Calhoun: The NPS Institutional Archive

Theses and Dissertations Thesis Collection

1964

Water jet air pump theory and performance.

Higgins, Hugh W.
Pennsylvania State University

[Link]
 School
U. S. Naval Postgraduate
Monterey, California
 4
The Pennsylvania State University

The Graduate School

Department of Mechanical Engineering

Water Jet Air Pump Theory and Performance

A thes is in

Mechanical Engineering

by

Hugh W. Higgi ns

Submitted in partial fulfillment

of the requirements
for the degree of

Master of Science

September \3&k
 j j

.Ida

ACKNOWLEDGMENTS

The Author wishes to express his appreciation to Dr. R. G.

Cunningham who suggested the subject, and under whose guidance this

thesis was prepared. The aid of Mr. J. Korman and others in the

construction of the apparatus is gratefully acknowledged. The Author

also acknowledges the use of the facilities of the Mechanical Engineer-

ing Laboratory.
 I I

TABLE OF CONTENTS
Page

Acknowledgments ,
;;

List of Figures , . iv

Nomenclature , . . V

I. INTRODUCTION 1

Statement of the Problem ...,,,... 1

Previous Investigations. ... I

II. WATER JET AIR PUMP THEORETICAL ANALYSIS 3

Derivation and Analysis of Equations 3

III. EXPERIMENTAL PROCEDURE AND RESULTS 10

Apparatus 10
Water Jet Air Pump. . 10
Water Jet Water Pump 11
Experimental Procedure 11

Results and Comparison with Theory 12

Nozzle Throat Spacing , 17
Observations of the Mixing Process 19
Comparison with Other Studies . 20

IV. DESIGN PROCEDURE 22

V. SUMMARY AND CONCLUSIONS 2k

Statement of the Problem , . 2k
Procedure of the Investigation 2k
Results , . . . . 2k
Conclusions , 26

BIBLIOGRAPHY 27

FIGURES 28

APPENDIX A: Apparatus Used as a Water Jet Pump:

Comparison with Other Studies kk

APPENDIX B: Sample Calculations 46

APPENDIX C: Test Data k$

 IV

LIST OF FIGURES

Figure No. Page

1. Jet Pump, Showing Mixing Process 28

2. Jet Pump Test Stand and Flow Diagram 29

3. Jet Pump Assembly Showing Nozzl e-to-Throat

Spacing, S 30

k. Nozzle Section Details , 31

5. Throat-Di ffuser Section Details , , . , 32

6. Rotometer Calibration 33

7-10. Head Characteristic and Efficiency vs. Flow Ratio

b = 0.2, 0.3, 0.4, 0.6 3k

11. Head Characteristic and Efficiency vs. Flow

Ratio., b = 0.6 for Pump at Optimum Spacing . . k0

12. Head Characteristic and Efficiency vs. Flow Ratio

b = 0.6 for Lucite Section Having Throat Length
L = 5.312 inches k]

13. Jet Pump Test Stand Showing Modification for Use as

a Water Jet Pump kl

\k. Experimental Pressure Ratio vs. Flow Ratio for

Water Jet Pump 43
 Table of Nomenclature

W = Mass flow rate, lb /min

m

P = static pressure, psf or ps

P = total pressure, psf

V = velocity, fps

J = density, lb /ft

A = area , ft

D - diameter, ft or in

L = throat length, ft or in

S = nozzle-to-throat spacing, ft or in

• ,
lb m ^
sc =
q dimensional constant 77
lb, sec 2
An
b - area ratio J7

W9 f\ Q9.
= flow ratio z _
t "HZ
' _£;
Q i

3
Q. = volumetric flow rate, ft /min

K, = nozzle coefficient

K = throat friction coefficient

Ki = diffuser friction coefficient

p
d
N = dimens ionl ess pressure ratio p _ p
n 1

I
£7^= dimens ionless pressure ratio —Ps
d

<
J~ - efficiency = P, N In o"L

ft lbf
E = power
sec

SUBSCRIPTS

1 = water

2 - ai r

m = mixture
 VI

m = mean

n = nozzle entry

s = throat entry

t = throat exit, diffuser entry

d = diffuser exit

= side flow entry to jet pump

fl = friction loss

ml = mixing loss

c = isothermal gas compression

 CHAPTER I

INTRODUCTION

A jet pump is a device that utilizes a momentum transfer action

from a high velocity jet fluid to pump another fluid, either the same

or different from that of the jet. The liquid jet gas pump is simple

in construction, has no moving parts, and is used primarily for pump-

ing gases containing some condensable vapors, such as water vapor,

solvent vapors under a vacuum, or others which would be harmful to

mechanical pumps.

Statement of the Problem

The jet pump, although in use for many years, has received

relatively little attention in the technical literature. The majority

of papers have considered either the case of compressible flow such as

the analysis by Keenan and Neumann (1) , or the case of incompressible

flow such as the analysis by Gosline and O'Brien (2). Therefore, the

objective of this investigation was to develop a theoretical analysis

of the liquid-jet gas pump and to evaluate the theory by comparison

with experimental results.

Previous Investigations

The water jet air pump has been investigated by Flugel (3),

Pawel -Rammengen (k) , and others, and some performance data have been

reported by Folsom (5). In this latter study Folsom (5) concluded

that the performance of the jet pump should be dependent on diffuser

characteristics, however in all the studies listed above, no satis-

factory analysis of the pumping system was made.

-Numbers in parenthesis refer to the Bibliography.

 The most recent study on liquid jet air pumps was reported by

Takashima (6), wherein data were obtained for a jet pump in a water

circuit using two-inch pipes. The conclusions of this study were

based on assumed values for the nozzle discharge coefficient, throat

friction factor, and diffuser coefficient.

 CHAPTER I I

WATER JET-AIR PUMP THEORETICAL ANALYSIS

Derivation and Analysis of Equations

Although the details of the mixing process of the water and air

are highly complicated, application of the momentum theory permits a

solution by treating the initial stream before mixing, and after mixing

has been completed. As shown in Figure 1, the water jet penetrates the

slowly moving air, and as mixing occurs momentum transfer accelerates

the air in tha throat. As the flow of air and water progresses the

mixture stream spreads until at the throat exit, it touches the wall of

the throat.

The following assumptions are made:

(1) The flow streams are one-dimensional at throat entrance and

exi t

(2) fixing is completed in the constant area throat, against an

adverse pressure gradient.

(3) Suction chamber pressure is constant throughout the chamber,

that is, the pressure at the nozzle tip is equal to the pressure at

the throat entrance.

{k) Isothermal conditions prevail in the throat-di f f user section

because of the relative heat capacities of air and water.

p
(5) Air acts as an ideal gas, that is 2 = rt , and water is
2
T>
incompressible, J
= constant.

Equation of Motion for Liquid Gas Mixture

Assuming negligible change in elevation and steady flow conditions,

the equation of motion, including a friction loss term, can be written

 dP + JU)£ + f dx 1_ = (I)

Where: f = friction coefficient

/^ - mixture density defined as

W + W
A_ _J 2 subscripts and 2 refer to 1

^^ Qi + Q9 liquid and gas respectively.

Let the volume ratio £> =

/
—
Q 2
, and neglect the mass of gas W
2
then,
<il

Q, (1 + (p ) 1 +0
Let ^ represent Q? y , the flow ratio at pump inlet, and assume

that the gas phase behaves as a perfect gas under isothermal

conditions then,

A !&
P
(2)

Within the isothermal flow tube density is a function of absolute

pressure only.

Equation (1) can be integrated from a to b using Jm

'.

P
b
- P
a + V - V/ + /
f " 2dx
- (3)

in which the mean density is defined by

P. - P P. - P
b a _ b a
^m ^DdP 75 , i P

^ =
P

I
b-
[P +Cj>
P

s
P
a

s
In P]
b
a
- ^ P
b-
(P
b-
P
a
+
P
a>

<Ps
P
s
,nP
J>
(4 >

6^1 P
Thus equation (3) becomes:
2 b 2
P - P &
"s P
In P u V.
2
- V (
f V dx ==
(5)
b a , s b
P~
+
j.
—2g
_\x a +
J
a
uTg"
</l </l

Diffuser Equation

Applying equation (5) to the flow from the throat exit (t) to the

diffuser exit (d) ; and basing the friction loss term on the density and

velocity at t:

\ ^V
2 2
P + fj£ . p .
</,V
t
-
t
-
^> S P S
In P,
(6)

The friction factor K, corresponds to f-jr for flow in a pipe. It is

convenient to replace the first two terms with P ,, a total head

expression. Although this is approximate (through use of o^\ instead

oft/)
d
) the
t error is negligible since the kinetic head at the pump

discharge (d) is small.

The throdt and nozzle velocities are related by continuity

V = b (1 + <fi t
) V
t Is (7)

Where b= J. and ^ = At = £% P
(|sotherma]
t ^ t flow)

The diffuser equation can now be rewritten as follows:

/? 2
~
P
d
' P
t
=
~lP-
c
[b2(1 +
S^t
)2
" V 2(I +
^t )] '^s
P
s
,n
^
Pt
(8)

The Throat Equation of Motion .

Equation (5) can be applied between the throat inlet (s) and

exit (t) . In addition to friction, the throat equation must include a

second loss term, E ., which is required for the irreversible mixing

losses that accompany the entrainment of gas by the liquid jet.

P
t£ lnP
~7V~ ~7T
- P
s +
P
s

p
t

s
+
V
t

2
'"is
^c
Xh fV d *

D2 9
c
+
E
ml =0 (9)

For the liquid/liquid jet pump it was shown in reference 7 that the

mixing loss is of the form

2

"ml !i£ [f(b,C2$

w )]
2g
c

where (X> is the mass ratio of secondary liquid to primary liquid

stream.

This relation is obtained by writing both the momentum equation

and the equation of motion across the jet pump throat, and eliminating

the pressure
K difference term, P^ - P . It can be shown that the
t •
s

expression reduces to the following form when (p =0, that is, when

the secondary liquid flow is zero:

2
V
E
m1
= 1»- [1 - 2b +b 2 ] (10)
yc

This will be recognized as the usual "sudden enlargement loss"

for a pipeline, where b = _a, the ratio of pipe area before and after
Ab
the enlargement.

Just as the gas phase energy input term at (s) can be neglected

due to the small magnitude of W„ compared to W. , the useful gain in

momentum by the gas phase in the mixing process will be neglected,

permitting use of this sudden enlargement loss in equation (9):

2 2 l 2 2
P -P d) P lnP f V -V, ( fV dx, V (1 - 2b + b^O (11)
t s + /s s t + t 1 s + \ + Is
'

~7T ^T P
s
29
c > D2
9c
29
c

With the friction loss again based on density and velocity at station

(t) , and using the continuity equation (7), the throat equation (11)

becomes:
P -P
s
=
% ]S

c
[2b-b
2
(l +
^ t
)
2
-b -K b
2
3
2
(l + <j6p-j6
s
P
s
In p^
'
(12)

The diffuser and throat equations can be added to obtain:

(13)
9
c s

Where K,. = K_ + K, , for the throat and diffuser.

Nozzle Equation .

The primary or nozzle fluid is of constant density thus: ,

2 ? ?

^ 2g
c ^ 2g
c
2g
c

where the final term represents the friction losses.

Let P = P + c/^
^ Vn . the total head, thus
n n 1

2gr
c

2
+ K.)
r e^V,
P - P = (1 lM
n s " 1 s
,
(14)
2g3
c

Overall Pressure Drop Equation.

Subtracting equatipn (13) from equation (1^),

P
n
-p
d
- ^ V)s [1 + K,-2b + b
2
+ K
4
b
2
(l +^ t
)] +^ s
P
s
ln ^d
P
(15)
2g c

Jet Pump Efficiency.

Assuming isothermal compression of the gas phase in the throat

and diffuser, the useful power out is:

P P
E = W d ln d
out 2
A, ~s

Since for all practical purposes P ,

= P ,
 p
d

Efficiency )/ =
"2/1 r
d '" F7
1

(16)
P
d In d

« n
n d s

P,
With N representing the first term and o(^= —
s

^ = ^>d N ln^ (17)

Significance of the In Terms.

The term r ( )
s s In __b appears in the equation of motion when

the mixture density varies i sothermal ly. This represents work

(ft lb, /lb liquid) by the liquid on the gas phase and as such is an

energy loss to the liquid phase. Note that the term subtracts from

the pressure difference just as the friction loss term does.

Viewed from the gas phase, an observer moving with a gas bubble

would see the bubble change in diameter, and the work performed would

iJz? -
w> \
)•
p
^ - 4r
<?
ln
P
a
ft
lb
m
lb f
9 a5

for isothermal conditions, where cy \s the gas density. The corres-

pondence of this term to the "loss" term in the equation of motion is

obvious

The first law of thermodynamics applied to the isothermal ly

compressed bubble shows that a quantity of heat must be transferred

out of the gas, which is equal in magnitude to the work term.

This heat is received by the liquid phase and a temperature rise

must result. Because of the relatively large heat capacity of the

liquid this temperature rise is negligible. Note that equations of

motion, rather than energy equations (which would include internal

energy and heat terms) are adequate and appropriate for this analysis

The isothermal work of compression term, however, stems from the gas-

phase energy equation.

It is interesting to note that the expenditure of mechanical

enrgy of the liquid as work on gas bubbles, and the necessary return

of heat energy to the liquid corresponds in net effect to fluid

friction, that is mechanical energy is degraded to heat energy,

appearing as a liquid temperature rise.

 CHAPTER I I I

EXPERIMENTAL PROCEDURE AND RESULTS

Apparatus

Water Jet Air Pump

A schematic illustration of the experimental water jet air pump

and the auxiliary apparatus is shown in Figure 2. The jet pump used,

as shown in Figures 3, 4, and 5, was designed to facilitate investi-

gation of the variable area ratio b, and spacing S, and was patterned

after a pump used in a previous work by Cunningham (7). It consists of

a one inch pipe tee body, with two interchangeable nozzles and two !

throat-dif fuser sections. Therefore four jet pumps could be assembled

in one body. All pumps are identified by three numbers; e.g. No.

141/316/300, the numbers referring to nozzle diameter, throat diameter

and nozzle-to-throat spacing in thousandths of an inch.

In addition to the two throat-di f fuser sections previously

described, another throat diffuser section with throat diameter 0.224

inches was made using transparent lucite in order to observe the flow

patterns in the throat and diffuser.

Water was supplied to the nozzle from a 130 gallon tank thrpugh

a centrifugal pump, a one-inch throttle valve, and a rotameter.

A calibrated bourdon tube pressure gage was used to measure

nozzle pressure. Air was taken in the top of the pump through a

precision wet test meter, a glass water trap, and a throttling valve.

The vacuum in the pump body was measured with a U-tube mercury mano-

meter, also containing a water trap. The pressure at the diffuser out-

let was measured with two calibrated bourdon tube pressure gages,
inserted upstream of the back pressure valve. Following the back

pressure valve, the water-air mixture was discharged to aweigh tank by a

flexible hose.

Water Jet Pump

Before the water jet air pump performance data were taken, the

apparatus described above was modified to act as a water jet water

pump in order to obtain data to compare with well-established perform-

ance data previously published by Gosline and O'Brien (2) and

Cunningham (7). This modification consisted of disconnecting the air

supply to the pump, and connecting a one-inch water supply line through

a throttling valve from the pump to the supply tank as shown in Figure

13,

Experimental Procedure

All of the tests were made at a selected nozzle flow rate, W. , and

suction chamber pressure, P , The diffuser discharge pressure P , was

varied from the "cutoff" pressure where the air flow rate (L equals

zero to the maximum obtainable, where the back pressure valve was wide
3
open,
r while maintaining3 P , P , W, constant, air flow rate 0_ o in ft /
n s 1 2

min was recorded.

The rotameter used to indicate primary flow rate W. , was calibrat-

ed as shown in Figure 6, and served only to show steady-flow conditions

during a run since the mass flow rate W. was determined by timing the

pump discharge into a weigh tank..

The five measured quantities were then recorded and converted to

the dimens ionl ess flow and pressure ratios N,o^and (p , as will be

shown in a later chapter.

 12

Results and Comparison With Theory

An initial check on the apparatus was made by modifying the water

jet air pump to function as a water jet water pump as described

previously. The data obtained compared satisfactorily with data

published by Gosline and O'Brien (2) and Cunningham (7) as shown in

Appendix A.

Values for the nozzle and throat-di f f user coefficients, K. and

K_. respect ivel y, were computed from test data taken at a flow ratio

approximately 2/3 of the maximum-efficiency flow ratio. These values

were found to be K. = .18, and K_. = .051. These compare with

K. = .10 and K_. = .30 which Cunningham found at high Reynolds

numbers using oil as the fluid (7).

The apparatus was then converted to function as an air pump and

tests made. From these data the dimens ionless flow ratio G£> , was

computed and plotted as a function of thetwo dimens ionless pressure

ratio's N and c^as shown in Figures 7 through 10 for various water

flow rates and area ratios. Performance of the theoretical model is

shown on these curves as a solid line. Numerical results are listed

in the test data section, Appendix E.

The relation between P - P, or P, - P and [V, , b, K. , K_, , U> ,

n d d s Is I ,3^ ' s

(£) +» (pA>] ' s °f interest as a means of constructing an N In ^ vs

(p , curve. Favorable comparison with test results would validate the

treatment

Obviously there are too many unknowns to permit direct calculation

of K's from the test data. This would be possible only if P were known.

(Folsom pointed out the need for P when considering the possibility

of a theory-experimental data comparison.)

 13

The friction coefficients can better be measured with liquid/

liquid flow or taken from previous works (Reference 7). Nominal (high

Reynolds number) values of K.= 0.1 and K-.= 0.3 will be used.

Equation 13 can now be solved directly for d) and hence P ,

using test data, thus giving one use of the theory. Rearrangement of

equation 13 yields:
p
d

U.
r ::
2 _± ^s P
fs s
ln P
s V d
P
s
t K,,b "''k,, Ktt-JTT-* mV.v.J (18)

2
\ 2gc

Sample Calculation:

Pump Number - 173/224/120

Barometer - 28.57 in Hg

Air Temp - 83.5°F

Water Temp 53°F

3
0_ - 0.121 ft /min at 14.03 psia
2

W. - 71 lb /min
1 m
P = 99.0 psig Corrected)

P -5.8 in Hg or 11. 18 psia

P
d
= 60.0 psig (corrected) ~ P = 74.03 psia

N = Pd = 7^03 m , g
P -P . 99.0-60.0
n d

P 7Z+ - 03
/ d
0\ = 7T-
P
= VT-To
,, ,8 •
= 6.64
s

W
^ =
w
2

1
^J
^2d
= 12.22 ^2 = 0.02
P
d

& =
<t>A
Pd =-02 7^°3 = 0.132
s
 14

K_. =0.3 (assumed)

W 71
v = 10-**
= 116 ft/sec
<7^ 60(62.4)1.631 x

2 2
c/,V ls 2 = 62.4 (116) = 13,040 lb /ft
f

2g 64.4

Substituting in Equation 18:

M t
- 2 - 1
-_J_ - 404 - 9050
I (0.3) (0.6) 0.3 (0.3) (0.36) (13040) (0.3) (0.36) 13040

0-09
(fi t

P =
t ^s P s = ( 0.132) (11.18) = 16.4 psia
j2T~ 0.09

Thus the theory provides a means of finding the throat pressure

for measured performance.

Construction of Theoretical Characteristic Curves

Equation 13 could be used as a means of constructing N lnov and

uj , N In «^\ (or efficiency) curves if the throat volume ratiora

could be eliminated as an independent variable, The measured value of

(f) .
(above) is close to the average value,
+
(t>
rt avg = s ^d = 0.132 + 0.02 = 0.076
2 2

Suggesting the approximation:

0, + #.
A -
2 (19)

with K-. 0.3 and the above approximation for y9 equation 13 can be

solved for Q) :

P In P ,

s d
P " P
?: 2 d
1 +-§.
V s
1
L. + (1
2
) " ' "
^4" K b2 ^V 1s 2 ^0)
V'^h' 2 g.
d 34
2g^
 15

Sample Calculation :

Find (P for P. = 60 psig 74.03 psia with P = 11.18 psia and

all other data from the above sample calculation.

[2.17 + 0.575] = 11.1 - 1 - 3.33 - 6.41

<f> s

Similar calculations for other P , values result in the following:

Assumed P ,
Theoretical <p Theoretical
d

74.03 0.131 0.02

54.03 1.00 0.207

4o,.03 1.82 0.51

30, 03 2.68 0.996

20, ,02 3 . 82 2.13

15. .03 5.28 3.93

11 .18 6.78 6.78

The theoretical curve construction can be completed by solving

equation 20 for P ,-P at (p =0. Here the right side must equal zero,

and this shows that P ,

= 77.48 psia, therefore,

M
N ,n
, J
<^ 77.^8 .

,n
77.48
TTT8
s
.
99
k 22'

113.03 - 77.48
The experimental data points are above the theoretical curve at

low flow ratios for area ratio b = 0.2. This was assumed to be due to

the failure of the flow to conform with the throat wall which in turn

could reduce friction, as was originally suggested by Cunningham (7).

Referring to figures 7 through 10 it can be seen that the actual

pump performance coincides with theory up to a certain point, and then

falls away rapidly. Figures 7A and 10A are complete characteristic

curves for figures 7 and 10 respectively. The point at which this

 16

rapid decline occurs coincides with the point at which mixing no longer

occurs in the throat. This is shown in the accompanying flow diagrams.

Since the theoretical calculations were based upon complete mixing in

the throat, they can not be expected to predict actual performance when

jet breaking occurs in the diffuser. It can be seen that when the

water jet penetrates the entire throat length and does not merge into a

water-air mixture until well into the diffuser section, the mixing loss

is magnified resulting in a marked loss of efficiency.

The theoretical calculations were derived in Chapter II assuming

that the nozzle tip coincided with the throat entrance, that is the

pressure at the nozzle tip was assumed equal to the pressure at the

throat entrance. In practice, however, the nozzle tip is withdrawn

from the throat a distance S in order to obtain optimum performance,

and the behavior of the water jet across this distance S could be

thought of as a fluid friction energy loss as described by Cunningham

(7). This friction loss was calculated and found to be negligible since

it was dependent on the density ratio of air to water.

This comparison between theoretical model and actual performance,

coupled with visual observations of the change in mixing patterns has

shown that (a) the theoretical approach is valid and (b) maximum

efficiency demands that the liquid jet gas pump be designed and/or

operated to provide jet break-up before entry into the diffuser.

Approximate Flow Ratio at Maximum Efficiency Point .

McElroy (8) observed that the flow ratio at which maximum efficiency

occured coincided as a rule with the state of equal momentum at the jet

and at the throat discharge.

ie: W„ V = W. V.
t t Is
 17

in this case

Vt^Vu
V
]s
b (1 + f t
) = V
Is

mep b

For comparison with the water jet air pump, consider the pump with

b = 0.2. The above expression yields <Z) = k. With the assumption

that d> = ( <p + £J )/2, we find that S^ = 6 -9- A similar

d

calculation for b = 0.6 results i n (f)

d = 1.21. Comparison with
mep

Figures 7A and 10A show - k.O and 1.0 respectively, a fairly

(2),
mep
good check. Apparently this approximate rule of thumb is useful for a

water -jet air pump.

Nozzle Throat Spacing

The experimental jet pump as shown in Figure 3 was designed not

only to observe the effect of area ratio b on pump performance, but

also to study the effect of nozzle-to-throat spacing, S. The nozzle

was held in place in the pump body by a nut and 0-ring seal. By

loosening the nut the nozzle could be moved in and out to change the

spacing while the pump was in operation. Scribe marks were placed on

the nozzle sections at 0.1 inch intervals.

The optimum value of S was found by initially setting the spacing

at one nozzle diameter, and then setting the operating flow ratio Cl) .

equal to approximately two thirds of the value of that obtained at

the maximum efficiency point by adjusting the back pressure P ,. The

water flow rate W. and the suction port pressure P were held constant.

Optimum spacing was then found by moving the nozzle section in 0.1 inch

steps to obtain maximum air flow 0_2 while still holding Wj , P , Ps ,

 18

and P , constant,
d

The results are listed in Table I below:

TABLE I

OPTIMUM NOZZLE - THROAT SPACING

Nominal Area
Ratio b Pump No S I nches S/d
n

0.2 141/316 >1 .92 >13-6

0.3 173/316 >1.92 >11.1

0.4 141/224 0.2 - 0.4 1.42 - 2.84

0.6 173/224 0.65 3.75

The optimum spacing S is listed in addition to the ratio of

optimum spacing to nozzle diameter, S/d . Where more than one value

of S is listed the air flow rate remained essentially constant over the

range of S values. In other words spacing had a negligible effect on

efficiency is the range indicated.

For the nominal area ratios of 0.2 and 0.3 no optimum is listed

since the air flow Q_ increased with increased spacing up to 1.92

inches, the physical limit of the pump.

The first series of tests for all four area ratios were completed

using the spacing S recommended by Cunningham (7) for the liquid-

liquid case. The experimental data were then used to plot performance

curves as seen in Figures 7 through 10, and the information from these

curves was then used to select the flow ratio (Z> to calculate the

value of optimum spacing.

The S/d values listed in Table I shqw an i neons stancy which

i
'
is
n

probably due to an interaction with the jet break-up location, that is,
 19

as spacing was increased the jet break-up receded and influenced the

efficiency. Pump number 173/224/650 (b = 0.6) is the best guide to

optimum spacing since the value of CP', used to determine optimum

spacing showed least divergence from theory.

The most important conclusion to be drawn from this section is

that pump spacing should be set approximately five times the value of

optimum spacing listed for the same pump for the liquid-liquid case.

Observations of the Mixing Process

Observations of the mixing process in the transparent lucite

throat-di f fuser section indicated that for low back-pressure operation

the water jet penetrated the entire length of the throat and part of

the diffuser before merging into the water-air mixture. As the back

pressure was increased the break-up of the water jet receded until

mixing was taking place in the throat as it should be. The flow pat-

terns are shown in Figures 9 and 10.

Comparison of the flow patterns and performance curves leads to the

conclusion that optimum performance is obtained when the water jet

penetrates most of the throat length, and then the water-air mixture

merges to fill the diffuser entrance.

In order to accomplish mixing at the diffuser entrance, section t,

it was concluded that either (1) the nozzle tip should be withdrawn to

increase S or, (2) the length of the throat should be increased.

The first case was explained earlier in this section, and optimum

values of S were found for area ratios b = 0.4 and b = 0.6. The

performance of the b = 0.6 pump for optimum (S = 0.650) is shown in

Figure II. Comparison with Figure 10 (where S = 0.130) shows that peak

efficiency for the optimum spacing was somewhat better and the
 20

efficiency was higher over a wider flow ratio range.

The second case was tested by constructing a special lucite throat-

diffuser section identical in detail with the first lucite section, with

the exception that the throat length was increased to 5-13 inches.

Performance curves for this section are as' shown in Figure 12. Mixing

of the water and air took place prior to the diffuser entrance for the

entire flow ratio range with the exception of the high back pressure

portion just prior to "Cutoff" where air flow goes to zero. Here the

flow pattern, as sketched in Figure 12, has an unsteady appearance

similar to cavitation. This phenomenon occurs over the range of (D .

values that include the narrow peak on the efficiency curve, and may

explain this sudden change in performance. In this area the back

pressure has little effect on the flow ratio; e.g., an increase in

the back pressure from 19 psig to 23 psig decreased the flow ratio

from 0.0736 to 0.0556.

Comparison with Other Studies

In general, these data compare satisfactorily with performance data

reported by Folsom (5). The water jet air pump used in Folsom's study

had an area ratio, b, of 0.^5 and the primary and secondary flow rates

were larger because of the difference in the size of the equipment.

Despite these di f ferences the performance curves show the same trends.

Peak efficiencies of about 16% were reported by Folsom. The maximum

efficiency obtained in the present study was about 13% for an area

ratio of b = O.k.

Takashima (6) observed that as the diffuser section back pressure

increased the solid jet stream into the diffuser section backed up until

complete break-up was accomplished in the throat. In addition he

 21

observed that nozzle-to-throat spacing had little effect on pump

efficiency. As shown in Figure 10 the flow patterns observed in the

present study tend to corroborate Takashima's findings. However, as

shown in the previous section the nozzle-to-throat spacing does have an

effect on pump efficiency.

 22

CHAPTER IV

DESIGN PROCEDURE

Water jet air pump performance can be described by the water and

air flow rates, and the nozzle side port, and discharge pressures. The

pump itself is described by the nozzle and throat areas, a and A

respectively. In addition the following dimens ionless flow and pressure

ratios are used (ZA, N ando<^which are defined in Appendix B.

Design Example :

A water jet air pump is to be designed knowing the discharge

pressure and the air flow rate, with high flow ratio required,

Given: W- , P ,

l d

Find: W. , P , b, s, a, A
1 n

Since (7), at maximum efficiency decreases as b increases, the

smallest area ratio used in this work, b = 0.2, will be selected in

order to attain the desired "high flow ratio" characteristic.

Inspection of Figure 7 shows that high efficiency and the limit

of agreement with the theoretical solution occurs at £Z?, = 0.24.

With this value of (p , where N 1 n cA^ - 0.37, efficiency ( is equal to

(T) , Nlnc?\= 8.9 per cent. Nozzle flow rate Q. can be found from

(T) - 2s from which W. can be calculated using W. = /^iO,

,
(/ i

is a constant.) (p can be found using (7) P = 0J d 3

- Values of K. and

K_. can be assumed to be 0.1 and 0.3. S ince P is usual ly near

atmospheric the ratios of equations 13 and 15 will yield a value of N

which in turn can be used to calculate P . Nozzle area, a, is found

n

from: 9

p
n
- p
r
s
—J
= \"

g
Is r,
+ K]] {]k)
 23

and the continuity relation

v
i *n ls

The throat area, A, is found from b = 5..

A
Nozzle-to-throat spacing is found by referring to table I
 2k

CHAPTER V

SUMMARY AND CONCLUSIONS

Statement of the Problem

The purpose of this investigation was to develop a theoretical

analysis and compare this with experimental results, and to determine

the effects of nozzle-to-throat section area ratio and nozzle tip to

throat entrance spacing upon pump performance.

Procedure of the Investigation

The theoretical analysis developed in Chapter II was based on

integration of the differential equation of motion, treating the density

of the liquid-gas mixture as a variable. The theoretical treatment is

based on the use of K. and K_. coefficients measured with liquid/liquid

flow condi t ions.

The nozzle pressure was varied from 10 to 100 psig, the back pres-

sure flow to 71 psig, and the suction chamber pressure from 14.03 to

[Link] rpsia. The flow rate of water was varied from *+3.3 to 71 lb /min,
m
and that of air from to 1.075 cu. ft. /min.

Resu ts
1

A function of the dimens ionless pressure rat ios //and e^was plotted

as a function of the flow ratio 0., as shown in Figures 7 through 10.

Numerical results are listed in the test data section, Appendix C.

The actual pump performance coincides with the theory, shown as a sol i d

line in Figures 7 [Link] 10 , up to a certain point and then falls away

rapidly. The point at which this rapid decline occurs coincides with

the point at which mixing no longer occurs in the throat, as shown in

the accompanying sketches of the mixing patterns. Since the mixing

 25

occurs in the diffuser and the mixing loss is directly proportional to

the area, the mixing loss increases rapidly causing a sharp decline in

efficiency.
 26

Concl us ions

(1) The performance of the water jet air pump can be satisfactorily

predicted by the equations developed in Chapter II.

(2) Comparison of theory and experiment plus observation using the

specially constructed transparent throat-di f fuser section has

shown:

(a) Flow ratfl'o (l), for maximum efficiency decreases with

increasing area ratio b as predicted by theory.

(b) Maximum efficiency is obtained when jet break-up occurs in

the throat. Hence the pump should be designed and operated

to maintain this break-up in the throat.

(3) At an area ratio b = 0.6, using an unusually long throat, an

unsteady "front" resembling cavitation occurred in the throat.

Under these conditions pump performance agreed well with theoret-

ical predictions. When the cavitation zone moved out into the

diffuser entrance, actual efficiency decreased rapidly below the

predicted value.

(h) Although it is possible to calculate P , direct measurement would

be highly desirable in any future work. This measurement would

permit direct calculation of friction coefficients from experi-

mental data.
 27

BIBLIOGRAPHY

1. "A Simple Air Ejector," Keenan, J. H. and Neumann, E. P. Journal of

Appl ied Mechanics vol. 69, 19*+2.
,

2. "The Water Jet Pump," Gosline, J. E. and O'Brien, M. R, University

of California Publications in Engineering vol. 3, no. 3, 193*+,
,

pp. 167-190.

3. "The Design of Jet Pumps," Flugel, G. NACA T.M., 982, 1939.

k. von Pawel -Rammi ngen, G. Doctors' Dissertation, Braunschweig, 1936.

5. "Jet Pumps with Liquid Drive," Folsom, R. G., Chemical Engineering

Progress vol. kk, no. 10, 1948, pp. 765-770.
,

6. "Studies on Liquid-Jet Gas Pumps," Takashima, Y. Journal of the

Scientific Research Institute (Tokyo), vol. k6
,
, 1952,pp. 230-
2k6.

7. "Jet Pump Theory and Performance with Fluids of High Viscosity,"

Cunningham, R. G., American Society of Mechanical Engineers
Transact ions Nov. 1957.
,

8. "Design of Injectors for Low Pressure Air Flow," McElroy, G. E.

Bureau of Mines Technical Paper G78, 19^5-
,
 23

CO
CC CO
O
o Lu
O
> O
5: H
< o Q-
UJ
(X
o
-J
h- Uj
oo >
O
UJ
1Z
>< iD

o
a:
CO
cr a.

ID
> Dl
<
D LU
Z
o
o
UJ

>"
 29

MANOMETER MANOMETER. (f|)

THERMOMETER
X
1

/V ROOM AIR IN

PRECISION WET
TF5T METER fo 2)

•* — THROTTLE VALVE
-®n
WATER TRAP
A A WATER TRAP

i£
GA6E ( Pn)(/ s~~\ GA6E(^j)
j—
back pressure
COfy/TROL VALVE
-* ®
JET PUMP

WATER
SUPPLY
ROTAMETER (vujV
TANK
PUMP

TO DRAIN

FIG.2 JET PUMP TEST STAND

AND FLOW DIAGPAM
30
 31

i_ HCO

CO
_J
<
CO Q
LU o
a.
o_
o
LU
CO O
o LU
00
C5
-J
Nl O
Ld
O CO
LU
CD oo LJ
< o
^ *
N
T INI
V7//
2
O
o « •

«
*)
o
5*
 32

CO

v\
<
Z) I-
^0 Ld
1

C3
^
IS *0
Q Ll
ro C^
O
CD 00
* t

o oo tN
O
LU co C"-.

U IX ^f- LU
Id Id CO
LU K) 00
oa —
rD
D CM cv lu
«
UL
Ll.
CD ^>
Q -J CD Csj
C£>
00• — Ll.

<>h CD
DC * <M «

-o CM
% •
<
o
^*-
CO
i ro
h
o

Li-
 33

o
/c

9o

80
CALIBRATION

. 70 READING

. ]^ GO

-
SCALE

50

ROTAMETER

40
ROTAMETER

JO

6
20

FIG.
/0

<
d ti> to ^p to cm i: o
Niw/ai 31Va r\Q-U Q3/\83S80
 3^

EXPERIMENTAL
0.7
O — Nln«X

see fig. 7A for complete

characteristic curve.

o
u
THEORETICAL HEAD CHARACTERISTIC
u
1-4
Eh
O

w
Eh
o U
z
H
u
1—4
u
Q
<

0.1 02 Q3 0.4 0.5 0.6 0.7 08 0.9

FLOW RATIO, (J)

Fig. 7 Head Characteristic and Efficiency

vs Flow Ratio, b = 0. 2.
 35

ft

-20 £-

u
w
I—
u

10 15 20
FLOW RATIO, $

Fig. 7A Head Characteristic and Efficiency

vs Flow Ratio, b = 0. 2.
 36

0)
o
u
<u

!*
U
a
u
I—

0.2 0.3 0.4

FLOW RATIO, (J)

Fig. 8 Head Characteristic and Efficiency

vs Flow Ratio, b = 0. 3.
 37

EXPERIMENTAL
0-Nln©<

4->

u
u
0<

O
1-1

0.2 0.3 0.4

FLOW RATIO, <$>

Fig. 9 Head Characteristic and Efficiency

vs Flow Ratio, b =0.4.
 38

EXPERIMENTAL
O — Nlno<

see fig. 10A for complete

characteristic curve.

U
l-H

w
H
u
<

X
u

0.2 0.3 0.4

FLOW RATIO, <j)
d

Fig. 10. Head Characteristic and Efficiency

vs Flow Ratio* b = 0. 6.
 39

X
c

u
l-H
H
en

w
H
u
<1

u
Q
<

2 3 4 5
FLOW RATIO, (j) ,

Fig. 10A Head Characteristic and Efficiency

vs Flow Ratio, b = 0. 6.
 ko

U
t—
H
i—

w
H
u

u
Q
<
w

02 03 0.4
FLOW RATIO,

Fig. 11Head Characteristic and Efficiency vs

Flow Ratio, b = 0. 6. (For pump at optimum
spacing.
 k]

PUMP MO. 173/224/130

PN * 1)2.3 PSIA

a 02 0.04- 0.06
FLOW RATIO (J>d

FIG, I 2 MEAD CHARACTERISTIC AND EFFICIENCY

VS. FLOW RATIO b=0.6 FOR LUCITE
SECTION HAVING THROAT LENGTH
L» 5.312 INCHES
 42

MANOMETER (rJ

T.

WATER TRAP

I
^
THROTTLE
VALVE
e-
GA6«F (PN
)

6A<SE (Pj)

i &
0ACK PRESSURE
JET PUMP CONTROL VALVE

WATER
ROTAMETER? M pump
Supply
tank

? ^- TO DRAIN

FIG. 13 JET PUMP TEST STAND

SHOWING MODIFICATION FOR
USE AS WATFP JET PUMP
 43

PUMP NO. I4l/3I6/^I50

P. = 114,2 PSIA

30.0 -
OS

X / '

24,0 0.4

H
£ Q?
O i 1

CL
LlI Q? a?
0_
/8.0 - O0.3
1-
f> <
a:
>-
o Id
Z (V
iJJ ^ *«
—
o
.~
^.0 - w0,l A- EFFICIENCY
in
ti- LU /
(V
ll.
/ 1

Ld Q.
/
/ CAVITATION y
^|
Co - 0.1

r i

O.S hO
ill A 5*
FLOW RATIO -^-
w,

FIG. 14- EXPERIMENTAL PRESSURE RATIO AND

EFFICIENCY MS FLOW RATIO
FOR WATER JET PUMP
 kk

APPENDIX A
 hS

Apparatus Used as Water Jet Pump:

Comparison with Other Studies

As described in Chapter III the experimental apparatus was

modified to act as a water jet pump in order to measure experimentally

the friction coefficients K, , K_, and to permit a comparison with prior

studies. See Figure 13*

The tests were conducted by holding constant the primary flow rate

and the suction chamber pressure while the back pressure was varied

from the "cutoff" pressure (where the secondary mass flow rate equals

zero) to the pressure producing the maximum flow obtainable, ie:

where the back pressure valve was wide open. At each point nozzle

pressure, suction chamber pressure, diffuser discharge pressure,

primary mass flow rate and secondary mass flow rate were recorded.

The ratio of secondary to primary mass flow rates was then plotted

as a function of the ratio of the back pressure-suction chamber pres-

sure difference to nozzle pressure-back pressure difference as shown

in Figure \k for jet pump 1^1/316/350. Similar results were obtained

for other area ratios and will not be shown. Numerical results are

listed in the test data section, Appendix C 4

The approximate theoretical approach used by Cunningham (7) was

used to calculate values for the nozzle and throat-di ffuser constants

K. and K_. respectively. With K. = .18 and K_. = .051 calculated from

the experimental data a satisfactory comparison was made with Figure

\k (7), that is maximum flow ratio (2) and zero flow ratio N were

calculated for the area ratio used (b = 0.2), and found to be N = .*+77

and (P = 2.6. These compare with N = .kf and (D = 2.5 taken from

Figure ]k (7) .
 ke

APPENDIX B
 47

Sample Calculations

The tests were conducted by operating the pump at a fixed primary

flow W. and fixed suction port pressure P . The discharge pressure

P , was then varied from zero where the air flow rate was a maximum to

the "cutoff" pressure where the air flow rate went to zero. The water

and air flow rates, and the three pressures

r P , P , and P , were
n s d

recorded for each run. These data were then used to calculate the

pressure ratios N and^/^and the flow ratio (p , as shown below. The

water flow rate was set by rotameter and then measured by weighing

against a stop watch. The air flow rate was measured by a precision

wet test meter.

Test Date: 4/13/62

Pump Number: 173/224/120, b = 0.6

Barometer: 28.57 in Hg corrected

Air Temperature: 83.5°F

Water Temperature: 53°F

3
0, = 0.121 ft /min (at 14.02 psia)
2

W, = 71 lb /min
1 m
P - 99.0 psig corrected for gage calibration error

P -5.8 in Hg or 1 1. 18 psia

P ,
= 60.0 psig corrected for gage calibration error

^ = _d
p

P
. 74.03
11.18
= 6 .64
s

P
N - A ,

r 74.03
P -P 99-0 - 60.0
n d
 48

N In o^= 3.6

=12.22 Q2 =0.02
r w ,-4
d
P
d

:
0.02 (7^.03) = 0. 132
(11.18)
s

The calculated N ln<S7\values were then plotted against flow ratio

&s , and a smooth curve faired through the points. An efficiency

curve 19 = CO N lne^-was plotted by multiplying N lno^values from

the curve by the corresponding value of /^,-

 49

APPENDIX C
 50

Date 4/10/62 Barometer 28 9? in '

H S"
pSia
Rotameter 59 Reference: Figure 7 '

Tj. 54 »f
b °- 2 141/316/308 T 2 78°F
P
,n |p
f Pd lbm „ cuft cX
psig
s
dn. Hg -psig '
¥V
lmin Hfl mm N In Oi Nln<* ®d \%
1. 20
-4. 8 43.3 . 950 . 1423 . 182 . 0259 J[ 1. 345 3.48
1. 282
1 . 035 . 1540 • 248 .0382 1.10 5 4. 22
1. 370
2 . 835 . 1659 . 315
• 0523 1.036 5.41
1.4 51
3 . 760 . 1779 . 372 . 0662 .887 5.88
1. 535
4 . 700 . 1900 . 429 . 0815 . 774 6. 30
1. 6?]
5 . 640 . 2025 . 483 . 0979 .671 6. 57
1 710
6 . 580 . 2150 . 1152 . 577 6.6
. 536
1. 790
7 . 547 . 2285 582
.
. 1330 . 518 6.9
1. 875

8 . 500 . 2415 . 6*28 . 1515 .452 6.87

1. 960
9 .480 . 2555 . 673 . 1720 .416 7.15
Z 045
10 . 458 . 2690 . 715 . 1924 . 379 7. 3

2. 210
12 .420 . 2980 . 793 , 2360 . 323 7.61
2. 465
15 . 380 . 3440 3100 . 261 8.1
.902
.

2. 630

17 . 360 . 3760 966 . , 3630 . 232 8.42

2. 890
20 . 336 . 4280 1.0 6 .4540 .1975 8.96
3.0 55
22 . 313 . 4§50 520 174 9.05
1. 117 , .

3. 310
25 . 272 . 5240 1.196 .626 . 1415 8.86
3.470
27 . 232 . 5650 1. 245 . 704 . 1135 7.99
3. 730
30 .087 . 6310 1.318 . 832 . 0405 3. 37
 51

Date 28. 50 in Hg.

4/13/62 Barometer
Reference: Figure 14. 00 psia
Rotameter 93 8
T]_ 53*F
b °* 3 T2 82°F
173/316/237
Pn Ps 5 lbm cuft 1

<X
W]mi n Q^mm N NlnOC
.
P si g in. He. psie In (X *>d ^>oo
1.075 i. "*n?
98,5 -6.6 71 1.075 .142 264
. .0375 .941 .0353
. 883 1.488
2 .882 .1658 398
. .066 .676 .0446
.754 1.673
4 .755 .190 5 . 515 .0982 . 514 .0504
.620 1 8*8
6 .628 . 216 . 619 .134 . 382 .0512
. 54 2.045
8 . 54 . 243 . 715 . 174 ,
300 .0*22
2. 23
in .48 . 271 .802 . 217 . 245 .0532
2.41
17 :4R0 ™i 88ft ?.A* .21 .0557
2.69
15 , 377 . 347 .989 . 343 .1592 .0546
% iA
20 .355 .433 1.15 .498 .128 .0638
3.62
25 . 336 . 531 1. 285 .682 .10 56 .072
4. ftQ

30 .319 .642 1.41 .908 .0889 .0805

4 56
35 ?.qn . 77?. i k? 1 171 07?,S, .085
*m
40 . 247 . 924 1.61 1.485 T
056 .0832
5. 30
43 .185 1025 1.67 1.712 .0398 .0681
5.48
45 .084 L104 1. 70 1. 88 .01745 .0328
S.62
46. 5 L162 1. 725 2.005

.
 52

Date 4/12/62 „ 29.02 in Hg.

Barometer 14. 25 psia
Rotameter 50 Reference: Figure 9
T x 54°F
b 0.4 141/224/291 T 2 84°F
Pn Ps 9psig W Ibm cuft c*
N N
psig in Hg. 1
mm Ql mm In CX
In of Ai ty
: 90 3 1. 104
70 -2. 7 Q 36.6 . 90 5 . 2035 .099 .02015 1. 51 . 304
. 680 1 ?^R
2 . 680 . 239 . 2296 .0548 r
99? .0 547
. 565 1.41
4 . 565 . 276 .3435 .0947 . 737 .0698
.454 1.566
6 .455 .316 .449 .142 JSJM .0757
.382 1 721
8 .382 . 359 . 543 . 1952 .408 .0798
1. 877
10 . 330 . 404 . 629 . 254 . 324 . 0824
2.03
12 . 290 .453 708
. . 321 . 263 .0845
2.185
14 . 258 . 504 . 781 . 393 • 2175 . 0855
?.. ^4
16 . 220 . 56 . 869 .487 . 173 .0843
2. 495
18 . 205 . 621 . 914 . 568 .1514 0875
2. 65
20 .191 . 685 . 975 . 668 .1325 . 0885
2. 96
24 .163 . 788 1.085 . 856 .1015 . 0870
V 27
28 .152 .005 L 185 1.192 .0857 . 1022
3. 58
32 .142 1.217 1. 275 1. 55 . 073 .1133
3.89
36 .137 1.475 1. 359 2.0 .0649 . 129
4 04
38 . 105 1.635 1. 396 2. 28 .047S . 1094
4. 2
40 . 041 1.81 1.43 5 2.6 ,0180 .0467
4.275
41 1. 905 1.452 2. 77
 53

4/13/62 28. 57 in Hg.

Barometer
14. 03 psia
Rotameter 93.0 Reference: Figure 10
T]_53°F
b °' 6 173/224/120 T2 83. 5°F
Pn Ps Ibm c*
wJ-xxun Q2^£
P si S ,
inlfe psig u mm N
In <X
1. 255
NlnO< *>d
\&
29__ -5.8 71 1.055 .1418 227. .(5322 r 9? .0296
1.435
2 . 988 . 1652 . 361 .0596 . 751 0448
1. 613
4 . 875 . 1896 .478 .0905 . 593 .0537
1. 79
6 . 748 . 2155 . 582 .1255 .455 .067
1 Q«
8 . 6<^5 ?4? 683 16,^ ~*A? n^Q7
2. 33
1?. 485 299 . 84 5 . 2 52 . S?,?,^, . 0573
2.6
15 .415 . 346 .955 .331 .1745 .0577
3.04
20 . 330 .431 1. 112 .480 . 1185 .057
3.49
25 .279 . 528 1. 25 . 66 .08.74 . 0577
1 Q4.
™ ?*} A^ft 1 11 87^ ,,.il64J n^Ai
4. 39
35 .195 . 767 1.48 1. 135 .0486 .0551
4. 83
40 .167 . 916 1. 575 1.441 .0377 .0544
* ?»
45 .150 1.093 1. 665 1.82 .o?ii .0566
e; 11
50 136 1 307 1 74^ ?. ?.R n ?. *)=; fl=iq?
6. 19
55 .129 1. 57 1. 82 2. 855 .0228 .0651
6.64
60 .121 1.9 1. 892 3.6 .01998 .0719
7 OR
65 . 112 2. 3 2 1. 956 4 =14
nj.7.,1.,.,
078^
7. 52
70 . 108 2.9 2. 02 5^.86 . 0157 .092
7.61
71 .0 3.03 2.03 6.15