Sunteți pe pagina 1din 6

ELSEVIER JSAE Review 15 ( 1994~ 31)9-314

Improvement of prediction accuracy for torque converter performance:


One-dimensional flow theory reflecting the stator blade geometry
Masaaki Kubo, Eiji Ejiri, Haruo Kumada, Yoshihiko Ishii
~is,~an Research Center, Nissan Motor Co.. Ltd., 1 Natsushima-cho, Yokosuka-shi, Kanax~awa, 237.1apart

Received 15 February 1994

Abstract

Specific fuel consumption is a key word these days, and the torque converter can play an important role in impro',ing it. This makes it
essential to improve the accuracy of methods for predicting torque converter performance. In this work, the accuracy of performance
predictions was improved by replacing the flow angle and total pressure loss in the angular momentum theory with corresponding values
obtained by analyzing the flow around the stator using a general flow analysis code. A quick and highly accurate prediction of torque
converter performance is made possible with this method.

1. Introduction mensional flow lines, and torque converter performance is


predicted using equations of angular momentum and con-
The percentage of passenger cars equipped with an servation of energy. These equations are solved using
automatic transmission has exceeded 70% in Japan in empirical values for the friction loss coefficient and outlet
recent years, and there have been increasingly stronger flow angle of each element. One disadvantage of this
demands for automatic transmissions that are smaller and method is that it cannot fully predict the effects on torque
lighter and deliver higher performance. In this connection, converter performance of minute modifications made to
the performance requirements for torque converters have the stator blade geometry as a result of tuning the perfor-
also risen. Specifically, there is a need to shorten the axial mance characteristics of the stator. The reason for this is
length, i.e.. achieve a flatter design, to facilitate easier that the stator, like the pump and turbine, is assumed to be
mounting in front-wheel-drive cars. There is also a need to a flow channel, even though it would be more rational to
improve torque converter performance to obtain better fuel treat it as a cascade.
economy. To meet these requirements, it is necessary to Attempts have been made to overcome this problem by
have analytical tools that yield accurate performance pre- treating the stator as a cascade and analyzing flow around
dictions in the shortest time possible. it as the basis for making performance predictions. Minato
Predictions of torque converter performance have been et al. [4] used the finite difference method to analyze
made by Abe et al. [1] and Fujitani et al. [2] using two-dimensional flow around the stator: the outlet flow
three-dimensional models to analyze the pump, turbine and angle thus obtained was then incorporated in the one-di-
stator, the three elements of a torque converter. With the mensional angular momentum theory to predict torque
present level of technology, however, their approach is not converter performance.
entirely practical because of the enormous time needed to In this work, the authors used the finite volume method
prepare a calculation grid and perform the calculations. to perform a two-dimensional analysis of flow around the
A simpler approach employed heretofore has been to stator. In addition to the outlet flow angle of the stator,
make performance predictions using the one-dimensional stator loss, which is thought to have the largest influence
angular momentum theory [3]. With this approach, flow on torque capacity characteristics, was also incorporated in
through a torque converter is expressed in terms of one-di- the one-dimensional angular momentum theory to make

11389-4304/94/$07.1t0 ;e 1994 Society of Automotive Engineers of Japan, Inc. and Elsevier Science B,V. All rights rcserved
SSD11)389 4304[ 94 )[1[)1)31-N JSAE9434866
310 M. K u b o el al. / ./,%~E Ret iew 15 G 974) 3 0 0 - 714

Energy Equation
3
~.~,Ti co i/~o g Q =h 1+h2+~3
t --1 Flow Analvsis
_. Ptlinp - "~ PeCifications of stato-r-t
h: (h;sx head) f
•Turbine ..... esh to be generated |
L~, ~ (outlet 11(),,'~anglc~i ~ ,~221 112 (boSS[Itcatl}
............ ~ - i : > :
S*',tor iwo inlet flow angle a 31_.j~-. . . . . . . . .

i f,Dd]TF~(((~P / ~ .7 (outlet flow angle)[

outlet flow angle a 32 ]


~ :1, ~ I l l ~ i
Torque Converter
total pressure loss coefficient ,; i

Fig. 1. Characteristics of proposed method.


i ~"=f( a 31) i
performance predictions. The results obtained indicate that ,_i j
i one-dimensional
this approach improves the accuracy of torque converter [" ----~ angula
ular momentum
performance predictions. i theory
I characteristics --q
of torque converter _j

2. Calculation method II

2.1. Outline
Fig. 2. Flowchart of proposed method.
A notable feature of the method proposed here is that it
uses the results of flow analysis in addition to the outlet
flow angle and loss of the stator, the two values conven- outlet flow angle, 0t32, in relation to the inlet flow angle,
tionally employed in the one-dimensional angular momen- 0~3~, of the stator. The total pressure loss coefficient, (,
tum theory (Fig. 1). A flowchart of the overall method is here is given by
shown in Fig. 2. The following sections describe in detail
the procedures used in analyzing flow around the stator _£= Pt i -- P t 2 ) / P t 2 (I)
and in predicting torque converter performance.

2.2. Flow analysis around stator

In the flow analysis, the stator was treated as a two-di-


mensional developed cascade in the design path. The
analysis was performed with STAR-CD, a general flow
analysis code based on the finite volume method. The
standard k - e model was used as the turbulent model, and
the log law was employed to find the flow velocity distri-
bution near the wall. Figure 3 shows the grid distribution
around the stator. The grid used in this analysis consisted
of 3400 cells. To obtain better calculation accuracy, a fine
mesh was used in the vicinity of the wall.
The boundary condition in each region was defined as
follows (Fig. 4):
Region 1: uniform inlet flow condition (calculations
were performed for varying inlet flow angles);
Region 2: constant pressure condition (P2 = 0);
Region 3: cyclic condition with region 4 (pressure and
flow velocity of corresponding mesh locations coincided);
Region 5: wall (flow velocity = 0).
A flow analysis was conducted under the foregoing
conditions to find the total pressure loss coefficient, ~', and Fig. 3. Grid distribution around stator.
,44. K u b o et al. / . I S A E Re~'iew 15 (1994) 3 0 g - 3 1 4 311

Region3
Blade A Blade B I
Region I ~egion5
inlet 0 inlet
-36' %
Region4~ I Region2
:outlet
Fig. 4. Boundarycondition.

where ~" and ~_~ are defined by the following expressions


~ = 58.3 ~
~ / outlet
'~" _68 °

as functions of ~<:
Fig. 6. Stator specification',.
¢-F(,~3~ ), (2)
,~: = G(~ l )" (3) where
The functions F and G are found as blade characteristics
unique to each stator. hf, = X , c : / ( 2 g ) . {1 + (tan-~o~,,l + tan~h.: ) / 2 }, (9)
h~i = & f ( 2 g ) . { ( r , . , % - c tan o~,.1)
2.3. P e r f o r m a n c e predictions
---(F i 1.2(Di i - - c t a n c t 1.2)(ri 1.2/Yi,l) (10)
The performance of a torque converter is predicted from where A~ and &i are the friction loss coefficient and the
the following equations of angular momentum and conser- shock loss coefficient, respectively. For thc stator, loss is
vation of energy: expressed in terms of the total pressure loss coefficient, ~",
Equation of angular momentum: and the outlet flow angle, a3,, and is given by
3
~f=0. (4) h3 = ( ' / ( 2 ,g ) - c-'( 1 + tan-'a3._,). ( I 1)
i I

where
T = pQ{( rm_ % - c tan ~i.:)rl.a
(r, l.ato i c tan oq l.e)ri 1,-'}; (5)
Equation of conservation of energy:
3
;2, - y_, T,,oi//,,~O (6)
i I

where It I is the sum of the loss heads of the three


elements and is given bv
h L = tt I + h~ + h,~. (7)
For the pump and turbine, loss is expressed as the sum of
the friction loss head, hf, and the shock loss head, h~, and
is given by
e=0 (Blade A) e=0.8 (Blade A)
2
h 1 +]12= E ( h l i q - h , , i ) (S)
t I
,'5,

I. e=0 (Blade B) e=0.8 (Blade B)


Fig. 5. R c l a t m n of end wall area to blade surface area. Fig. 7. S t r e a m l i n e around stalors.
312 M. Kubo et at./JSAE Reciew 15 U994) 309-314

1" [an &, -~!


-2
--[,
....................
I
+ . . . .
z ~-t--

0,8- High speed ratio o s~ Low speed ratio


Blade A |
0,6- -l i Blade A
-1..51 ................... ~ ......
0.'t-
High speed r a t i o N . ~/~Low speed rmio 24 [1lade B

/ -2 5. i t,
4"
./

f2 -1 ; i
tart a 3~

Fig. 8. T o t a l p r e s s u r e loss c o e f f i c i e n t .
Fig. 9. O u t l e t f l o w a n g l e

In this work, the value of ff was found by analysis, of these end walls is proportional to their area (Fig. 5) and
treating the starer as a two-dimensional cascade in the can be expressed by
design path. For an actual stator, however, it is necessary m = 1 + cos /3/(0-. a). (13)
to take into account the effects of the end walls of its
upper and lower surfaces [5]. To do that, if' is corrected Torque converter performance can be calculated by
using the correction factor m and is given by simultaneously solving the system of Eqs. (4)-(12). In this
work, certain given values were assumed for the friction
~" = ~'X m. (12)
loss coefficient, shock loss coefficient and slip angle of the
It is also assumed that the increased loss due to the effects pump and turbine.

4.0 i
. Present Calculation

1
3.8 1
. . . . . Conventional Calculation
3.6 \x. o---o Experimental
3.4) \.

Present Calculation 32
-- Conventional Calculation
3.0 1)
6.0 F 3.0 3.o I
5"6 1- 28
2.8
2.8 ~
\,,,r
~-'E2.6 2,6
5.2 I" 2.6
2.4
4.8~- 24 2.4
2.2 2.2
O "

100
~ 2.0 - o2.0 I00

90 " ~,1.8 90
a.61-~ 1.8 ~16 80
80 cr " " ~1.6
I-'-
I- 1,4 - 14 70
70
1.2
~ 2.4 i- 12 6o~ - 1.2
}
).,..
so~ 1.0 - 1.0

/
2.0 1.0
"D
I 0.8 - 0,8 4 0tu
~
1.6 I- 0.8
0.6 - 0.6 t I** 30
1.2 [- 0.6 30
0.4 - 0.4 20
0.8[- 0.4 20
0.2 - 0.2 i0
10
0 0
0
I , I I I I I I I I

' 01.2 01.3 o,I o'., o',


0.1 6 o.'8 o'9 to 0+1 0.2 0.3 014 0mS 0.6
Speed Ratio
0.7
o
0.8 0.9 110
Speed Ratio e

(a) Blade A (b) Blade B


Fig. 10. P e r f o r m a n c e p r e d i c t i o n s .
M. Kubo et al./JSAE Rel'iew 15 (1994) 309-314 313

3. Analysis results the values of the friction loss coefficients used for the
pump and turbine in the predictions made with the pro-
3.1. Test stator posed method. In comparison with the conventional
method, the method proposed here achieved virtually the
Figure 6 shows the profiles of the two stator blade same prediction accuracy for the torque ratio and effi-
geometries used in the analysis. Blade A was a thick type ciency of both blades A and B. With regard to torque
of blade, capable of maintaining a low loss level over a capacity, the proposed method showed a flat characteristic
wide range of speed ratios. Blade B was a thin type of curve, i.e., torque capacity displayed little change as the
blade, designed to reduce loss to a very low level within a speed ratio increased, in the low to medium speed ratio
narrow range of speed ratios. range. That torque capacity characteristic was obtained
because the proposed method made it possible to calculate
3.2. F l o w analysis results the sharp change in loss resulting from the flow separation
that occurred behind the stator.
Figure 7 compares the stream lines around the two However, the proposed method tended to underestimate
stator blades for speed ratios of e = 0 and e = 0.8. Fig. 8 the speed ratio at which separation occurred. As a result,
shows the relationship between the total pressure loss the predicted capacity differed somewhat from the experi-
coefficient and the stator inlet flow angle. Fig. 9 shows the mental data. The following can be considered as possible
stator outlet flow angle in relation to the inlet flow angle. reasons for that difference:
In Fig. 7, it is seen that separation occurred behind the (1) The point of separation in the flow around the stator
suction side of stator blade A at a speed ratio of e = 0. was not calculated accurately in the flow analysis.
Also, a region showing an exceptionally slow flow veloc- (2) Even though the flow showed a tendency to sepa-
ity formed on the pressure side at a speed ratio of e = 0.8. rate, the effect of that separation was not fully calculated
The effect of this separation is thought to have caused the (a problem of three-dimensionality). That tendency might
sharp increase in stator loss seen in Fig. 8 on the both the have been caused by lower solidity of the periphery of the
high and low speed ratio sides. stator than at the design path or by stagnation of a low-en-
In contrast, since blade B achieved a smooth flow with ergy secondary flow. These are points that need to be
little resistance (Fig. 7), it showed much less loss in the examined in future work.
high speed ratio region (Fig. 8). However, as the incident
angle was increased, separation occurred from the front of
the stator suction side at around a speed ratio of e = 0.5, 5. Conclusion
and the separation region soon extended over the entire
The present work focused on flow around the stator,
suction side. As a result, loss increased sharply once the
which has the largest effect on the torque capacity charac-
incident angle coefficient ( = tan o~3~) exceeded 0.5, as
teristics of a torque converter. An attempt was made here
indicated in Fig. 8.
to improve the prediction accuracy of the conventional
As seen in Fig. 9, the stator outlet flow angle of both
one-dimensional angular momentum theory by applying
blades A and B increased as the inlet angle increased,
the results obtained in a two-dimensional flow analysis.
except when the inlet angle of blade B was large. That is,
The following points were made clear as a result.
the slip angle, representing the difference between the
(1) This improved method makes it possible to predict
outlet blade angle and the outlet flow angle, increased as
changes in torque capacity characteristics due to flow
the speed ratio decreased. Yet, no large change is observed
separation that occurs on the suction side of the stator. It
in the outlet flow angle overall. This is attributed to the
fact that the separated flow became reattached behind the has not been possible to identify such changes heretofore
stator. in one-dimensional performance predictions.
(2) Because the method proposed here uses a general
flow analysis code, it enables performance predictions to
4. Performance predictions be made easily and in a relatively short period of time.
Consequently, it is sufficiently practical for use in design-
Using the results described in the previous section, ing actual stators for torque converters.
performance predictions were made and the predicted per-
formance was compared with experimental data for an Nomenclature
actual torque converter. The results obtained with the
method proposed here are compared in Fig. 10 with perfor- = torque (Nm)
mance predictions made with the conventional one-dimen- w, = angular velocity (rad/s)
sional angular momentum theory. To facilitate this com- c = circular flow velocity ( m / s )
parison, the same friction loss coefficients were applied to Q = volumetric flow rate (m3/s)
all three elements in the latter performance predictions as rij = radius (m)
314 M. Kubo et al./.ISAE Ret'iev~ 15 (1994) 309-314

~£ii :
inlet and outlet f l o w angle (deg)
]7 = pressure (Pa)
Pt = total pressure (Pa)
p density of fluid ( k g / m :')
g = acceleration o f gravity ( m / s : )
total pressure loss coefficient
hi = loss head (m)
angle relative to axial direction of blade chord
1 = length o f blade chord (m)
A -- pitch (m)
H = blade height
¢F : solidity ( l / s )
aspect ratio ( H / l )

Subscripts
= 1: p u m p
2: turbine
3: stator
= 1: inlet
2: outlet

f-

References

[1] K. Abe et al., Three-Dimensional Simulation of Flow Through a


Torque Converter, (in Japanese), Proceedings of JSAE, 902, October
1990.
[2] K. Fujitani et al., Computational Study on Flow Through a Torque
Converter, SAE Paper No. 881746.
[3] T. Ishihara, A Study of Hydraulic Torque Converters, Report of the
Institute of Industrial Science, University of Tokyo.
[4] K. Minato et al., A Performance Prediction of Hydrodynamic Torque
Converter, SAE Paper No. 900555.
[5] W.L. Stewart et aL, A Study of Boundary-Layer Characteristics ol
Turbomachine Blade Rows and Their Relation to Overall Blade Loss,
Rolls Royce, Maserati, Saab, Aston Journal of Basic Engineering, Transactions, ASME, pp. 588-592
Martin, Innocenti, Jaguar, and Lotus !September 1960).
have one thing in common.
NGK spark plugs.
All Japanese car makers also
specify NGK as original equipment and
factory-authorized replacement parts.
The unique copper core in every
NGK plug eliminates overheating and
reduces fouling. That means improved
heat dissipation for peak performance
under any operating conditions.
No wonder NGK is recognized as
the perfect plug for any engine.

S-ar putea să vă placă și