Abstract: This paper describes modelling of the transient dynamics of an automatic transmission
during gear changes. A brief introduction to the automatic transmission system and the dynamic
characteristics of the transmission components during the gear changes are presented. Then, detailed
mathematical models of a four-speed automatic transmission manufactured by BTR Automotive,
Australia, are developed. A mode description method is used to describe the transient shifting process
and a modular structure of the transmission system, which consists of a torque converter module,
geartrain module, hydraulic system module and modules of clutches and bands, is presented. As an
application, the developed simulation system is applied to investigate the transient performance of
the automatic transmission during the 12 shift process. The output torque pro les predicted by the
model simulation correlate very well with the experimental data measured from vehicle tests.
NOTATION r radius
R drum radius of the band
A(t) ow area Rd inner radius of the clutch friction plates
c ori ce owrate coeYcient Ri outer radius of the clutch friction plates
o
C radial clearance of the valve t time
dR diameter of the valves T torque
F force V(t) oil volume as a function of time
H Heaviside step function X stroke of the valve/piston
i speed ratio b band wrap angle
bB( p)
I mass moment of inertia
bulk modulus as a function of pressure
k ,k clutch/band sti Vness
0 1 b bulk modulus at p=p
c0 0
l ow length between ends 1 and 2
air ratio at pressure p
ca
L length vectors
i air ratio at p=p
e ,0 k 0
M modes of operation
n i
oil property constants
number of contact surfaces in the multidisc h angular velocity
clutch pack h angular acceleration
p pressure q dynamic viscosity of the oil
p pressure rate m static or dynamic friction coeYcient of the
p atmospheric pressure C
0 clutch
p p pressure drop across the ends 1 and 2
Q1 2
owrate
m
B
static or dynamic friction coeYcient of the
band
Q generalized coordinates
i
Subscripts
B band
The MS was received on 8 October 2001 and was accepted after revision B1 band B1
for publication on 26 June 2002.
* Corresponding author: Faculty of Engineering, University of
C clutch
Technology, Sydney, PO Box 123, Broadway, NSW 2007, Australia. C2 clutch 2
I07201 IMechE 2002 Proc Instn Mech Engrs Vol 216 Part I: J Systems and Control Engineering
332 N ZHANG, D K LIU, J M JEYAKUMARAN AND L VILLANUEVA
ematical description of the hydraulic elements is required The owrate passing through the valves and ori ces
in order to execute the simulation. As an example, the can be de ned in terms of Reynolds number; i.e. the
ow characteristic (pressure drop/ ow resistance) of the standard ori ce ow equations of turbulent ow are
hydraulic circuit at low temperature is dominated by
(t) =cA(t) p (t ) p (t)
Q (13)
the viscosity eVect rather than the turbulent ori ce ow; 1 2
thus the pressure drop due to viscosity eVects should be and viscous ow conditions for laminar ow are
incorporated at low operating temperatures. Math-
p p (t) dc3 3 e 2
ematical representation of each hydraulic element can (t) = 1 (t ) 2 R 1+
C A BD
Q (14)
be formulated from rst principles. The pressure changes l 12q 2 C
R
in terms of various ow quantities are described by the
The stiVness characteristics of the clutch can be divided
diVerential equations for uid dynamics, and the ow-
into two regions; one is a linear region where the stiVness
rate passing through valves and ori ces is described by
is governed by the clutch return spring and the second
laminar and turbulent ori ce ow equations.
region is dominated by the compressibility of the facing
In particular, Newtons second law is used to derive
material of the clutch. Based on the measured exper-
the equation of motion of the regulator and control
imental data, the clutch non-liinear characteristics can
valves, which is primarily dictated by pressure diVer-
be written as
ences, spring forces and jet forces. Fluid dynamics
models are used to de ne the pressure changes in the F (t)=k X +H(X X )k (X X )2 (15)
C 0 0 1 0
hydraulic system. As an example, sudden pressure
As an example, simulation results of the hydraulic
changes in a valve signi cantly aVect the dynamics of
system are compared with measured pressure pro les in
the valve. The di Verential equations that describe the
Fig. 4. In particular, Fig. 4 compares the band pressure
pressure changes in terms of various system ow
resulting from a typical current pro le of the trans-
quantities can be written in the following generalized
mission control unit ( TCU ). The pressure pro les show
form:
very good correlation, with the exception of the period
(t)
Q between 4.8 and 5.3 s. It is noted that the inertia phase
p (t)= b ( p) (12) begins at t =4.8 s and the system enters the non-linear
V(t )
transient gear state.
where the bulk modulus can be de ned as a function of
pressure, i.e.
b pk 4.4 Clutch and band model
b ( p)= 0
( 1 c )pk+ c b Frictional torques of clutches and bands are obtained
a a 0
and from empirical formulae [9]. The torque transmitted by
clutches is proportional to the axial force across the
c ( p /p)0.7 clutch plates and the coeYcient of friction on the contact
c = 0 0
a 1 c + c ( p /p)0.7 surfaces. The torque transmitted through the multidisc
0 0 0
the rst gear ratio (i =2.39) during the torque phase torque pro le with the torque hole (rapid decrease in
FS1
between t # 0.25 s and t # 0.49 s. The speed ratio then output torque) during the torque phase and gradual
continuously decreases to the second gear ratio (i = torque increase during the inertia phase. To examine the
FS2
1.45) during the inertia phase between t # 0.49 s and sensitivity of the time width and torque hole, the 12
t # 1.47 s. Consequently, the engine speed also decreases shift simulation is performed with various combinations
during the inertia phase. Figure 9 shows that the corre- of band pressures and frictional coeYcients.
sponding output speed pro le of the ring gear increases Variations in band forces are common in high-volume
slightly during the 12 shift. Figure 10 shows the output production transmissions because of the dimensional
I07201 IMechE 2002 Proc Instn Mech Engrs Vol 216 Part I: J Systems and Control Engineering
338 N ZHANG, D K LIU, J M JEYAKUMARAN AND L VILLANUEVA
tolerances associated with manufacturing transmission begins at t=0.25 s and ends at t =0.55 s, which is longer
components, in particular, high-precision hydraulic than the shift event presented in Fig. 10 (i.e. t=0.25 s
elements. In addition, variations in oil properties such and t =0.49 s respectively). Similarly, the inertia phase
as frictional characteristics, oil viscosity, oil ageing, etc., begins at t =0.55 s and ends at 1.525 s. The inertia phase
alter the performance of the hydraulic system and con- duration of 0.975 s is similar to the one shown in Fig. 10
tribute to variations in the band supply pressure. To (duration of 0.98 s). Results con rm that a small change
simulate the variability of a typical production trans- in the dynamic friction coe Ycient has a signi cant eVect
mission, simulations are performed using the four on the time width and torque hole in the torque phase
slightly diVerent band force pro les shown in Fig. 11. of the shift process. The static and dynamic friction
These band force pro les are obtained by changing the coeYcients vary during the service of automatic trans-
regulator valve (an important hydraulic system compo- missions as a result of changes in oil temperature, oil
nent) dimensions within the production tolerances. The viscosity, and components and oil ageing.
output shaft torque pro les corresponding to the four The developed simulation model was then rearranged
input forces are compared in Fig. 12. Cases 1 and 3 show to investigate what B1 band pressure and torque pro les
a sharp decrease in output torque at the end of the torque were required to achieve a desired output torque pro le.
phase and case 2 shows the longest duration of the shift In the rearranged model, the desired shaft torque is used
and sharp changes in output torque at the end of the as one of the input parameters and the corresponding
inertia phase. Alternatively, case 4 shows the best torque B1 torque is calculated as part of the simulation. As an
pro le; i.e. it exhibits the minimum torque hole and example, a desired output torque pro le, shown in
output torque disturbance. Results presented in Fig. 12 Fig. 14, which is smoother than the torque pro le pre-
show that variations in the band force have a signi cant sented in Fig. 10, is used as an input to the simulation
eVect on the output shaft torque variations. model and then the 12 shift simulation is executed. The
To investigate the eVect of the dynamic friction resulting B1 band torque required to produce the desired
coeYcient of the band material, the 12 shift simulation output torque pro le is obtained and shown in Fig. 15.
is performed with a 2 per cent decrease in the friction Here again the results show that a smooth B1 band
coeYcient. The results corresponding to the output torque pro le is required to achieve smooth output
torque pro le are presented in Fig. 13. The torque phase torque variations during the 12 shift. This model can
Fig. 12 Output torque pro les corresponding to the band forces shown in Fig. 10
I07201 IMechE 2002 Proc Instn Mech Engrs Vol 216 Part I: J Systems and Control Engineering
340 N ZHANG, D K LIU, J M JEYAKUMARAN AND L VILLANUEVA
Fig. 13 Output torque pro le with a 2 per cent decrease in the friction coeYcient of Band B1
Fig. 15 Band torque required for the desirable output torque pro le shown in Fig. 14
Proc Instn Mech Engrs Vol 216 Part I: J Systems and Control Engineering I07201 IMechE 2002
MODELLING OF DYNAMIC CHARACTERISTICS OF AN AUTOMATIC TRANSMISSION 341
I07201 IMechE 2002 Proc Instn Mech Engrs Vol 216 Part I: J Systems and Control Engineering