Professional Documents
Culture Documents
R. G. CUNNlNGHAi
tie Liquid Jit Gas P u p
Professor of Mechanical Engineering,
The Pennsylvania State University, Gas compression with a liquid jet occurs isothermally and hence with minimum work.
University Park, Pa. Performance characteristics oj the liquid jet gas pump (efficiency arid compression
ratio versus inlet volumetric flow ratio) are predicted accurately by a one-dimensional
R. J. DOPKIN analysis providing (he mixing zone remains in the throat. Jet breakup was investigated
to enable prediction of required throat length and to improve efficiency. Effects of
Engineer throat length, nozzle contour and spacing, nozzle-throat area ratio (0.15 to 0.4-5), jet
E. I. DuPont de Nemours and Co.
Textile Fibers Dept. velocity and suction pressure were investigated. Optimum throat lengths were found;
Spruance Plant corresponding efficiencies exceed 40 percent. Two jet breakup flow regimes were
Richmond, Va.
found: impact and jet disintegration. For the impact regime, jet breakup length
depends on inlet velocity ratio, jet Reynolds number and nozzle-to-throat area ratio.
Optimum throat lengths were found to be an empirical function of nozzle-to-throat
area ratio and ranged from 12 to 82 throat dia. These results, coupled with the one-
dimensional model, permit design of efficient liquid jet gas pumps.
-Nomenclature-
JP= jet Reynolds number rjmt mixing throat efficiency
A = cross-sectional area, ft2
D = diameter, ft
o- _ liquid surface tension lbf/ft
3
Kn!, Ken, MlSc p = density, lbm/ft
friction loss coefficients Vio'A.pi V = velocity ratio at the throat
K,h, Kdi — IjJ— jet Weber number inlet, TWFio = <t><,/c
LJG = liquid jet gas pump 2
L, = throat length, ft a = diffuser area ratio, At/A a 111 = liquid viscosity, lbf sec/ft
b = area ratio, A„M i 4>o, <t>l, 4>d =s: volumetric flow ratios, Qi„/
LT = throat length/dia ratio L</
Di
1 - b Qi, Qj./Qi and Qu/Qi
c — area ratio, Ai0/A„ — —;—
Lm = jet breakup length, ft o Subscripts
p = pressure, lbf/ft2 e = energy rate, ft lbf/sec 1 = liquid primary flow
P = total or stagnation pressure, gc = gravitational constant ft 2 = gas secondary flow
lbf/ft* lbm/lbf sec* 3 two-component mixture of
3
Q = volumetric flow rate, ft /sec in = mass flow rate, lbm/sec gas in liquid as the con-
R = gas constant, ft lbf/lbm °R 2ZWc tinuous medium
S = nozzle-to-throat spacing in
n = jet pump number ——
Po i, n, s, o,
throat diameters, Si/D, rt0, rat, r,,s = pressure ratios Pt/P0, Pa/ I, d = location, see sketch
St = nozzle-to-throat spacing, ft P„ Pa/P, f = force
T = temperature, °R xo = specific work ft lbf/lbm = fric = friction
V = velocity, ft/sec W/m 9 = gas partial pressure
W = work rate ft Ibf/sec y = density ratio, pio/pu where he = kinetic energy
z= Pio = Pto/RTia ideal gas m = mass or mixing length
jet. velocity head,
2<? c relation opt = optimum
lbf/ft2 V>p = ie^ P u m P efficiency V = vapor
rc=
cooling water
thermometer
and the isothermal compression work on the gas by the liquid.
manometer The throat pressure rise is also diminished by the mixing loss
-~~rfloi",'
which occurs whenever two streams of dissimilar velocity are
mixed [1J. This phenomenon is normally the largest source of
energy dissipation and it controls the overall performance of
Fig.2 Test stand and instrumentation the LJG pump. Mixing loss is a function of the inlet velocity
2Zb2c
ratio v == V'o/V lo and the jet pump number n == - - ; and the
Po
Pressure rise in the throat: design of the LJG pump involves selecting the ar£a ratio b to
permit operation at the lowest possible jet pump number n, in
order to minimize mixing losses. In addition to v and n, the dif-
fuser compression ratio rd' and the overall pump compression
ratio rd, are also functions of the nozzle-throat area ratio band
(2) the diffuser area ratio a, i.e., rdl (v, n, b, a) and rd,(v, n, b, a).
1.0
mixing process were measured with a series of static pressure
taps in the throat sections.
The influence of nozzle design was investigated and five of the
nozzle contours are described in Fig. 4. The key design param-
eter is the nozzle-to-throat area ratio 6 = A„/At] in the case of
jets formed by the square-edge orifices, the cross-sectional area
A„ is taken to be that of the vena contracta. The contraction
coefficients, measured by mass flow tests, and orifice diameters
are listed in Fig. 4 for nozzles sized to provide three area ratios:
b = 0.153, 0.300 and 0.449.
Nozzle-to-throat spacing was <S = 3 throat dia. Performance
appeared to be relatively insensitive to spacing (from S = 1 to 7
throat dia) when corrected for the S effect on total jet length
from nozzle tip to break up of the jet [15].
An
Shown in Fig. 5 are results for a b = 0.300 pump in terms of
efficiencies and compression ratios for the pump, and for the
mixing throat alone; the separation of the curves is of course a (
result of the diffuser compression process. Note that the experi- /
Theory; D
mental points agree well with the theoretical predictions up to
the "departure point" flow ratio which is labelled $„' on the Tjj> Z = 96.3 psi
D
curve. The flow ratio <po (the independent variable) is deter- 10 P = 13.0 ppsia
o
mined by adjustment of the pump discharge valve and hence Kt„ - .39 b = 0.300
pressure Pd- At the maximum Pd setting, the mixing zone is Kdl. - .26
located near the throat inlet; further increase of P d will cause
flooding or backflow into the air suction port. As Pd is reduced
in finite steps, air is progressively induced in greater quantities
and simultaneously the mixing zone moves downstream in the i Flow Ratio
throat. The initial data points (at <£„ = 0.6 in Fig. 5) correspond o
to Lm = 8, that is to mixing completed about eight diameters Fig. 8 Theory-experiment comparison including behavior of mea-
sured friction coefficients
01—
1.0 1.2 1.4 1.6 in Figs. 7 and 8. Apparently Kdi tends to be smaller for LT
2.0
values near the optimum length and this is attributable to the
* 0 Flow Ratio
beneficial effects of intense mixing at the diffuser entry which
Fig. S Throat and diffyser las* coefficients versus flew ratio mea- prevents or delays phase separation. Variations of the K curves
sured with throe throat lengths for 4>0 > 4>o, i.e., beyond the departure points are not of practical
concern since this is an off-design condition. The Kth values rise
for 4>o > 4>o', because the pressure recovery from momentum
the test parameter with b = 0.300 and n = 4.91. The LT = 10 transfer is partially lost as mixing occurs in the diffuser. Con-
pump gave poor results, i.e., a maximum efficiency of 30 percent versely, Kdi values drop for 4>, > 4>J because a limited amount
and a departure flow ratio <j>0' of about 1.1. Increasing the throat of mixing in the diffuser entry is beneficial. Negative Ka values
length to LT = 16, 22, and 28 dia provided much better <j>„' which occur for <$><>>> <$><>' simply attest to the departure from
levels. The longer throat lengths caused greater Motional losses the essentially one-dimensional homogeneous flow which is ap-
and thus depressed the efficiency at a given flow ratio (compare parent for 0,, < <t>„'.
at 4>0 = 1.0). The curves suggest that an optimum exists since
the $0' gain for LT - 28 over the LIT = 22 pump was marginal Summary of Friction Coefficients. Friction coefficient results
and at the price of a drop in pump efficiency. The theoretical may be summarized as follows: Ka< — 0.26 to 0.35 for all con-
curve shown in Fig. 7 is for KM, = 0 and this curve of course lies ditions; Kth = 0.32 to 0.48 for optimum throat lengths; Ka =
above all four sets of experimental data. (By replacing K,?, = 0 0.15 and = 0.60 for short and for very long throat lengths, re-
with the appropriate K,n values [see Fig. 8] experiment-theory spectively. Because the gas throat-entry pressure loss is neg-
comparison similar to Fig. 5 would result.) ligible, Km = 0 is recommended. Nozzle coefficients may be
taken from direct tests or from the liquid-liquid jet pump liter-
The corresponding K curves for LT = 16, 22, and 28 appear ature. For square-edged orifice nozzles with accurate contrac-
in Fig. 8. The constancy of K values is again evident; the aver- tion coefficients (jet area is based on the vena contracta) K„, = 0
age Ku, and Kdi values in the 4>„ < </>/ regions are shown for is applicable. Nozzle coefficients corresponding to Fig. 6 are
each curve. Note that K,h increases from 0.15 to 0.43 in propor- K„, = 0, 0.10, 0.25, 0.14 and 0.16, respectively, for the SE1,
tion to length LT. Additional data including effective friction El, CI, 7H and E3 nozzles.
factor values/(A = Ktli/LT were reported in [15]; the approxi-
mately linear increase in Kth with LT will facilitate prediction Shown in Fig. 9 as a function of throat length are maximum
of performance with excess LT. Subsequent tests showed that jet pump efficiencies and corresponding inlet flow ratios <f>„''.
LT = 19 - 20 was optimum for 6 = 0.300, i.e., slightly shorter These data were cross-plotted from performance curves for pumps
than for the LT = 22 center curves in Fig. 8. The flatness of the with mixing throat lengths from ten to 38 dia in length. First,
K(4>„) curves for <j>o < 4>o will fail under extreme LT conditions: confirming the suggestion in Fig. 7, there is an optimum length
(a) very short lengths prevent complete mixing in the throat (19-20 dia) for this 6 = 0.300 pump operated at a jet pump
and the K's vary continuously; (or LT = 10 the range was 0.3 number of n = 4.91. Secondly the plot of flow ratio <f><>' as a
to 1.4 for K, and —0.2 to 1.5 for Kdi [151; (&) very long throats function of mixing throat lengths reveals two flow regimes.
h
at low area ratios may promote phase separation and the re- Two Flow Regimes. In Fig. 9 there is a relatively rapid re-
sulting impairment of pressure recovery will be reflected by sponse of air induction effectiveness (hence 4>o) to increased
abnormally high Kdi values. throat lengths up to the optimum value of LT = 19-20 dia. The
The diffuser coefficients are relatively immune to LT varia- slope then changes abruptly, i.e., there is a significantly reduced
tions. A value of Kd< = 0.30 is representative for the series response of ($><,' gain for throat lengths beyond LT ~ 19-20.
M
30
J
20
10
Oi—. 1 _J 1 I I i i
0 1.0 2.0 3.0 *.0
* 0 Flow Ratio
222
Downloaded From: /SEPTEMBER 1974
https://fluidsengineering.asmedigitalcollection.asme.org Transactions of the ASME
on 06/30/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
~r— r™" ~~~ 1™ - — — ~ T ~ ~ — < — ~ T -
Nozzle SE1
b = .153
Z = 152 p s i
/ °^r^'^
-P = 13.0 psia
-
r, = 3.03 p
S = 3
a -/ 0
Data Type Method -
P $ ' Departure Point A
Sk
a J- A $ n " Departure Point 8
&/o O Pressure Distribution C
" /
Fig, 14 Mixing l@ngth-f!ow ratio correlations for three jet vaiochioa
(mathod C pressure distribution data)
• Flow Patio
1.2 1.4x10"-
The remaining task is to identify a dimensionless correlation
that will collapse all of the data to a common curve. Free jet Fig. 17 Mixing lengths versus flow ratio/Reynolds number ratio
breakup studies [21, 22] have been expressed in terms of length
in jet diameters plotted versus the reciprocal of the jet Reynolds
number. Fig. 17 shows that this correlation is apparently ap-
plicable for the L J G pump. Mixing lengths in Fig. 17 are non- of pumps with LT :S LTovt whenever the mixing zone is posi-
dimensionalized with jet diameter Du and the flow ratio 4>„ is tioned (off-design) at Lm < LT0!,t as a result of setting Pd a t
divided by the jet Reynolds number ??. As a result data forming
discharge pressures in excess of the design P,i level.) Breakup
the several curves in Figs. 14, 15 and 16 are collapsed to one
length in the impact regime is affected by jet velocity Via,
relation, L,n/Du = 7.86 X 10° <£»/£?, up to the three breakaway
velocity ratio v, area ratio 6, and by the jet Reynolds number.
points. Included are data for b = 0.153 and 0.300, and in addi-
The linear relation permits calculation of Lm from flow ratio (j>0
tion 0.449, and for three values each of jet velocity heads Z
and conversely.
and suction port pressure P,. The breakaway points are evident-
2 The breakaway point Ln — 15Di0c represents the optimum
ly dependent on area ratio 6, while velocity and suction port pres-
LT values for the range of V\„, P, and b values in this study, us-
sure effects have been "eliminated" through use of these coor-
ing water and air as the liquid and gas components. Apparently
dinates.
jet velocity and suction port pressure exert only second order ef-
The final correlation is shown in Fig. 18. By dividing the di- fects on the optimum value of LT for the confined mixing of a
mensionless length L/Di0 and the parameter 4>0/1? data by the liquid jet and gas annulus. The formula optimum lengths in
Aio 1 —b throat and jet diameter terms are:
area ratio c = —- = —-— , all of the data up to breakaway
An b
fall on a common curve Lm/Dioc = 7.86 X 10° vfR. The coor- LJD, LJDi0
dinates of the common breakaway point, in Fig. 18 are v/1?. =
1.91 X 10-6 a n d LJDloc = 15. 0.153 32.4 83.0
The significance of Fig. 18 may be summarized as follows: 0.300 19.2 35.0
0.449 12.3 18.4
1 The empirical expression Lm/Dioc = 7.86 X 10« v/1Z. de-
scribes the jet breakup length behavior in the impact flow regime The data used in arriving at Lm/Duc = 15 as the optimum point
up to the optimum LT length. This flow regime obtains for an were based primarily on the 6 = 0.153 and 0.300 pumps. D a t a
LJG pump with LT < LTopt. (It can also occur in the operation for 6 = 0.449 were few in comparison, b u t Figs. 17 and 18 show
that the breakaway point (hence optimum length) for b = 0.449
fits the behavior pattern established with the smaller — b pumps.
3 From a design standpoint, 1 and 2 describe the useful
l * 152.0 pst
Ps • 13.0 psia
5 • 3
Nozzle b _b "j L_
O SE! .153 .153 13.0 152.0
A SE2 .300 .153 12,5 244.0
.300 13.0 40.4
.300 13.0 96.3
.300 13.0 152.0
.300 9.0 95.3
.300 4.58 96.3
.449 13.0 71.5
Fig. 16 Mixing lensth-flow ratio correlation for b =0.153 and 0,309 Fig. 18 Mixing l«ng ths ¥ersus inl©t velocity ratlo/Reynolitt number
(msthod C data) ratio