resistance can be simulated precisely.

mr 2 + I w

I f inertia of engine flywheel e engine speed.

During the simulation of driver, powertrain and

running resistance, the precise simulation of
acceleration resistance is the key factor.

r dt

inertial of wheel Ted engine transient torque

. PRINCIPLE OF THE HYBRID

SIMULATION

+ IEC

(2)

In the above two formula,

The hybrid simulation of driver means that the

operating objects such as engine, clutch and
transmission are in reality, and the driving acts and
levels are simulated by the control system. The
hybrid simulation of transmission system means that
the actual clutch and transmission are simulated by
substitutive clutch and transmission, and the
difference between the actuality and simulation are
compensated by the control system. The hybrid
simulation of vehicle running resistance means that
the accelerating resistance is mainly simulated by the
flywheel, and the difference between the actual and
simulated accelerating resistance and the roll
resistance, air resistance and uphill resistance are
simulated by the proper control of loading equipment.

igio dva

d e
dt

Ted = Tc + I f

de
dt

(6)

(7)

'

In which i g gear speed of substitute

transmission

'

T1

I ' j and I , the difference between

and I j , is compensated by dynamometer.

flywheel

I'j

transmitting efficiency of
'

substitute transmission T c transmitting torque

of substitute clutch TD dynamometer braking

. CAD OF THE TEST BED CONTROL

SYSTEM

torque TDf dynamometer friction torque I D

4.1 Principle of optimization

dynamometer rotating inertia I s inertia of the

In the control system flywheels inertiaI,

control step lengthT and predictive time period
N are several important parameters. Theoretically
speaking, the inertia of flywheels should be the same
as that of the actual vehicle to reduce the loading
burden of the eddy-current dynamometer. Actually,
the simulation of speed has error. If the inertia is big,
the loading burden of the eddy-current dynamometer
should be increased to compensate the error.
Therefore, the inertia of the flywheel should be
selected reasonably to increase the simulation
accuracy maximumly.
Popularly, the shorter the control step is, the
higher the control accuracy is. But, if the step is too
short, the more computer resource should be
occupied. The control step should cooperate with
other parts of the system. If the response time of the
actuator is too long, the reduction of the control step
has small result. Therefore, the control step should be
selected compensately.
The selection of predictive time length affects
the control accuracy deeply when the dynamic matrix
predictive method is adopted. If N is big, the
robusticity is good, but the dynamic responsibility is
bed. If N is small, the dynamic tracking performance
is good, but the robusticity is bed. With the increase
of N, the loading accuracy increases stepwisely. If N
is too big, the real time performance can not satisfy
the requirement.

shaft and coupling device I j flywheels inertia

I ' f inertia of which is fixed coupled to the

engine crank shaft in the test bed c speed of
transmission input shaft.
When transmission is in gear and clutch does not
slide e = c put (6) into (7), then

Ted =

(I + I + I j )
TD + TDf
d
+[ D 2S
+I f ] e

dt

i g T1
i g T
1

(8)
The simulation purpose of engine dynamic test
bed is to make Ted in test bed be the same as that of
actual operation. If the inertial force is simulated by
flywheelthe corresponding coefficient should be
equal in (4) and (8), then

Ij =

(mr2 + I w )i

ig io T1 T2
2

T1

+i

2
g

T1 (I f I

(I D + I s )

(9)

TD = (Ff + Fi + Fw + F )ri g T1 /(ig ioT1T2 )
TDf

4.2 Evaluate criterion of optimization

Based on the characteristics of developed test
bed, two sorts of evaluate criteria are used. One is
basic, which includes vehicle speed tolerance and
that of engine torque. The other is the coefficient of
determination, which is used to evaluate the whole
simulation accuracy.

(10)

The inertia of flywheel should be selected

according to (9), and its value can be the average of
that of every gear in actual use.
If there is difference between
supposing

I ' j and I j

4.3 Optimization of the coefficients

The inertia of the flywheels can be choosed as
the following values: average of theory simulated
inertia at every gear speed, half of the average, one
quarter of the average, null (electricity simulation
entirely). The control step can be 1000ms, 500ms,
250ms, 100ms, 50ms. The predictive time can be
2000ms, 1000ms, 500ms and 0(normal feed back
control). The parameters of the control system are
optimized based on the simulation of ECE 15 mode
and US LA4-CH mode(150s300s) with Shanghai
Santana 2000 car.

I j = I ' j + I then

TD = (Ff + Fi + Fw )rig T1 /(igioT1 T2 ) TDf

+

I de
ig dt

(11)

The simulation of acceleration resistance can be

divided into two parts. The main part is simulated by
3

vehicle speed. Figure 4 illustrates the profiles of ideal

and simulated engine torque. Figure 5 is the error of
simulated engine torque. Figure 6 is the compare
between the profiles of simulated (real line) vehicle
speed and measured(dotted line) on the chassis
dynamometer.

According to the above optimization criteria,

the optimized parameters are: simulated inertia(half),
control step(100ms), predictive time(0) when ECE
15 mode is simulated. Figure 2 is the profile of
simulated vehicle speed, in which dotted lines are
error bounds. Figure 3 is the simulated error of
Va(km/h)

60
40
20
0
-20

t(s)

50

100

150

200

Figure 2 Simulated vehicle speed

Va(km/h)

0.5
0
-0.5
t(s)

-1

50

100

150

200

Figure 3 Simulated vehicle speed error

Te(Nm)

60
40
20
t(s)

50

100

150

200

Figure 4 Ideal and simulated engine torque

Te(Nm)

100
50
0
-50
-100

t(s)

50

100

150

Figure 5 Simulated error of engine torque

4

200

Va(km/h)

60
40
20
t(s)

50

100

150

200

Figure 6 Simulated and tested vehicle speed

When US LA4-CH mode is simulated, the

optimized parameters are: simulated inertia(half),
control step(500ms), predictive time(500ms). Figure
7 is the profile of simulated vehicle speed, in which
dotted lines are error bounds. Figure 8 is the
simulated error of vehicle speed. Figure 9 illustrates
Va(km/h)

the profiles of ideal and simulated engine torque.

Figure 10 is the error of simulated engine torque.
Figure 11 is the compare between the profiles of
simulated (real line) vehicle speed and
measured(dotted line) on the chassis dynamometer.
Te(Nm)

200

100

150
50

100
50

0
150

t(s)

200

250

300

350

Figure 7 Simulated vehicle speed

0
150

t(s)

200

250

300

350

Figure 9 Ideal and simulated engine torque

Va(km/h)

Te(Nm)

200

100

0
-2
150

-100
t(s)

200

250

300

350

Figure 8 Simulated vehicle speed error

t(s)

-200
150

200

250

300

350

Figure 10 Simulated error of engine torque

Va(km/h)

100

50

0
150

t(s)

200

250

300

Figure11 Simulated and tested vehicle speed

350

The hybrid imulation of driver ensures the

automation of test- bed and improves the

. CONCLUSION

repeatability of test. This EDTB can simulate driver,

transmission system and running resistance easily,
and it is an economical and efficient method to test,
study and develop engine and vehicle. All the work is
all-important to improve the technical level of engine
test and to enhance the designing and developing
ability of engine and vehicle.

Using the foreign advanced engine dynamic test

bed as reference, the hybrid simulation method is
adopted to develop national engine dynamic test bed
based on domestic eddy-current dynamometer. The
hybrid simulation scheme of engine dynamic test bed
is introduced. The simulation principle is discussed.
Finally, based on the simulation of ECE15 and US
EPA FTP75 Driving Schedules with Shanghai
Santana 2000 car, the control system is designed .The
results are compared to the actual test results on the
chassis dynamometer.

. REFERENCES
[1]
J.Koustas and N.Waston,"A Transient Diesel
Test Bed with Direct Digital Control,"SAE
40347,pp.201-215.
[2] C.C.Qui and F.B.Yuan,"Microcomputer
ontrolled dynamic engine test bed,"c24/87,ImechE
1987,pp.325-332.
[3] Zhang Junzhi and Lu Qingchun, "Study on the
Engine Dynamic Test Bed", Proceedings of the IEEE
International Vehicle Electronics
Conference,IVEC99, pp. 406-409.

It can be seen from the theoretical analysis and

test results that the limitation of eddy-current
dynamometer, including absence of driving ability
and bad response performance to the control
instruction, is compensated efficiently with low
cost and easily. This ensures the simulation
accuracy and enlarges the application range. The
F.F. vehicle can be simulated by the F.R. method