You are on page 1of 7

Modeling and Design of Jet Pumps

D.T. Hatzlavramldls, SPE, Mobil R&D Corp.

Summary. Models for jet pumps currently are derived under the assumption that the power and well fluids are incompressible liquids
that, in many cases, are assumed to have equal densities. When either the well or the power fluid contains gas, current design practices
still use the equations for incompressible liquids and account for the presence of gas by modifying the mass-flow-rate ratio and the
friction-loss coefficients. This paper proposes a new approach to modeling pumps operating under multiphase-flow conditions.

Introduction
Jet pumps were initially used in the oil industry for artificial lift. 1,2 the mixture of the two fluids in the throat leads to the following
Only recently have these pumps been used in EOR processes 3-5 governing equations:
whenever lowering the bottomhole pressure (BHP), extending the
Pi-Po=(I+Knz)Z, ................................ (1)
production stage, and limiting cavitation and/or sand erosion prob-
lems are desired. FpFJ
The concept of jet-pump operation is simple. Power fluid enters Ps-Po=(l +Ken)--Z, .......................... (2)
the nozzle at high pressure and leaves as a high-velocity jet (the Fla
nozzle acts as a converter of potential into kinetic energy) that is
J,
directed into a constant-area mixing tube or throat filled with stag-
nant or slowly moving well fluid. As the high-velocity jet pene-
trates the well fluid in the throat, momentum is transferred from
Pt-Po =
4 2
2FAn -(2+Kth )FAn (1 +FpFq)(1+Fq) +
2FpFJ Fln
I-FAn
.................................... (3)
the fast- to the slow-moving fluid elements, and mixing between
the two fluids occurs. The mixture stream of the power-fluid jet and prPt=Z(1 +FpFq)(l+Fq)Fln(1-Kdi-Fld)' ....... (4)
and the entrained well-fluid elements spreads, while the undisturbed
well-fluid core gradually decreases as the flow progresses. At the In terms of compression ratio and efficiency, Eqs. 1 through 4
throat exit, the mixture stream occupies the entire cross-sectional yield
area of the throat and has substantial kinetic energy. It is subse- rp = [2FpFJ Fln (1-FAn)+2FAn(1-FAn)2 -(1 +Kth +Kdi )
quently directed into an expanding area diffuser, where part of its
kinetic energy is converted into potential and its pressure rises to XFln (l-FAn )2(1 +FpFq)(1 +Fq)-FpFJFln (1 +Ken)]/
a level sufficient to lift the mixture of the two fluids to the surface. [(1 +Knz)(I-FAn )2 -2FpFJFln (I-FAn)-2FAn(1-FAn)2
CUnningham 6 proposed comprehensive mathematical models for
liquidlliquidjet pumps in the late 1950's. Although his early models
+ (1+Kth +Kdi )Fln(1-FAn )2(l +FpFq)(1+Fq)] ...... (5)
address systems of equal-density liquids, they have been adopted Q2(Pd-Ps) (w/Ph(Pd-Ps)
in their exact form by some designers 7.8 not only for liquidlliquid and 1/= = =Fqrp . ............ (6)
systems of different densities but also for systems with free gas. QI(Pi-Pd) (w/ph(Pi-Pd)
For the latter, slight modifications of the volumetric-flow-rate ra- The method of deriving Eqs. 1 through 6 is similar to that for
tio and the friction-loss coefficients 7,9 have been suggested. Re- deriving the governing equations for the gas/liquid pump (Appen-
cent derivations of the liquidlliquid pump model 10, 11 account for dix A).
density differences between power and well fluids. In addition to It can be shown that the equations of two recent model deriva-
his liquidlliquid pump model, Cunningham 12 proposed in the early tions at the Inst. Francais du Petrole 10 and Delft U., 11 with proper
1970's a complete model for a jet pump used to pressurize a non- symbol conversion, reduce to the equations presented above. When
condensable gas. No model is available for a jet pump that uses the symbols in Ref. 10 are converted into our notation and the iden-
gas as the power fluid. Although it is recognized that both power tifications R=FAn , k=Fp , M=Fq , K I5 =Knz, K25 =Ken, K56 =
and well fluids generally can be multiphase mixtures, the design K th , and K67 = Kdi are made, the two derivations become equiva-
of jet pumps for oil wells is based totally on a model for single- lent. Similarly, in Ref. 11, if the identifications FpD = rp' FAD = FAn'
phase, incompressible liquids. FwD=FpFq' Ktd=Knz-FpFJFln/(l-FAn)2, Ppf/Pf=lIF p , and
Ppf/PM=(l +Fq)/(1 +FpFq) are made, the equations shown above
Governing Equations are obtained. In contrast, Cunningham's equations in the form used
In the work presented here, Cunningham'sl2 nomenclature from by some designers, 7 with the identifications R=FAn , M=FpFq ,
his liquid/gas model (i, n=the power fluid at the nozzle inlet and KID =Kth + K di , KN=Knz + F~ FJ FIn /(1- FAn)2, yield
exit, respectively; s=well-fluid intake, o=throat or mixing-tube rp=[Fln (1-2FAn)F~Fi +2FAn (l-FAn )2 -(I-FAn )2Fln
inlet; and 1,2=power and well fluids, respectively) is retained.
x(1 +FpFq)2(1 +Kth+Kdi)]/[(1 +Knz)(1-FAn )2 -2Fln
Liquid/Liquid Pump. The following assumptions are made. X(I-FAn)F~FJ -2FAn (l-FAn )2 +(1-FAn )2Fln
1. ID flow conditions apply.
2. Power and well fluids are incompressible. X(1+FpFq)2(1+Kth+Kdi), ........................ (7)
3. The pump operates isothermally and under steady-state con- which is equivalent to those presented here and in Refs. 10 and
ditions. II if and only if Fp = 1, i.e., if the two fluids have equal densities.
4. The kinetic energies at the inlet and outlet of the jet pump are
negligible. . Gas/Liquid Pump. Retaining the same assumptions as for the liq-
5. The fluids are perfectly mixed at the throat outlet. uid/liquid pump, except that now the power fluid is an ideal gas
6. The average jet velocity is maintained between the nozzle outlet and the well fluid is an incompressible liquid, we obtain (Appen-
and the throat inlet. dix A) the following governing equations:
With these assumptions, application of the macroscopic mass,
Po In(p;lpo) =(1 +Knz)Z, ........................... (8)
momentum, and energy balances for the power fluid in the nozzle,
the well fluid between the suction port and the throat entry, and FpFJo
Ps-p o =(I+Ken ) - - Z , .......................... (9)
Copyright 1991 Society of Petroleum Engineers Fla
SPE Production Engineering, November 1991 413
Finally, the problem can also be cast in terms of dimensionless
variables FAd, FAn' and (Pd-Ps)/Z or Po/Z, where the last two are
pump Euler numbers.
An important aspect of the problem in question is that, to ensure
physically realizable solutions to the governing equations, one needs
Po ] to place bounds on the parameters FAd and FAn and the indepen-
X-(I+F qt )+2FAn ............................ (lOa)
Pt dent variable Fq' These bounds result from enforcing the con-
ditions
Pd-P s <? (Pd)min -Ps <?O ............ : . (15a)
and 1/s1 ........................................ (15b)
for a positive F q' Thus, for a certain combination of flow-area ra-
tios, FAd and FAn' there exists a range of Fq for which predictions
of the jet pump are not physically realizaole.
To accommodate for the presence of gas in either the power or
the formation fluid, instead of the modifications to the flow-rate
ratio followed in current design practices, 7 I propose the follow-
ing. If the formation fluid contains gas, the jet-pump design can
be treated as an intermediate case between the liquid/gas and the

X(Po
Pd
y (1 +F qd ) 2 -KdiFln Po Fqo(1
Pt
+Fqt)]-~ln(Pd).
Fqo Pt
liquid/liquid cases, for which designing equations have been avail-
able for some time. If, on the other hand, the power fluid contains
gas, the jet-pump design can be treated as an intermediate case be-
tween the gas/liquid and the liquid/liquid cases. Because no equa-
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (11) tions were available for the gas/liquid pump, they are derived here .

In terms of performance efficiency, Eqs. 8 through 11 yield Model Subtleties


1/=Fqo(Pd-Ps)/[Po In(p;!Pd)]' ..................... (12) To determine the optimum flow-rate ratio, Fqopt, in addition to the
The governing equations for the liquid/gas pump 1 are given in area ratios FAd and FAn' the pressure-loss coefficients (Knz , Ken'
Kth' and K di ) need to be provided. In principle, these coefficients
Appendix B.
depend on the area ratios, the flow rates, and the viscosities of the
From energy analysis of the gaslliquid pump (Appendix C), the
fluids.
energy dissipation associated with the fluid mixing is
Area-ratio dependence can be introduced easily. At high Rey-
nolds numbers, the suction port can be approximated as a sudden
Z [ 1-2FAnFqo-Fln (Fqo)2
m=--- - (1 +FpFqo)(1+Fqt )2 expansion, the nozzle as a gradual contraction, and the diffuser as
PloFqo F qt a gradual expansion.
Viscosity and flow-rate dependence can be introduced together
F 2F F3 F2 F F3 ] through the Reynolds number. This dependence is of particular im-
+2F2 ~(I+F F )(I+F )F _ p qo An +~ portance for the pressure-loss coefficients in the throat and the
An F p qo qt qo I-F F2
qt An Aa diffuser where power and well fluids flow together. The flow mech-
anism is quite complicated in these regions.
The power fluid enters the throat as a high-velocity jet while the
-~ln(Pt). ................................ (13) well fluid flows originally through the annular area between the
PloFqo Po throat wall and the jet boundary. It is assumed that the throat inlet
coincides with the nozzle exit, although in practice the nozzle exit
Methodology is withdrawn upstream by a length equal to several nozzle diameters.
The problem of interest for oil wells is the following. Given the Two flow regions can be distinguished in the throat. In the an-
desired oil production rate, Q2s' the wellhead pressure and BHP, nular region, which begins at the throat entrance, formation fluid
Pwh and P bh, respectively, the depth, D, and the properties of the
is entrained by dragging action in the jet, high-velocity droplets
power and well fluid, Fluids 1 and 2, respectively, select the opti- of power fluid are shed (by wave disturbances) into the annular
mum size jet pump, i.e., the pump with the appropriate area ra- space, and a mixing layer forms on the jet boundary. As the jet
tios, FAd and FAn' and flow-rate ratio, F q , associated with the moves away' from the throat entrance, the mixing layer spreads out,
highest perfoniJ.ance efficiency. while the undisturbed high-velocity core progressively decreases.
The current methodology 7 for solving this problem is iterative, Eventually, a point is reached where the undisturbed high-velocity
does not consider the diffuser (FAd = 1), and does not involve com- core of the jet disappears and the jet is fully developed. Another
prehensive maximization of the efficiency, 1/. It is proposed that point of interest is the point where the outer boundary of the jet
a methodology that honors this maximization should be based on reaches the throat wall. Downstream of these points is the mixing
obtaining optimal FAn and F q from zone, where the two fluids undergo chaotic mixing. When the throat
is short or the backpressure, Pd, is sufficiently high, the mixing
(o1//oFq)(Fq ,FAn )=O ............................. (14a) zone is located near the throat entrance, the potential core is
and (01//oFAn )(Fq,FAn ) =0 ........................... (14b) diminished relatively little, and the mixing is poor. Very high back-
pressure causes flooding or backflow at the suction port. If the back-
Another methodology, which also considers diffuser size, is pro- pressure is reduced, the mixing zone will move downstream in the
posed as a parametric study with regard to FAd, FAn' and Pd' The throat. An optimum is achieved when the downstream of the mix-
area ratios FAd and FAn are chosen in the interval [0,1] (FAd=I ing zone is at the throat exit. Reduction of the backpressure be-
means no diffuser; FAd=O denotes a perfect diffuser), and a low- yond this point causes the mixing zone to move into the diffuser,
er "bound," Pwh +P3dgD (friction losses are negligible compared a situation where the pressure losses substantially increase and the
with static contributions), is placed on Pd' If the diffuser cross- performance efficiency decreases. According to Cunningham and
sectional area is selected to be equal to that of the tubing, Z is known. Dopkin, 13 the characteristic lengths for total mixing and jet break-
With FAd, FAn' and P d selected and with Z known, F q can be cal- down can be correlated with the nozzle Reynolds number, N Re ,
culated from the quadratic throat-pressure equation. Optimal FAd and the throat Reynolds number and Weber number, N Ret and
and FAn are selected as those that maximize 1/. N we , respectively.

414 SPE Production Engineering, November 1991


50

..- 0

40

-
0
as
a:
M


n=1
n=2
n=4
..
CD
:J
30 0
6
c
n=8
n = 10
n = 15
..
III
III
CD n = 20
a.
20
n = 25

-..
as
0
.c 10
ZML

I-

1 234 5 6 8 9 10
Inlet Flow Rate Ratio,

Fig. 1-Pressure ratio, 'to, vs. inlet flow-rate ratio, F at different pump sizes, n. Gaslliquid
let pump FAd =0.0, FAn =0.10. "

If friction losses are neglected, the pressure-loss coefficient for pump design,1 the above limit is modified by approximating
the annular region can be estimated 14 to be 5/144. In the mixing Ps-Pv by P s and Pi-Ps by Z. For flow-rate ratios higher than
zone, the fluid that wets the tube constitutes the continuous phase, Fq,c/, cavitation is predicted; thus, pump operation is restricted to
regardless of the mechanism of jet disintegration (disappearance ratios lower than cP ct. Another way to determine the cavitation-
of the undisturbed high-velocity core or breakup of the fluid thread limited flow rate is to conduct parametric studies in which a and
into droplets or bubbles as a result of flow instabilities and subse- b are kept constant but the flow-rate ratio changes. Through these
quent droplet or bubble coalescence). The nonwetting fluid is in studies, a critical flow ratio can be defined, beyond which pump
the form of small droplets, small bubbles, or coarse slugs, depending efficiency decreases as the ratio of formation to power-fluid flow
on its viscosity. Good estimates of the viscosity of the mixture in rates increases.
this region, in addition to the viscosities and the volume fractions
of individual fluids, require knowledge of the surface tension, flow Ideal Pump
rate, and dispersion type. When the sand content of the formation In this work, "ideal pump" means a pump for which friction is
fluid is not negligible, oil viscosity must be modified to account negligible and the mass-flow-rate ratio is infinitesimally small-
for sand. i.e., Knz=Ken=Kth=Kdi=O and FpFqo=O. Under these condi-
Pressure-loss coefficients have been determined through best fit- tions, Eqs. 1 and 3 reduce to
tings of available experimental data. Cunningham 12 determined
from experiments that Knz varies with the je~ Reynolds number as rio =p;/P o =exp(Zlpo) ............................. (17)
NRe Ih for 500:sNRe :s20,OOO. At very high Reynolds numbers,
however, Knz approaches a constant value, 0.10. Other investiga- and ( -pt)2 - [2ZFln 1
- - - ( ---Fqo ) +1 JPt 2ZFl
_+ _ _n_=O,
tors 8,9,l1 recommend Knz=0.03 to 0.05. Cunningham l2 re~om Po Po FAn Po Po
mends Ken =0, particularly when Ken is combined with K nz , while
other investigators S suggest that Ken =0.03. Cunningham's experi- .................................. (ISa)
ments show that, for NRet~3,SOO, Kth+Kdi=0.30. Other inves- respectively.
tigators S,9,l1 recommend that Kth + Kdi =0.20. In gener.al, The dimensionless form of Eq. ISa is
constant-value pressure-loss coefficients are valid at high Reynolds
numbers where viscous effects become less important. rro -(nFi +l)rto +n=O, ........................... (ISb)
One of the most important problems in jet-pump design is the where n=2ZFln/Po and Fi= lIFAn -Fqo are the pump Euler num-
prediction of the onset of cavitation, which can be induced as a ber or pump size and a dimensionless flow-rate ratio, respective-
result of increased velocity of the power-fluid jet, vI~' or reduced ly. Because F qo ~ 0, Fi:S lIFAn for physically realizable solutions.
suction or discharge pressure, Ps and Pd, respectively. When cavi- Finally, for the ideal pump,
tation occurs, the flow of the well fluid at the suction port ceases
to respond to backpressure reductions, and vapor cavities form at
the throat entrance to implode farther downstream where the pres-
sure is higher. In pumps handling liquid/vapor or liquid/gas mix-
tures, it is also possible for the velocity of the two-phase mixture
to reach the sonic velocity, which is another manifestation of cavi-
tation. - F:o In(:~:). ................................. (19)
For the liquidlliquid pump, Cunningham et al. 15 defined a
cavitation-limited flow ratio as Eq. ISb has a minimum, equal to (nF;*+ 1)/2 at F;*=(2n Ih -1)/
n. Its solution is
F q,c/=C.J(Ps-Pv)/1.3Z, .......................... (16)
r,o = [nFi + l..J(nFi + 1)2 -4n ]/2 ................... (20)
where pv=vapor pressure of the power fluid. This limit is applied
to the case where the power-fluid vapors fIll the suction chamber, From the two roots, only the root with the" + .. sign is accepted
and suction flow for the well is drastically reduced. In current jet- because only this leads to a finite-bounded Mach number.

SPE Production Engineering, November 1991 415


100

110 o
~-
0
~- 80

70
//// o
i "-""'LG
~ ......... n 1 : rclo a:
---n.1: rto ""''''LG
i 80 '-'LG
.rt....... ...n.4 : rdo f ---"GL
~n.4: rto :I
i 50
....... ..n 10 : rdo ---GL
.---GL
1i no
a: 40
............., ...,
---n .10 : I -LL
-LL
f 30 -LL
:I
20
...................................................
f
II.
10 ................n
o ..................................
10 2
inial Flow Rata Ratio, F'"
a
a
~~-----------------------, ..i
0
ia: .LG : n .. 0.08
.........._. LG : n .. 0.32
.... LG : n .. 0.80
! ---.. -- GL : n .. 0.08
--..n 1 ; rdo :I
- n 1 : rto
.... -....n 4; rdo
---.-- GL : n .0.32
---.-- GL: n. 0.80
f _LL:n.O.08
- n 4: rto Ii.
10 .. ---~-n 10; rdo
_LL:n.O.32

~
-LL:n .. O.80
L-----:::::::::::::::::=i - - - n
.....................................................
10; rto
:I

=
is
----_......-.. _-_ . -_. -.. _...-......- ........---_.-_..- -- 1
0 1 2
1.1 1.2 1.3 1.4 1.5 1.8 1.7 1.B LSI 2.0 b inial Flow Rata Ratlcl,
Inlet Flow Rate Ratio, FI
b
..8
0
-- -... --- ...
ia:
............
..................
f
............. .LG : n. 0.08
:I ...LG : n .. 0.32
8 .... LG : n .. 0.80
---.. --GL : n '" 0.08
~
-"--"---n 1 ; rdo
I ---.--GL : n .. 0.32
- n 1 : rto ---.--GL : n .. 0.80
.. .... --n 4 ; rdo - L L : n .. O.08

~

- n 4 : rto _LL:n.O.32
-"--4---"
....__... _--.......... ....-.
10 ; rdo - L L : n .. O.80
:I
...................... ..........
. . __ .. _-_ ..................----_ ... _- _
- n 10 ; rto
=
e .. ............................................ 4 ....
1i
e ::::::........................................................... ..
...... ,

. . ------..--a---------.-.. .------..---.. . -----


1.00 1.02 1.04 1.08 1.08 1.10 1. 12 inial Flow Rata Ratio,
C
c Inlet Flow Rata Ratio,

Fig. 3-(a) Throat pressure retlo, "0' (b) diffuser pressure ra-
Fig. 2-Pressure ratio, '10
or, do' vs. Inlet flow-rate retlo, F tio, rdI' and (c) overall pressure ratio, rdo' vs. Inlet flow-rate
at different pump sizes, n. Gas/liquid Jet pump FAd = 0.0; (a) " ratio, q, 0' at different pump sizes, n. FAd = FAn = 0.20.
FAn =0.10; (b) b=0.50; (c) FAn =0.90. (GL = gas/liquid; LG = liquid/gas; LL = IIquld/llquld.)

For the ideal gaslliquid pump, the equation for the fluid-mixing containing gas is bounded by the performance of the liquid/gas and
energy dissipation is liquid/liquid pumps, while the performance of a pump that han-
dles power fluid containing gas is bounded by the performance of
Em-
_ npo F; - (Fqo
- [2 )2] -In rto ' ............... (21) the gaslliquid and liquid/liquid pumps.
Fig. 1 shows the throat compression ratio vs. inlet flow-rate ra-
2PloFqo Fqt
tio at different pump sizes in the case of the gas/liquid pump. In
The energy dissipation becomes zero at an inlet flow-rate ratio Fig. 1, the points corresponding to zero fluid-mixing energy dissi-
wZ, which, from a combination of Eqs. 20 and 21, is found to be pation are designated as ZML points. Meaningful solutions to the
Ff=(2rfo In rfo-rfo+1)/rfo(rfo-l) . ................. (22) throat compression ratio exist at F; fii=Fl-
In Figs. 2a through le, the diffuser compression ratio vs. inlet
Meaningful solutions to the equation of the throat compression flow-rate ratio at different pump numbers and different nozzle/
ratio exist for F; <:!.Ff. throat-area ratios are presented. They are for pumps with an ideal
diffuser-i.e., FAd=O. It can be seen that the diffuser compres-
Results sion ratio is more sensitive to the throat/inlet flow-rate ratio than
The results presented here are for ideal pumps, liquid/liquid, liq- the corresponding throat compression ratio. As F; increases (i.e.,
uid/gas, and gaslliquid only. These results reinforce the sugges- as Fqo decreases), the compression ratios for the ideal gas/liquid
tion that the performance of a pump that handles formation fluid pumps increase. However, as FAn-+l (i.e., no nozzle), the same

416 SPE Production Engineering, November 1991


1.0

0.9

0.8

0.7
s=- .......... LG : n = 0.08
... +. LG : n = 0.32
>-
C)
a
0.8
=
........ LG : n 0.80
c 0.5 ---a-GL : n = 0.08
4D ----GL in = 0.32

....-..
C)
0.4 ---GL : n = 0.80
-LL;n=O.08
W 0.3 -LLin=0.32
---LL in=0.80
0.2

0.1

0.0
0 2 3
Inlet Flow Rate Ratio, Fqo

Fig. 4-Efflclency, fI, va. Inlet flow-rate ratio, q,o' FAd = FAn =0.20.

ratios become weak functions of F j At constant F j , as the pump to be true by use of the governing equations, if we make the ap-
size, n, increases, both the throat and diffuser compression ratios proximation rso = 1). At constant pump size, efficiency increases
increase, regardless of the nozzle/throat-area ratio. as the throat/inlet flow-rate ratio increases, regardless of the pump
Figs. 38 through 3c show the results of a comparative study of type. At constant throat/inlet flow-rate ratio and pump size, the ef-
three types, i.e., liquid/gas, gas/liquid, and liquidniquid of ideal ficiency of the liquid/liquid pump is higher than that of either the
pumps. liquid/gas or the gaslliquid pump. Depending on the values of the
Fig. 3a shows that, at constant inlet flow-rate ratio, the throat pump size and the throat/inlet flow-rate ratio, the efficiency of the
compression ratio increases as the pump size increases, regardless gas/liquid pump will exceed that of the liquid/gas pump or vice
of the pump type. At constant pump size, the throat compression versa.
ratio decreases as the throat/inlet flow-rate ratio increases, regard-
less of the pump type. Also, at constant throat/inlet flow-rate ratio Conclusions
and pump size, the throat compression ratio of the gas/liquid ideal 1. The governing equations for the liquidlliquid pump were de-
pump is higher than the compression ratio for the liquidlliquid ideal rived and shown to be identical to those recently derived indepen-
pump, which in turn is higher than that of the liquid/gas ideal pump. dently by other investigators. It is also shown that the equations
Fig. 3b shows the diffuser compression ratio. It can be seen that, used by some designers are limited to fluids of equal density.
at constant throat/inlet flow-rate ratio, the diffuser compression ratio 2. A number of alternative methodologies, iterative and direct,
increases as the pump size increases, regardless of the pump type. were proposed for selecting the optimum size and flow-rate pump
At constant pump size, the diffuser compression ratio increases as for a given production rate.
the throat/inlet flow-rate ratio increases, regardless of the pump 3. A critical review of design correlations was made and numer-
type. Also, at constant throat/inlet flow-rate ratio and pump size, ous mod~fications regarding the dependence of the pressure-loss
the diffuser compression ratio of the liquid/gas ideal pump is higher coefficients on area ratios, flow rates, and fluid viscosities were
than that of the liquid/liquid ideal pump, which in tum is higher suggested.
than that of the gas/liquid ideal pump. 4. It was proposed that design of a pump that handles well fluid
Fig. 3c shows the overall compression ratio r do =(rdt)(rto ). Ac- containing gas be considered an intermediate case between the liq-
uid/gas and liquid/liquid pumps, while the design of a pump that
cording to Fig. 3c, at constant throat/inlet flow-rate ratio, the overall
handles power fluid containing gas should be considered an inter-
compression ratio increases as the pump size increases, regardless
mediate case between the gas/liquid and liquid/liquid pumps.
of the pump type. At constant pump size, as the throat/inlet flow- 5. Governing equations fo~ the gaslliquid pump were derived and
rate ratio increases, the overall compression ratio decreases for the an energy analysis was made for the same pump.
gaslliquid and liquidniquid ideal pumps and increases for the liq- 6. Results of calculations of compression ratios and efficiencies
uid/gas ideal pump. The overall compression ratios for the gaslliquid for all three pump types (liquid/gas, liquid/liquid, and gas/liquid),
and liquid/liquid ideal pumps are about equal. At constant throat/ operating under conditions of negligible mass-flow-rate ratio and
inlet flow-rate ratio (lower than "" 3) and constant pump size, the friction, were compared to determine the effect of gas and its loca-
overall compression ratio of the gasniquid ideal pump is higher than tion (as power or as well fluid) on performance.
that of the liquidlliquid ideal pump, which in tum is higher than 7. In general, the presence of gas in either the power or well fluid
that of the liquid/gas ideal pump. At constant throat/inlet flow-rate reduces pump efficiency. When gas is present in the well fluid,
ratio (higher than "" 3) and constant pump size, the overall com- the overall compression ratio of the pump is lower than that of a
pression ratio of the liquid/gas ideal pump is higher than that of pump with gas in the power fluid operating under identical con-
the gaslliquid ideal pump, which in tum is higher than that of the ditions.
liquidlliquid ideal pump.
Finally, Fig. 4 shows the performance efficiencies for the three Nomenclature
types of ideal pumps. Note that at constant throat/inlet flow-rate A = cross-sectional area, ft2
ratio, as the pump size increases, the efficiency increases for the C = constant in Eq. 16
liquid/gas and gas/liquid pumps, while the efficiency of the liq- d = diameter, ft
uid/liquid pump is not a function of pump size (this can be shown D = depth, ft

SPE Production Engineering, November 1991 417


FAa = annular area ratio, (At-An)/An=(I-FAn)/FAn Reference.
FAd = throat/diffuser-area ratio, At/Ad 1. Gosline, J.E. and O'Brien, M.P.: The Water Jet Pump, U. of Califor-
FAn = nozzle/throat-area ratio nia Publications in Engineering, U. of California Press, Berkeley, CA
F; = dimensionless ratio for ideal gaslliquid pump, (1942) 3, No.3, 167-70.
2. Brown, K.E. and Petrie, H.: The Technology ofArtificial Lift Methods,
(lIFAn )-Fqo
PennWell Publishing Co., Tulsa, OK (1980) 2b, Chap. 6.
F;* = F; at which r,o becomes minimum 3. Galate, J.W. and Mitchell, R.F. m: "Downward Two-Phase Flow Ef-
Ff = F; at which becomes zero fects in Heat-Loss and Pressure-Drop Modeling of Steam Injection
Fq = volumetric flow ratio, q2/ql Wells," paper SPE 13622 presented at the 1985 SPE.Califomia Regional
Fq,cI = cavitation-limited volumetric ratio Meeting, Bakersfield, March 27-29.
4. Galate, J.W.: "Heat Loss Factors Affecting the Design of Deep Arc-
Fqd = volumetric flow ratio, qU/qld tic Steam Wells," Proc., ASME 31st Annual Petroleum Mechanical
Fqo,Fqt = volumetric flow ratios, q20/qlo' q2,1qlt Engineering Conference, Mexico City (Sept. 1976) 244-53.
Fp = density ratio, P2o/Plo 5. Tjondrodiputro, B., Gaul, R.B., and Gower, G.H.: "Hydraulic Jet
FpFqo = mass flow ratio, w2/wl Pumping in a Remote Location," World Oil (Dec. 1986) 35-37.
g = acceleration of gravity, ft/sec 2 6. Cunningham, R.G.: "Jet Pump Theory and Performance With Fluids
of High Viscosity," Trans., ASME (1957) 79, 1807-20.
gc = 32,174 (lbm-ft)/sec 2 -lbf 7. Petrie, H.L., Wilson, P.M., Smart, E.E.: "Jet Pumping Oil Wells,"
Kdi = friction-loss coefficient, diffuser World Oil (Nov. 1983, Dec. 1983, Jan. 1984) 51-56,101-08,111-14.
Ken = friction-loss coefficient, throat entrance 8. Smart, E.E.: "Jet Pump Geometry Selection," Southwestern Petrole-
Knz = friction-loss coefficient, nozzle um Short Course, Lubbock, TX (1985).
9. Jiao, B., Schmidt, Z., and Blais, R.N.: "Efficiency and Pressure Recov-
Kth = friction-loss coefficient, throat
ery in Hydraulic Jet Pumping of Two-Phase Gas/Liquid Mixtures,"
n = jet-pump number, 2ZF},,/po for gas/liquid pump SPEPE (Nov. 1990) 361-65.
NRe,NReI = Reynolds number, vlodnPlo/lLl> V3tdtP3,1IL3t 10. Corteville, J.C. et al.: "Research on Jet Pumps for Single and Mul-
N We = Weber number, v~tdtP3,1u tiphase Pumping of Crudes," paper SPE 16923 presented at the 1987
P = pressure, psia SPE Annual Technical Conference and Exhibition, Dallas, Sept. 27-30.
11. Grupping, A.W., Coppes, J.L.R., and Groot, J.G.: "Fundamentals
q = volumetric flow rate, ft 3 /sec of Oilwell Jet Pumping," SPEPE (Feb. 1988) 9-14.
rio = pressure ratio, p;lpo 12. Cunningham, R.G.: "Gas Compression With the Liquid Jet Pump,"
rp = compression ratio, (Pd-Ps)/(p;-Pd) J. Fluid Eng. (1974) 203-15; Trans., ASME, Series I, 196.
rto,rdt = pressure ratio, P'/Po' Pd/Pt 13. Cunningham, R.G. and Dopkin, R.J.: "Jet Breakup and Mixing Throat
R = gas constant divided by molecular weight, Lengths for the Liquid Jet Gas Pump," J. Fluid Eng. (1974) 216-26;
Trans., ASME, Series I, 96.
ft-psi/(lbm- OR) 14. Bird, R.B., Stewart, W.E., and Lightfoot, E.N.: Transport Phenome-
T = temperature, OR na, John Wiley & Sons Publishers, New York City (1960) 220-22.
v = velocity, ft/sec 15. Cunningham, R.G., Hansen, A.G., and Na, T.Y.: "Jet Pump Cavita-
w = mass flow rate, lbmlsec tion," J. Basic Eng. (1970) 11-20; Trans., ASME, SeriesD, 92,11-20.
Z = jet velocity head, Plovl~/2gc' Ibf/ft 2
Appendix A-Governing Equations for the
= specific energy dissipation, lbf-ftllbm
Ga./Uquld Pump
7J = pump efficiency, Fqrp
IL = viscosity, Ibf-sec/ft2 Energy Equation for the Nozzle.
P = density, Ibmlft3 0dp. JO [0
u = surface tension J; -P+I ; v1dvI+j; dio=O ...................... (A-l)

Subscripts gives
bh = bottomhole
d = diffuser RTo In(p;lpo) = IhPlov1n + IhKnzPlov1n, ............ (A-2)
f = friction but vln=vlo ..................................... (A-3)
i = entrance to nozzle
and RTo=Po/Plo' ................................. (A4)
m = mixing loss
min = minimum Hence, Po In(p;lpo)=(l +Knz)Z. .................... (A-5)
n = nozzle
o = exit of nozzle, throat inlet Energy Equation for the Suction Port.
opt = optimal Ps=Po+ IhP2v~o + IhKenP2v~, .................... (A-6)
s = suction port
but v2o=(Fqo/FAa )Vlo ............................. (A-7a)
t = throat exit
v = vapor and P2o=FpPlo' ................................. (A-7b)
wh = wellhead Hence, Ps-Po=(1 +Ken)(FpF~o/FA~)Z, .............. (A-8)
1 = power fluid
2 = well fluid Momentum Equation for the Throat.
3 = mixture of power and well fluid wlvlo +w2 v 20 -(WI +W2)V3t-IhKt~tP3tV~t=At(pt-Po)'
Superscript .................................. (A-9)
z = zero mixing loss
WI+W2 wI(l +FpFqo)
but P3t=---
Acknowledgment. qlt+q2t qlt(l +Fqt ) l+Fqt
I acknowledge Bill Bartlett and Donn Pedersen of Mobil E&P Serv-
ices Inc. for introducing me to this problem. I am grateful to Peder- qlt+q2t qlt(l +Fqt ) (Po/Pt)Qlo(l +Fqr)
sen for providing me with useful references and information on field and V3t=
At An(l/FAn) An (l/FAn )
conditions and for many helpful discussions. I thank Mobil R&D
Corp. management for permission to publish this paper. =FAn(po/pt)(l+Fqr)Vlo' .................... (A-H)

418 SPE Production Engineering, November 1991


Hence, Author
Pt -Po =Z[2(1- FAn)(FpFio IF1a) -(2 + Kth)F1n
DlmHrl Hablavramldls Is a research
x(l +FpFqo)(P oIPt)(1 +Fqt )+2FAn ]. .............. (A-12) associate with Mobil R&D Corp. In Dal
las. He previously worked for Arco 011 &
Energy Equation for the Diffuser. Gas Co. In Plano and Shell Development
Co. In Houston. From 1981 to 1984, he
ddp was assistant professor of chemical en
I +IdV3dV3+rddi=o, ..................... (A-13) glneerlng at the Illinois Inst. of Technol
t P3 t t ogy. He holds a BS degree from the Natl.
Technical U. of Athens, Greece, an MS
but degree from the U. of Manchester, Eng
P3=PI(1+FpF qo )/(1+Fq) ...................... (A-14a) land, and a PhD degree from the U. of
illinois, all In chemical engineering.
and V3d =q3dlA d = FAdFAn (1 +Fqd)Vlo' .............. (A-14b)
Hence,
FpFJO]
+--Z - - -Po
I n(Pt)
- -Plo(J +m)=O, .......... (C-2)
FAd Fqo Po
where J and m are the friction and mixing losses at the throat,
respectively.

Fqo(1+Fqt)]-~ln(Pd).
The momentum balance for the throat can be written as
x(1+Fqd)2-Kdi F1n Po
Pd Fqo Pt Z [2FpFJo 2 Fqo
Pt-P o= - ----(2+Kth)FAn(1+FpFqo)-
................................. (A-15) Fqo I-FAn F qt
The above derivation used the following relationships:
Polplo =plpl' ................................. (A-16a) X(l+Fqt)Fq~+2FAnFqoJ. ........................ (C-3)

pql =Poqlo' .................................. (A-16b)


Combining Eqs. C-3 and C-4, and in view of
and PoIFqo=pIFq . .............................. (A-I6c)
KthF1n (1 +Fqt )(1 +FpFqo)Fqo=PloJ' .............. (C-4)
Appendix a-Governing Equations for the we obtain the following expression' for the mixing losses:
Llqul~/Gas Pump (after Cunningham'
___ Z_[ _ 2 ( Fqo)2 2
Pi-Po=(I+Knz )Z, .............................. (B-1) m- I 2FpFqo+FAn (1 +FpFqo)(l + Fqt )
PloFqo F qt
Po In(pslpo)=(1+Ken)(FpFioIF1a)Z, ............... (B-2)
Pt-Po=Z[2FAn -(2 +K th )F1n (1 +FpFqo)(1 +Fqt ) F 2F F2 F3 F F3
+2F z ~(l+F F )(1+F )F _ p An qo +~
+2FpFioF1nl(I-FAn)], .......................... (B-3) An F p qo qt qo I-F FZ
qt ~ ~
and prPt=Z(1 +FpFqo)[F1n (1 +Fqt )2 -F1dF1n (1 +Fqd)2
-KdiF1nFqo(1 +Fqt)]-PoFqo In(Pdlpt) ............. (B-4) _~ln(Pt)] . .............................. (C-5)
PloFqo Po
Appendix C-Energy Analysis for the
Gas/Llquld Pump For the ideal pump (negligible friction and mass-flow-rate ra-
The total energy balance can be written in the form tio), Eq. C-5 becomes

dp dp 2 2
wI-+ w2-+ w ldv I +w2dv 2 +(WI +w2)d=O . ..... (C-l)
ZF1n [( I
m= PloF FAn -Fqo
)2 -
(Fqo
F
)Z] Po (Pt)
- PloF In Po .
PI P2 qo qt qo
.................................. (C-6)
Eq. C-l, when integrated from the entrance to the exit of the
throat and rearranged, leads to
SPEPE
Fq Original SPE manuscript received for review Oct. 9. 1989. Paper accepted lor publication
Pt-P o =Z- [ I-FAn
2 ( - o)2 (1+Fqt)Z(1+FpFqo) April 3. 1991. Revised manuscript received May 10. 1991. Paper (SPE 19713) first presented
Fqo F qt at the 1999 SPE Annual Technical Conference and exhibition held In San Antonio, Oct. 8-11.

SPE Production Engineering, November 1991 419

You might also like