Sei sulla pagina 1di 81

UNCLASSIFIED

AD 27

5626

ARMED SERVICES TECHNICAL INFORMATION AOTCY


ARUNGTON HALL STATION
ARLUNGIN 12, VIRGINIA

UNCLASSIFIED

Best
Avai~lable
Copy

NDTICE:

When goverment or other drawings, speci-

fications or other data are used for any purpose


other than in connection vith a definitely rented
goverment procurement operation, the U. S.
Government thereby incurs no responsibility, nor any
obligation whatsoever; and the fact that th, Govern, way
ment Wey have formilated, furnished, or I
other
ed the said drawings, specificatiov
data is not to be regarded by implication jr othervise a.s in any manner licensing the holder or any
othsr person or corporation, or conveying any rights
or per,.ission to -nnufacture, use or sell any
patented invention that may in any way be related
thereto.

ARL

tI

62-318

5-

LAMINAR FLOW BETWEEN TWO PARALLEL


ROTATING DISKS

MARK G. BREITER
KAR. POHLHAUSEN

APPLIED MATHEMATICS RESEARCH BRANCH

AERONAUTICAL RESEARCH LABORATORY


OFFICE OF AEROSPACE RESEARCH
UNITED STATES AIR FORCE

rrI

ii
FOR ERRATA

A275562
AD

THE FOLLOWING PAGES ARE CHANGES


TO BASIC DOCUMENT

March 1962

ARL 62- 318

ERRATA - July 1962

Correct the last three lines on page 11 to read:

i wr

AERONAUTICAL RESEARCH LABORATORIES


OFFICE OF AEROSPACE RESEARCH
UNITED STATES P IR FORCE
WRIGHT-PATTERSON AIR FORCE BASE, OHIO

ADI275562

END CHANGE PAGES

NOTICES
N'e'-to G)-Jerlment drawings, specifications, or othecr tin .irc u-sed for any purpose other
than in euic1ion with a definitely related Government procurement operation, the Unite States
it-iirtiment thereby incure no responsibility nor any obligation whatsoever; and the fact that
thk 1L.11;.rnn ent may have formulated, furnished, or in any way supplied the said drawingg,
spe.lfiicatiow, cr other data, is not to be regarded by imprlication or otherwise as in any manner
licernsing th( h~ider or any other person or corporation, or conveying an~y rights or permission
to nmanufactirE, use, or sell any patented invention that may in any way be related thereto.

Qualifil'-d 7rquesters may obtain copies of this r o p o r t from the \,rmed Services Technical
TnformatiClL Agency, (ASTIA), Arlington Hall Station, Arlingoui 12, Virginia.

Thi :report has been released to the Office of Technical Services, U. S. Department of Cmnierce, Waishington 25, D. C. for sale to the general public.

Copies of AR. 'Technical Reports and Techn!ca! Notes should not be returned to Aero-autlcal
R-!x.arrc, Laboratory unless return is required by security con siderations, contractual oblgations,
-,r notic s on a specific document.

ARL 62-318

LAMINAR FLOW BETWEEN TWO PARALLEL


ROTATING DISKS

MA.9K C. BREITER

KARL POHLHAUSEN
APPLIED MATHEMATICS

RESEARCH BRANCH

"MARCH 1962

Project No. 7071


Task 70437

AERONAUTICAL RESEARCH LABORATOBY


OFFICE OF AEIROSPACE RESEARCH
UNITED STATES AIR FORCE
WVRIGHT-PATUERSON AIR FORCE BASE', OHIO
300 - May 1962

31.366 9 1367

FURE WORD
This report was prepared in the Applied Mathematics Research 13)
Aeronautical Research Laboratory, Office of Aeroispace ResearC-., Wr
Patterson Air Force Base, Ohio~.
The authors extend grateful acknowledgement to Dr. K. G. Guderlt
the Applied Mathematics Research Branch, for encouragement and muadvice.
The numerical calculations were carried out partly with the Recon
graciously made available by Professor R. T. Harling of the Air Forc,
of Technoulogy, and partly with the 1103A computer of the Digital C-Friip
Branch, Systems Dynamic Analysis Division.

Captain R. V. Hendon of

Computation Branch rendered valuable assistance in programming.


The work was carried out un.

?roject 7071, "Methods of Mathev

Physics", Task 70437.

BEST
AVAILABLE COPY

FOREWORD

i
This repo t was prepared in the Applied Mathematics Research Branch of the
Aeronautical Research Laboratory, Offlce of Aerospace Researc'., WrightPatterson Air Force Base, Ohio.
The authors extend grateful acknowledgement to Dr. K. G. Guderley, Chief of
the Applied Mathematics Research Branch, for encouragement and much helpful
advice.
The numerical calculdtions were carried out partly with the Recomp computer,
graciously made available by Professor R. T. Harling of the Air Force Institute
of Technology, and partly with the 1103A computer of the Digital Computation
Branch, Systems Dynamic Analysis Division.

Captain R. V. ilendon of the Digital

Computation Branch rendered valuable assistance in programming.


The work %L,

.a,

ried out under Project 7071, "Met.

Physics", Ta k 7043

ii

of Mathematical

ABSTRACT
The report inrestigates the viscous flow between two parallel disks rotating in
the same direction with the same velocity.

The fluid enters the space between the

two disks at a certain radius In the radial direction.

fcause of the shear forces,

it assumes a rotating motion with about the velocity of the disks.

The centrifugal

forces then build up a pressure increase in the radial direction.

The arrangement

corresponds to a centrifugal fluid pump, which may be advantageous if cavitation


is ai problem.
The general equations of viscous flow are simplified by the assumption that
the pressure difference normal to t..le disks is negligible (boundary layer assumptions).

One obtains a system of parabolic partial differential eq,,ations.

For large

radii the deviation from rigid body rotation (with the angular velocity of the disks)
is smail.

The lsi, _:- i equations which then result are s.

, analytically.

The

velocity profile. dePenc! ipon a parameter containing the kinematic viscosity, the
angular velocity and the distance of the disks, but not the radius.
The non-linearized parabolic differential equations are approximm.Led by a
difference scheme and solved numerically.

The results are given in non-dimen-

sional form with the entrance velocity and the distance of the disks as parameters.
Furthermore, the efficiency of the pump is computed from the gain of the total
pressure and the torque at the shaft of the rotating disks,

iii

TABLE OF CONTENTS
Page

Section
1. Introduction

Statement of the Problem

111.

Linearized Treatment and A symnptotic Solution

IV.

Numerical Solution of the System of Equations

11.

(The Inlet Problem)

10
14

V. Results and Conclusions


VI.

17

Appendix

iv

LIST OF ILLUSTRATIONS

Page
Figure
1.

Schematic diagram of the flow between two rotating disks

21

2.

k I and k 2 versus ad

22

3.

g(ad) versus ad

23

4.

Radial velocity u and relative tangential velocity V for


24

various values of ad
5.

Radial velocity 5 and relative tangential velocity V for


25

ii0= 1.0 d=0.5


6.

Radial velocity 5 and relative rangential velocity Vfor


26

0 =o. 5 d=0.5
7.

Radial ve jct,.. a ard relative tangential velocity N


d0

8.

= 0,

.5

or
27

.
C .5

Radial velocity d anid relative tangential velocity V for


28

= 0.5
0. 1 adi0=
9.

Radial velocity il and relative tangential velocity V for

ao = 1.0 a=l.O
10.

29

Radial velocity ai and relative tangential velocity V for

30

i0 = 0. 5 d=l. 0
11.

Radial velocity i and relative tangential velocity V for

5 0' 0. 25
12.

31

= l.0

Radial velocity 5 and relative tangential v,-locity V for

0o= 0. 1

32

=l.0

LIST OF ILLUSTRATIONS (Continued)

Page
Figure
13.

Radial velocity i

and relative tangnetial veloc*.Ly V for

a 0= 1.0 d- 1.5
14.

Radial velocity

33
i and relative tangential velocity V for
34

f0 =0.5 d=l.5
'A5.

Radial velocity d and relative tangnetial velocity V for


35

10- 0.25 d =1.5


16.

Radial velocity 5i and relative tangential velocity V for

36

a 0=0.1 j= 1.5
17.

Radial velocity 5 and relative tangnetial velocity V for


r,0 =0= n j - i7

18.

Radin. velocity i and relative tangential velocity V for

u0 =0.5 d=2.a
19.

38

Radial velocity 5 and relative tangnetial velocity V for

39

i0=0.5 d=2.0
20.

Radial velocity 5 and relative tangential velocity V for

5 o= 0.

40

2=2.0
d

21.

Radial velocity 5i ni versus r/r.1 for d=0.5

-&1

22.

Radial velocitv 5 m versus r/r. for d=l.0


1

41

23

Radial velocity 5i

42

24.

Radial velocity 5G versus r/r. for d= 2

25.

Fi()

versus r/r. for d=.5

for various values of 6 0 versus r/.

0v1

vi

42
for d=0.5

43

LIST OF ILLUSTRATIONS (Continued)


Page

Figures
26.

T"(Q) for various values of 50 versus r/r. for d= 1.0

44

27.

F(0

44

28.

P(r) for various values of a0 versus r/ri for d=2.0

29.

Moment, pressure, and efficiency versus r/r

for various values of -0 versus

for d:.5

for

d =0.5
10.

46

Moment, pressure, and efficiency versus r/ri for


d= 1.0

31.

47

Moment, pressure, and efficiency versus r/ri for


S=1.5

32.

45

Moment, pressure, and efficiency versus r/r


d=2. 0

48
i

fo49

I.

INTRODUCTION

Investigated in this report is the viscous flow between two parallel disks,
which rotate with the same constant angular velocity in the same direction.

The

fluid enters the space between the disks in radial direction through a cylindrical
surface (Figure 1).

It is set into a spiral motion whose tangential and radial

velocity components are due to the friction forces and to the centrifugal forces
respectively.

The pressure will increase with the radius because of the centrif-

ugsl forces, thus the arrangement can serve as a centrifugal pump.

One might

suspect that the absence of blades leads to low values of the efficiency, but if the
blades are spaced very closely and the mass flow is low, the efficiency may be
quite acceptable.

On the other hand, such a pump would not encounter the problem

of cavitation and also its characteristics over a wide operatin -ange might be
more favorable than in i conventional turbo pump.

These advd,,tages were recog-

nized by Mr. S.: . Hasinrer, Mr. L. G. Kehrt and Dr. J. P. von Ohain of the
Thermomechanics Research Branch of this laboratory.

Technical details and

experimental results will be published by them in a future report.


The present mathematical analysis will determine the velocity distribution
between the disks, the pressures, the torque applied at the shaft and the resulting
efficiency.
There exists an extensive literature on the single disk rotating in an infinite
medium*.

Assuming that the disk extends from radius zero to infinity, it is

*Advances

in Applied Matheratics, Academic Press, Inc..

1956, p. 166.

F. K. Moore:

New York, Vo) IV,

Three-Dimensional Boundary Layer Theory.

possible to reduce the partial differential equations governing this problem to a


sy3tem of ordin:'y differential equations by means of a similarity bypothesis.

But

an approach of this kind is not feasible in the present case because of the presence
of the seccid disk and the fact that the flow enters the space between the disks at a
radius different from zero.

However, another simplification can be made here.

For any practical application the distance between the disks must be very small,
for otherwise the friction would not be sufficiently effective to produce a tangential
f1jv,.

But then the approximations made in boundary layer theory apply reducing

the original system of Navier-Stokes equations, which is elliotic, to a system of


parabolic differential equations.

They are derived in Section II.

A first insight into the basic characteristics of the proble n is obtained if one
assumes that the velocities of the fuid relative to the disks are small so that
quadratic ternr. , -.-

gligible.

This is a condition whici

large radii; F a idea ii carried out in section Ill.

'ell satisfied for

It gives valuable insight into

the general character o" the velocity profiles; these profiles are governed by one
dimensionless pararmeter which contains the angular velocity, the vi-cosity and
the distance, but not the mass flow and the radius.
important for the practical design.

Naturally this parameter is

The solutions thus obtained can be considered

as an asymptotic expression for the solution of the actual non-linear system.


In the icinity of the entrance radius the linearization is not justified, in particular the tangential velocity relative to the disks is certainly not small.

This

region is most critical from the point of view of cavitation, and also the principal
losses are encountered there.
been integrated numerically.

Therefore, the system of differential equations has


The methods and the results are shown in section

IV and V and in an appendix.


2

STATEMENT OF THE PROBLEM

fl.

We use a systrn of cylindrical coordinates r and z (Figure 1), where the


z axis coincides with the axis of symmetry and r is the distance from it.

We

denote by

ai

the component of the velocity in the radial direction

the component of the velocity in the tangential direction


measured in a system of coordinates fixed in space

w
2d

the component of the velocity in the axial direction


the distance between the two disks

ri

the radius of the entrance hole

the density

the viscosity,

th. pru sure

the kinematic viscos'"

v =

he commant anguldr veloci 'y.

An incompressible axi-symmetric steady viscous flow is described by the NavierStokes equations


Ov +wav
ur

wT:

w
ur

+ u. v

z
2z
8z

.!

8w

u
+r

+ V

8z

-Or

Oz

+4v8V
22
8r

__ +
8z

--.+
r Or

(2.1)

8r2

au

vz

8
I Ip + V+ &u

A..
L

!u

a-:

p Or

8z 2

Or

and the equation of continuity,


a__2w =0

Ou +

Or

8z
3

(2.2)

By applying the boundary layer approximations, this system of equations is reduced


to
_U
iu

OU 2
MV+

-r

au
w 2-=.Oz

22u
Oz2

dr

(2. 3)
u

u. .v + ww K
r
z

Pr

Xv

2--

8w X 0
+r2uu+ L
r
r'

Or

(2.4)

8z3

It is convenient to introduce instead of v, which is the tangential velocity in a


fixed system of coordinates, the tangential velocity V relative to the disks

V1, v - rw

(2.5)

One then obtains


u2

Ou V

Vrr

8V

=-

iz

u[V+,rw] +
r

Along the disks, given by plane z -

V
z =

dp

Pdr

2u

O(2.6)

av
az 2

(2.7)

d, one has the boundary cnditions

u (r, *d) =0
(2.8)

V(r, *d)= 0

To determine the boundary value problem com.pletely, one must also prescribe
the velocity disti ioution in the entrance cross section.

u(ri, z)a

One has

u0

(2.9)
V(r

,z)-rw

IMI. LINEARIZED TREATMENT AND ASYMPTOTIC SOLUTION


Omitting quadratic terms in u, V and w one obtains from Eqs (2. 6) and (2. 7)

8z2

dr

V
0

does not depend upon z becauL


p&w

(3.2)

az

u
In these equations

(3.1)

che boundary layer

dr

assumption thai rhe pre sure does not depend upon z. The unknown w does not
appear, and so it is possible to solve these equations separately; w is determined
from Eq (Z.4) afterwards.

It is ecmarkAble that in these equations .!,ere appear

no parlai derivatives with respect to r.

This means that, except for the expres-

sion

p'aw

dr

rw = F (r)

(3.3)

which depends upon r a nd is undetermined so far, the equations can be solved


for each value of r independently.

The expression (3. 3) is the only inhomogeneous

term that occurs in the equations and in the boundary conditions; therefore, it will
enter the solution for u and V a a a factor which depends only on r.
5

This factor

is determined by the condition that the mass flow between the disks is constant for
Thus u and V as well as

every cross section r =const.

dr

can be found.

After elimination of u one obtaiv, from Eqs (3. 1) and (3. 2)

2
P

2rZ

Setting

4V
d V + 2V z F(r)
dz 4

(3.4)

one has finally

dV + 4a 4V
4

24F(r)

(3.5)

dz...

An the most general solution of Eq (3. 5) that satisfies the symmetry conditions at
the plane z v 0,

a
ac

V=A

.,: 1

sin!, (az) sin (oz)+A

cosh(az)cos(az)+ F(r)
2

where A I and A 2 are two constants which will be determined presently.

(3.6)

The

boundary conditiona (2.8) together with Eq (3.2) give fcbr z =d

V=0
(3.7)
1
22

d2 V
2
dz

Hence, by inserting Eq (3.6)

A,

F(r) k
1
1 2

k2

-F(r)

38
(38

with

a). sin (ad)


asi
2
cosh'.r.)+cos42ad)

(3.9)
k2

2 cosh(ad).cos (ad)
2 cosh (2ad)+cos(2ad)

Figure 2 shows k I and k2 as a function of ad


The shear force at the wall is proportional to dV/dz.

dV

F(!) ! (k

dz

One finds

, ) cosh(az) sin(az)+(kl +k 2 )einh(az)coo(az)

Let

t)he

a surface

flow of liquid (measured by volume) be given by 0.

The flow through

r = const extending between the disks is given by

+d
Q= 27r f
-d

Hence

(3.10)

(320

*a2

udz= I
a

-d

+d

2
df
dz

dz =

(3.11)

dV
a

-d

ith Eq (3. 10)

. ..
a

F(r). g (ad)

(3.121

where:

g(ad) a

Figure 3 shows

sinh(Zad)- sin (2ad)


cosh(Zad)+coo (2ad)

(3.13

g(ad) as a function of ad

irrom Eq (3. 13) we find the function F(r)

IFW

4 eQ

w- r.g(ad)

(3.14)

where

-w. g(ad)

(3.15)

One thus obtain, finally tor the velocity components, expressed in terms of the
mass flow

v U3 [

k,

,inh(az) sin (, z)+k. cosh(z) cos(az)- 1

(3. 16)

(3. 16)
Ua

-[kl

cosh(az)cos(aL.-k

sinh(a-) sin (az)]

Inserting these expressions into the continuity equation Eq (2.4) one finds w = 0.
The velocities in the middle plane between the two disks are found by setting z = 0.

Z,
=0 r

(3. 17)
zZO=

5U
r kl

Figure 4 shows the radial velocity component for different values of ad as a function of z/d for q/r = I.

For small values of ad, the profiles have a near para-

bolic shape with maximum velocity in the middle (z = 0).

For ad = w/Z the profile

is flat in the rviddle. for higher valie of ad the flow is moro and more concentrated
in the neighborhood of the disks.
The tangential velocity V is represented in Figure 4 for q/r = 1.

For small

values of ad it also has a parabolic shape, but with increasing ad the slope in the
vici;:ity of the disks becomes steeper and the relative velocity in the middle plane
ajroaches -1.
Actually the profiles for high values of ad must be viewed with considerable
caution.

One sees that in this case the velocity u may become n 1gative in the

middle; i. e. for such cases one has a radial outflow close to the disks with a
radial inflow in t'k

roddle.

to produce such

o, pattern physically,

for technical aFplications.

Obviously it would take a rathe

-ial arrangement

Theoe results ought to be disregarded

Important in any case is the role played by the dimen-

sionless parameter ad which determines the character of the profiles.


character is neither inflaenced by the mass flow Q nor by the radius r.

This
For a

given w and a given v, I. e. . a given fluid, the only quantity that can be influencid
in thi3 parameter is the distance of the disks.

If one wants to use high angular

velocities, one must try to lower the value of d by skillful design.


The pressure distribution is obtained from the function F(r) in Eq (3. 5).
Inserting Eq (3. 14) into (3. 3).
one finds

pw [r -2q

dr

(3.18)

and by integration
2
p=

[r.

(3.19)

. 2qlnr +const]

The momentum required at the ahaft of the rotor can be determined in two
different ways, either from the moment of the shearing forces on the inner side
of two disks, or by applying the law of conseration for the moment of momentum.
One has in the first case
r r z I dVd

M =-2- z.

'

ri

'-dr

(3.20)

iz zd

in the second
l4 u2,rr 2 p

u.vdz

(3.21)

-d

For the numerical ealuation the second one is preferable.


IV.

NUMERICAL SOLUTION OF THE SYSTEM


OF EQUATIONS (THE INLET PROBLEM)

The approximation shown in the previous section is unsatisfactory in the


vicinity of the entrance, where the velocity components are certainly not small.
Moreover, the lineariration causes all r-derivatives to vanish, i. e. it introduces
such i strong change of the character of the flow equations, that it becomes
impossible to satisfj the boundary conditions at the inlet.
portant for it accounts for most of the losses.

10

But the inlet is im-

As was mentioned above, the introduction of the boundary layer assumption


gives us a parabolic system of differential equations where lines r = const are the
characteristics, therefore, it is possible to proceed in the numerical solutions
from one line r = const Lu the next one without any need to go haclk.

Thus, we may

adopt a numerical scheme for the integration of the parabolic differential equations.
For our computations we assumed that the rotating disks lie in horizontal
planeR, and introduce a new system of coordinates x, y, whose origin lies in the
entrance cross section at the lower disk.
loder disk.

Thus the x-axis coincides with the

We have
x~rr
y = z+d

uO
0

ri

We introduce dimensionless variables in the system of ,'Iferential


(2.6), (2.7),

and (2.4)
uo

r.
1
r.

r.1
Y=Y"; r~

1
-m

11

equations

One thus obtains

-+

!:

+
+2

"-[

(4.2)

where for abbreviation

F'(r) a r.

1421r Z d
dr

The functions 5 ind 7 and v are given at the entrance (i01.


extends from j = 0 to

=d.
-

The channel

The boundary condition at y = Zd can bd replaced

by a symmetry coiidition at i = d.

0)

(4. 3)

0)=0

V,(; 0) =0

Thus, one has as boundary conditions

a,! (q(,a)

=0

aV(x,d)

=0

(4.4)

V(;, a) = 0

The natural method of solving such a system numerically is the introduction


of a difference scheme.

Here caution is needed from the point of view of stability.


12

The coefficients of the derivatives with respect to ;., in Eq (4. 2) may become very
sma)l for large values of i and italso is small in the vicinity of y
is necessary to use a very fine mesh size in the ;

direction, if one applies a

'"iirect" difference scheme where the derivatives with respvct to


along the line I = const which is already known.

d, but then it

are computed

Otherwise the procedure would

become unstable.
For this reason it was decided to use an ";nverse" difference procedure where
the Ierivatives are formed along the line ; a const for which the state is to be
-Imputed.

This brings about a rather severe complication, for the stat. along

the new line ; = const is determined by a system of simutaneous equations.


Since the original differential equations are non-linear, this system of equations
is also non-linear.

At first the non-linear system was solved directly by an itera-

tion method.

this method proved to be inconvenient,

Sirw

fied in the follo, ng manner: In going from one line

?proach vras modi-

= const to the next one we

computed the chanies of i, V and Z rather than the values themselves.

If a

difference procedure is admissible at all, these changes ought to be small and


second order terms in these changes are negligible.

In other words we use a

linearization which considers as a basic approximation the values at the line


= const which has Leen computed previously.

The state along the new line can then be obtained by a suitable super-position
of particular solutions of the linear equatioa which arises in this manner.

The

linear system of equation which arises in this process has a rather simple structure insofar as the matrix of systems has elements that are different from zero
on only a few lines that are paraUel to the main diagonal.
certain simplifications in the inversion of the matrix.
13

This brings about

For further details of the

numerical analysis see the appendix.


V.

RESULTS AND CONCLUSIONS

The numerical procedure described above gives a nearly complete description


of the flow pattern.

Assuming constant velocity u 0 at the entrance, we give the

results for a set of four initial radial velocities

1. 0, 0. 5, 0. 25 and 0. 1.
1i--

initial tangential velocity V 0 is always V0=-I because of Eq (2. 5).

The

For the

dimensionless distance between the disks we choose d = 0. 5, 1. 0, 1. 5, and 2.


Figures 5 thru 20 show the dimensionless velocity profiles versus z/d with
r/r

as parameter.

inlet velocities

These p:ofiles have a common characteristic: at the wall the

0 and ;00 wb ch are constant over the cross section, are immedi-

ately reduced to zero because oz the boundary conditions.

The reduction of the

radial velocities at the wall causes the velocities in the middl. *.o increase.
fore the radial ve)c -. iE , &the profiles close to the entrance c.
shoots the entra, "e prnf;,-, in the middle.

There-

us section over-

The effect decreases with increasing r

because of the increase of the available cross section with r, finally the profile
approaches the form given by the linearized theory.

Figures 21 thru Z4 show this

behavior again; here the radial velocities Vm in the median plane of the
(z = 0) are plotted versus r /r..I

i41.

The average radial velocities are obviously de-

termined by the condition of continuous flow.


The average tangential velocity is directly connected with the moment M of
the shear forces and thus with the work that must be performed to drive the pump.
Further.more it gives the main contribution to the dynamic pressure of the fluid
particles as they leave the pump, thus it influences rtrungly the efficiency.
mentioned before shear forces can be obtained from the slope of the velocity
14

As

profiles at the wall Eq (3. 18); but it appears to be preferable (from the numerical
point of view) to compute it from the radial and tangential velocities Eq (3. 19).
The pressure distribution is computed from the function F()
in the numerical procedure as a function of r/r
direction of r is required.

which is found

Here an integration in the

In our figures the contribution to p due to the term

in Eq (4. 3) is shown separately, (as a straight line).

The pressures are given by

the difference of the curve ;(M) from this straight line.


Properly speaking the dynamic pressure varies from streamline to streaznIi-..

For technical purposes only the average velocity rark be utilized.

The

dynamic pressure for the average velocity is slightly lower than the a-erage of
the dynamic pressures; in other words in defining the dynamic pr :ssure with
average velocities one takes into account the mmxing losses.

Thus, we have accord-

ing to Bernoulli'p e-,tion

PLI2

av

The efficiency ?I of the process in I.he pump is best defined by


Ps
11= where Q = Zirri .
31 show, for

Pd

2du 0 denotes the amount of fluid entering at r..

Figures 29 thru

- p = L, the moment M, the total pressure p8+ Pd and the efficiency

versus

for the chosen values of 40 and d.


0
According to the linearized approach the profiles at large values of r do not
r1

depend upon the flow field at smaller radii, thus dynamic pressure and torque are
determined only by the exit cross section.
15

However the stz.tic pressure arises by

an integration over the pressure gradient in the radial direction.


encountered appear in this analysis as Icsse

The losses

in static pressure.

The exit cross section does not play a special role in the computation; in
other words, each value of .r can be considered as the other radius of the rotor.
In the vicinity of the erncrance, one has always a slight pressure drop, for
large volume flow the pressure drop is more pronounced, and this fact is of
interest from the point of view of computations.
nrental from the point of view of efficiency.

16

High mass flows are also detri-

APPENDIX
For the numarical solution of the system of equations (4. 2) we set up a network of lines x-constant and y
along the x and ; axes.

constant with a spacing of h and I

The length

is chosen as uzi

from the wall to the median plane of the disks.

respectively

twentieth of the distance

The derivatives in Eqs (4.2) are

replaced by difference quotients, e. g.


-u

u i+l,kmUi, k

rv

i+l,k+l

1i+l, k-I

(A 1)

a2 j

i+l, k+l

I+l,
l2 k

t2

8j2

5I+1, k-I

Here i and i+ I correspond to successive lines i = constav,


k+ 1 to success 4 e linei

= conqtant with k- 0 at the wall.

and k - 1, k, and
Starting with given

values of fi,V, 7 at che line x = const (i = 0) corresponding to the inlet, we


determine the unknown values of a. V, 7 on the (+ 1 )st line from the known
values on the i-th line.
It is convenient to introduce as new ,.wikowna Lho-differences Aa,
Arvk

k'

defined by

Ak

'ui+

lk

"ui,.k

and similarly for AVk , and &Ck' and neglect all terms in the equations of higher
order in the A's.

(Naturally these A's will also depend on i, but only one value

of i is needed at a time).

17

Equations (4.2) then lead to the following linear difference equations in the
unknown differences

--

1,k

,ii,AV, and A

and the unknown pressure function f(i).

2 - i,k

-.

+a)+ (2 + 4L

k-+

Lz

-(

i, k-1(1+ -"

Wi,k) + Ui, I

I2

+I1k+l

rl,)

A-k

;wk

(ik+li,

Wi,k)

(l-w.

ik

)A

wik+-i-+

VkI

zx 2

ik(

(1+Z)

v itkl
x, (+ IZVk- I

2~

A2t

kI&d

I
k- I -h

h
X

Auk_

Itk1
itk~

k2i,kk-i

itk

z-I a1

v
ik

k-

i,k+l

+V

iLk-l

In addition to Eqs (A2) the complete specification of the problem requires


Eqs (A3), (A4), and (A5) resulting from boundary conditions.

18

Since

= W=

0 along the wall, one has

A 50 = A

Since

0.*0

(A 3)

and V are symmetric and v antisymmetric with respect to the

median plane of the disks,

AGz - AG1 =0,


A,21 Ai19
0

AV

-21

= 0,

a Iv20

(A4)

0.

In the continuity equation there appear only first derivatives with respect to
',

whiie the otl . r .':'

Eq (Al) the fir.

"ns contain second derivatives.

derivtive of

values wvat adjacent points.


formula for

is lost.

'v

0,

.11 notice that in

with respect to j has -iot been formed from

This was done in order to obtain a central difference

for a2-;- andB2


a jsa
fom , just as
a--

As a consequence or.

equation

Therefore, the following approximation is made in the vicinity of the wall.

In the last of equations (4.2) the term a(I


mated by its linear term.

is zero at the wall; a may be approxi-

Thus
5= f(4
=

8w,-

f',( ).

f,(i)
I
K-

-2

7v =A(;) y-2
19

=A

(A5)

From the last equation one finds

-~1

wi,l

- 4

1.w,2

and so
-

J.

An alternate form i.s

A*,' i
I "A'

( "1
i,

wt, ) )
-

Here the second term ought to be zero, but by proceeding in this fashion we are
more sure th.L C

'-:

equation of (AS) is not violated bec

of error accumu-

lation.
By numerically integrating the radial velocity a with respect to
fixed value of ;,

j for any

we obtain the total fluid volume passing through thf- cylindrical

surface characterized by ;,

and comparison with input volume provides some

check of the numerical results.

20

-c-d

0 +d

r.r

Fig 1. Schematic diagram of the flow between two rotating disks

21

1.0-'-

-I

kk
0.8i

~
K,~~~~4

-:u1-2

I'

1~

41

0.

if

T
k, an-ti+g4.4+
versus ad

-H

i f I

1,4

---

I.0

10

0.223

--

AVAILABLE COPY

-,-1

I-..,,-T,

-T-

4It

-----FT

1 L4

140
Z-t-

41,

- -

444

414

-444-

C;
Il

lo

or

do
T
v

2.7

fl

c;"

-----ME

IU
I M
I I

a~

I-

44

I f
44J

4-3

-I-!0H

C;

-4

t1

0T

go
--..

>

0;

0;

83

1--

0-

U.'1T1v

AR*

4-H

A
I

I:mI

~~

T-T

I -I-

I
-T

V... .. .

414+

*-7T

c4

Afl

ca
-tC

4-Z

ILa

-----

L A

t-

+4-

,~in

lu

Ao

0I
[i

-I J-I
of

IA

If

oPol

go

N0
0

NI

.1

a4

41,,

a I

4?P1
I II I L0

Lil

41

0*1
xt

44~
o

N
N

0_

4'

'

-U

26I

01

0 f

f i l

. 11

I I Lu

444-44

ffis

dI

II

IllI
I4'

In

Ill

04

0r

it I
IfI
44

Ifni

at

IF..
..
..
L0

H
R4

26

0.',

0. 1

r1.

109

.0.0

0.6

05

-. 4

0.241

27.

..

.2

..

0.4~~~~

II

o0.

a/dM.

4-

T/

M0

IC___4
-

F1.
g9.08

veo
-0.ia

and4
-0. reai.tneta.0.

euiy Vf

'&/d

0.

0.

-01

I I

fil

0TT

[ f i l

CI

4+1i0-i

-1 1

f l

.4:pl

-IN
.

..

IfJill

11

LL

-f

j449

Vti

p41

T-

- 1 1 Il

141

1-u

LX

0L

~O

II

-I

1-0

-_

t4

TR

IIIIi

IfI I

it

If I

I'
0

I
.*-

IA

Yv

C;

00

J I

SI

It.9

A
I

0.8

I IsIIh
0.

i4fw

0.

~iiif

-1.0a.0.9al

0.40

-0.d

.0.7iv

.0.6nia

-0.
veo0t.V5o

d0.
-0.0

0. 6

.......
__

17 1

..

-1

1010

04

.1.0

-0.9

.0.5

.0.'7

-0.6

-0.5

-0.4

-0.3

-044

.0.1

ai

-0.

~1.01

Fig. 11 Radial velocity Z and relative tangential velocity Vfor

31

0. 25

;,ilit

0.

.10s .0.

.1.0

.0.9

.0..6

.0.8

.0.7

-0

.0.6

..

..

03

.0Z

..

-..

-0.4

-0. 3

-0.

.0.1

4_

~ ~ ~ ~adrltv~ ~ a~e~a
~ 1i~ ~eoi'~ Vfr~ ~ ~-i

Fig.

33-

__

1d
!

0_

44t.vlct

t
-

ki- 4-10

IIt

H44

-,1T

T-T

I
t

IT

TRIT

IN

LI

T7

Jo

II

+H0
c;I

aT1

4.4

VIIa

A' t

ITT

UIA

4.

147

ILI

01

_4

33

0.4

. .q

7.'i
-

o0
r/r.

o.

0 . .0.0

0 3.

1..

080

'344

IaI.

."fIT

0.',

.1.0

.0.9

.0. ?

-0.8

-0.6

.0.s

.0.4

.0.3

.0.~

. 0.1
.-

.1.0

.0.9

.0.9

.0.7?

.0.6

.0.5

.0.4

.0.3

-. 4

.04-

.0.

T0
I

Fag
Rdil
1

+1

11

elciy

ndreatvetanenia

vloit

Vfo

..

*t.

.0.1

0
a/d

0. 4

.1.0

.0.1
.0.9

.0..

.0.
.0. 5

.0...0

01

00

.1.0

-0.
-0.9

.0.

.06

O.

-0.

.0 3

.0 a

N-

IV

1U.

414,

4--

4q

1:4-

--

44>

37r

-0

0711

tt4

41

4-44

4j2

I~fl

~~4 IT

U44

CI

37

.0

0.6
00

1.0

33:.

t.4.

-l.0

~da

17.1

tneta

eoiy~adrltv

eoiyVfri

*5

00

-1.0 -0.9
0.8

.0.7

.0.6

-0. 38

.0.

.0.3

..

-.

0.46

0.

15.0

.0.9

.0.8

.0.1

.0.6

.0.

.0.4

.0.3

.0.3

*0. 1
,qp-

-3.0

.0.9

.0.8

.0.?

.0.4

.0,9

.0.3

-0.4

.0.

Fig. 19 Radial velocityT and ritlative tangential velocity

)90

for 700.25

I.2.0

.. 0.1

0
9/d

.1.0

.0.9

.0.8

.0. 7

-0.6

.0.5

"1

~d,',y

Pa,~0 ani
ad..I
.'at~..
4.

-0.2

.4

.0.4

4~tF~

v~o~,y

.0.3

.0.2

.0.1

11'UHHiMi

fo

0.I

a*

W 111 1

61144

ol
Z.;

11-

44

I f [if I
-4-

-4- ---

4-

.'

4t-

-4-

*~~~~~
'-i

-,- ,4

Ar~

I2T~~
--

:> 1F3tK&

_I

-H-

4-4i

-.. l..
..

-pr

4-;;

K~

>:-

.....

....

rJrn-'

LC2

T.~j~
___

:1;.7
jj;~:~
:.&-:.

_____

4-

,--~.1-E

.4

44

LA

to

-fill

0E

44'4

41,

LA~

4.

14*
I

+14

Im.7

.21

1
U4
I

_H__14

jqpfT

I,

F!C;

r-42

in1

'4-

I+I

+4-4-I

loll

InI

AA

of

T-

4'
'4l

#4#I

I
I

'4II

ItI

IvI

f1

go -

I 1I
Io

v
I I

I I

10

1 1 1 11 r Tr

4441+4+

if

I Ir

lv++H
1111

if

-FT_

-T,

Q.;

I. T

44

4 rT

-H

Tt

4+

41
IT -,--t

Ti I

4-

t-*t-

_7

-t-0444 4Tl

I I fill

ttT

4
J4-

UO

1. 0

_tit
0.5

I-i 14t-44-

7-

.25
-4

0.1 ;-i-.6

4
r/r

Fig. 2 5

F(7)

for

Ydrious

val,,es

of
TO

43

versus

r/r

fo r d

0. 5

III,

TILLT

IF

+4-4
T-T4-

45

-'

KTTh

...........

in
-- -i--t- r

r -N -t

T-t

N:7Tr
*t-

"11 -

44

5-TEH
-

-44

*I

11

$44

-H+

-.

-- 4

i$4

744

_1444

10
1. .

EIU

11-14
.

. .

1
.

..

-H
1
.

1
.

AS IA0

I I

RMi

4--4

IfI

If

I
If

..

Tk

..
I

,
I

I
I

I
I

If
I

T
T-N
>4

'.0

14

HEH
FT-FT45

HI

tf
JillL

'

zr,-4

if

LIit

1477

1r

-4

t1

1*

0L
I I I

0t

4 i-446

111<

-tt

4--4
wi-

~44-L~iU4I

- -,-;

. ! I-

+I

[Ill[
-

I
I4 I

'1

-- -

111UT-r

l l

1 1
11
I i
Il

"

t44-

II

I4k

i-I-m

4
-H

-4
-

[i

LA

1
--

L-4
l

LIN

UM

1-t

0rrT7

UII

44 40

TVI

'V

'I'

LiT

LN

-LL

InI

*T11

fa l 1

"0

II

tII

Iai

V-l_

4+0

In-

fill
fill

A II

..

_.4-L

&

If.
11

47

4!

- +%

000
0

0fil0

*1ji

IL

lf A11

11
II
II

IH

i 07 ,IIIIit

IM0

UIN

f i

l l

C;

1~ 11 yrT

,it

0H

at-~-

IIt.

JITT-

IN

f1:3

,0

]if

48

00

C;C

ll

0f-T

in

II

III I I I

mii
In*

II L

I.

ft

4-_

Nor

14

-I-F

Iri

-ne

C;

P-V

li I

VT

f*l, I

4-i

4
IL N

T-

-I

0-0--0

-N-

FU

0U0

LfN

II
.. fa

in
N

T441,

Ili

4-4-

IiA'I

1Itt4h
L
t,
-1

ti I~4

-T

_49

1.54n

I'D- -C
k~c
06

a.

.,49!

->b.-- >
.c 'o

Sj
a

I.

v 'm

a;~.

".6

I..

X CL'U 4

0.

Aj-a "

2a,

so

-4

0-;,

.4a~

m w c

r.' 0

:-

1. a:17Z

W~ a

.0~.

.4

10
-

AtS

I"

'"

4o

oo

0'.'

vs

00.c
-oJ

cUV a -

.S4N

2.

-0

ale
jV
d 4

0I
aU'V~~

b.u

S.

-6a

4'.

v54

Ia

-6

-I

&1a

0 0.

-6j
.5

4o

4'4j
&~-

r.

4)~.

0~
31,'--

*2
-6

~.

4,

'-

1 ..4~' 2. loS v

Is.

a a
0

4'

1.-

I0

:2
-, 'U

cc
:i

Zo Is

r'~

U~a
3

0Cc

tti.

I.~~6
4.

3..

1rb

I
I

b.

.0

9L.

4,O0*CuI.,
4.
c

1 0.
'~

0~UC-

--

-w
:F

'

09

0.

0 go
a

r-41

=-

hi c

0CV M. at
U
r

- J Q,.0

'o
0'-.
m

-4

"Cc

*~a

44

b. L.l -

X.

-A

0c

CIO

cr4.

aa

b.- vs

I"

fen*

V b..-

$.
U

0~

0U

~0l-t

U7a

U.

*V0.

4..
j96C

. r'i

-0

.C
a

1~

,.

'

4.4

LCI-CtI 0

Ji

uq-

.71

3-~~

dc

IV,

4a

M
0

10Cj

i b.

-u
-

40 1-.

0~~~.-~0

Aw

:1.
l-0

-ClX
Jo

a,~ul

C,

.2f

-4

>

ccr

64.t~
U0
aJ
*

414
4 C~.
0

*U.
lw04

-cl w94.C.

.
3-

4iC al.",

10

*
.

4.