You are on page 1of 10

SPE

SPE6766
COMPARISON
OFCALCULATED
ANOOBSERVED
PRESSURE
MIXTURES
DROPSINGEOTHERMAL
WELLSPRODUCING
STEAM-MATER
by R.N. Upadhyay, J.D. Hartz, B.N. Tomkoria andM.S. Gulati, Members
SPE-AIME, Union Oil Company of California

@ Copyright 1977. American Institute of Mining, Metsllurgicsl. and PeIroleum Englneera, Inc.
Tfrie Paper was presented al trw 52nd Annual Fall Tectmwsl Conference and Exhibiion of the SOCielyof Petroleum En Ineere o AIME. held m Denver, Colorado. Oct. 9.12.1977. TIIJmaterial ia subject to
Mrrectlon by the auihor. Permleslon to copy Ie reatricled to an abatract ot not more then 300 words. Writs 620 1! N. CenI al ExfIy. Dallas, Texas 75206.

ABSTRACT
.
This paper contains comparisons of calculated andobserved flowing pressure profiles from geothermal wells
located in the United States and the Philippines. Comperisons are Included for tubular flow as well as flow
through the casing-tubingannulus, Our comparison
shows that for tubular flow, the Orkiszewski correlation makes the best prediction,whereas for annular
flow, no clear-cut choice of a correlation can be made.

wellbore heat loss-was used to calculate pressures in


a number of geothermal wells in which flowing pressure
surveys had been run. The paper contains comparisons
between the calculated and the observed pressure
profiles.
This comparisc:lencompasses a wide range of flow rates
and wellhead pressures, and includes tubular as well
as amular (between casing and tubi
flow. The three
v
correl tions used are: Orki zewski;
Hagedorn and
Brown;!!and Beggs and Brfll.3

INTRODUCTION
The capability to accurately predict flowing pressures
in a geothermal well producing steam-watermixtures
under various operating conditions is of value for
several reasons: general engineering essential to
evaluation of the geothermal reservoir and proper reservoir management; optimization of wellbore design from
well deliverabilityconsiderations;and minimization of
scale deposits in the wellbore.
This predictive capability is especially important because of the difficulty of running flowing pressure
surveys in geothermal wells. These wells are characterized by very high fluid velocities,which sometimes make it impractical for the pressure recorder to
traverse downward in.the well. There have been cases
of pressure recorders thrown out of the wellbore due
to high fluid velocities.

The phase behav or relationshipsused in this work are


for pure water,1 and do not include the effects of
dissolved salts. However, the waters in the wells we
studied were of low salinity. Also not considered is
the effectof non-condensablegases present in the
fluid. We do not ccnsider this to be a significant
limitation, because in otirobservation wells, noncondensable gases,consistingalmost entirely of
carbon dioxide, constituted a small fraction of the
steam phase.

This comparison of computed and observed pressure


drops in flowing geothermal wells can help determine
the degree of confidence an engineer should have in
results predicted by the three correlations evaluated.
The best correlation does a satisfactory jobof predieting pressure drops, and can be used in deliverability prediction calculations. Optimization of
wellbore design of ?uture wells in a partially deIn calculating flowing pressure profiles for oil i~ells,
veloped field can be accomplished by calculating prophase transfer between oil and gas requires a !Zher
duction rates for different flow string diameters ata
simple treatment, and is accomplished through the use
given wellhead pressure, and comparing the benefit of
of solution gas-oil ratio relationships. In geothermal increased flow rates against the higher cost of drillwells, however~ phase transfer between water and steam
ing and completing larger diameter wells.
attains critical importance, and calculationsmust incorporatethesteam tables accurately. Pressure profile Since the precipitationof calcium carbonate scales,
calculations for geothermal wells vary from those for
encountered in many hot water wells, is related to the
oil wells in another important aspect in that the tempresslreand temperature conditions in the wellbore,
perature of the fluid must be computed precisely.
calculations can be made to estimate the depth atwhici
scale precipitationwould cormnencefor various wirllbors
For calculations included in this paper, a computer
diameters and mass flow rates. This can assist the
program incorporatingthree previously published corengineer in the selection of operating conditions that
relations for predicting two-phase flow in vertical
will tend to cause scaling at shallower depths, thus
pipes-coupled with equations for phase transfer and
requiring easier clean-up operations
*

References and illustrations at end of paper.

COMPARISON OF CALCULATED AND OBSERVED PRESSURE DROPS


-1
IN GFQ&&ll
PRODQ NUIE8UAU3
MIXTURES

2
CORRELATIONSFOR TWOPHASE VERTICAL FLOW

1488 PLdvsl
Re =

Our calculationsof flowing pressure profiles were made


with three correlations:Orklszewski; Hagedorn and
Brown; and Beggs and Brill, Eachof these is awt?llknown correlation for two-phase vertical flow. Me
chose the first two correlations because they have been
widely used by petroleum engineers, and have, in our
own experience, done a satisfactoryjob of predicting
pressure drops in ofl wel1s. W! Included the Beggs and
Brill correlation fn our study hecause it is probably
t4emost recently published method.
.
Ti@ following is a brief description of these three
correlations;for a more detailed analysis, the reader
?hould refer o te original articles describing the
correlatdons~,2,~
Orkiszewski Correlation
This correlation takes Into consideration the slip be?%een liquid and vapor phases, and therefore includes
a correlation to calculate liquid holdup, I.e., the
fraction of the Rir)ecross-sectionalarea occu~ied by
liquid. It also-takes into account the flow regime:hat exists fn any section of the pipe; the flow
egimesconsidered are shown in FIGURE 1. The
lrkiszewskicorrelation is hybrid In that it uses d~$Ferentpublished correlations5fordifferent flow
*egimes: Griffith and41allis correlation for bubble
Flow regime, and Duns and ROS6 correlation for transition and mist flow regimes. For slug flow, Orkiszewskl
ievelopeda new tiorrelation2based
upon the experimental
iata of Hagedorn and Brown. Orkiszewski defines the
fariousflow regimes by the following imits:
Vs9/vm

slug flOW:

V$gjvm > B$ N~v < Ls

Wansition flow:

LM > NGV > LS

f@sl
Pm =

=!jo+36

LM =75+84

.(3)

. .

.(4)

Cl is a function of bubble Reynolds Number, NR b, and


C is a function of both N eb and liquid Reynofds
N6mber, NReL, defined belo!.

iReb=

1488pLvbd
ML

. . . . . . . . . . . . ...(5)

1488 pLvmd
ReL=vL

*(6)

Becauseof the interdependenceof v and Reb, the


calculation of vb requires an itera!ive procedure.
Forvm<
10,
...
6 = (0.013 log pL)/d138 - 0.681 + 0.232
. . . . . . . ...(7)

- 0.065 Vm

>10
0.799

(&tt4!i10fJvL)/d

- 0.709-

-v
with the limit d ~vm + bVb

(NLv)o75

. . . . . ...(8)

(1 - m/pL)

Pressure drop due to friction is given by

l+b[
s

()

l+ m
Vs /

2
-4v#$

. . .(1)

*
dzf=~

The pressure drop due to friction is given by


fpL(vs#HL)2

2.
pLvm
[(

+V
s1
vm+v;+~

)]

(g)

The friction factor, f, in equation (9) is calculated


from the Moody diagram using the following definition
of Reynolds Number:

In equation (l), vs is assumed to have a constant valu{


of 0.8 ft/sec.
Re =

dz f

LV

()

9+PL6

0.162 log m -0.88810gd


.

Liquid holdup in this flow regime Is given by the following equation:

!!E.

Vm + Vb

where vb = $C2

lubble Flow

HL=l.;

+Vb)+pvs

with the limit d >

Z 0.13

Orkiszewski uses the following relationshipto calculate two-phase dens~ty in this flow regime;

6=

#here LB = 1.071 - (0.2218 v#d)

Slug Flow

Forvm

GV

The friction factor, f, in equation (2) Is calculated


by using7the Reynolds Nwnber and the standard l~oody
diagram.
In this flow regime, pressure drop due to
acceleration ~s neglected.

< LB

>L

HL VL

log Vm - 0.42810gd

lubbleflow:

\nnularmfst flow:

SPE 6766

1488 pL dvm
PL

. . . . . . . . . . . . . (lo)

Pressure drop due to acceleration in the slug flow


regime is too small, and is neglected.

, . . . . . . . . . ...(2)

2gcd

The Reynolds Number for the calculation of friction


factor is given by the following relationship:

Transition Flow
This flow regime provides a transition from slug flow
to annular mist flow. The pressure drop in this regirn
is calculated by linear interpolationbetween the

R. N. UPADHYAY, J. D. HARTZ, B. N. TOMKORIA, M. S. GULATI

SPE 6766

pressure drops for slug and mist flow regimes. Mixture


density and frictional pressure drop terms are calculated for both slug flow and mist flow; they are then
linearly weighted with respect to a dimensionlessgas
velocity term and the limits of the transition zone in
waler to arrive at the terms applicable to this flow
regime.

pL
~
()

ND =120.872d

1/2
. . . . . . ..(17)

Liquid iscositynumber,
L

= 0.15726vL

1
pLU3
()

1/4
. . . . . . .(18)

!listFlow
The calculation of liquid holdup using this correlation
requires a three-step process: (1A From the liquid
viscosity number, NL, a quantity C L is eval ated;
(2) The value of CNL, together with NLV and NGVS is
used to calculate HL/v, i.e., li uid holdup divided by
a secondary correlation factor; ?)
3 An expression containing Nv, NL and ND is used to evaluate the secondary corre?ation factor, $, and thence HL.

In this flow regime, the gas phase is continuous, and


is the controlling factor. Because of high gas flow
rates, it is assumed that there is no slip between gas
and liquid phases, and the mixture density is given as
D

.pL

sl+
~
m

p
9

!s!4

........

. .(11)

The frictional pressure drop is based on gas phase


cmly, and is given as
9
(+)=%

~m=pLHL+@l-HL)

. . . . . . . . . . . . ..(12.

The friction factor. f. is obtained from the Moody


diagram and the gas-Reynolds Number, defined belo~:
1488P
Re =

Vs

d
(+)
f=

. . . . . . . . . . . .(13)

Amodified formof relative roughness factor (e/D) is


calculated for use with the Moody diagram; ~his.is done
to take into account the effect of the liquld film on
the pipe. The modified e/D is limited to values between
10-3 and 0.5, both inclusive.
Pressure drop due to acceleration can be given by the
following equation:
g=
()

..

. . . . . . . . . ,0

. .(14)

..,.,

. . . ...19)

IThe frictional pressure drop is given by

Pg

From the calculated value of liquid holdup, the mixturetlensitycan be calculated using the following
equation:

fw2

. . .(20)

7.116x1012xpmxd5

In equation (20), the density, pm, is based upon


liquid holdup. The friction factor, f, is read from
the Moody diagram using a two-phase Reynolds Number
defined below:
1488 pmvmd
Re

. . . . . . . . . . . . .(21)

Vm

where

Vm = Vsl + v
Sg
..
L (l-HL)
and Pm = ~L ~g
.

acc

The pressure gradient due to acceleration is given by


Haaedorn and Brown Correlation
This method considers slip between liquid and gas
phases, but does not divide flow into different
regimes. The correlationwas developed from a study
of flowing pressure gradients in a 1500-foot experimental well. Pressure drops were measured for twophase flow through l-in., 1 l/4-in., and 1 l/2-in.
nominal diameter tubing. Although liquid holdup was
not measured, it was calculated to satisfy the total
pressure drop measured after the contribution of friction and acceleration had been calculated. These calculated values of liquid holdup were then correlated
with flow rates, pipe diameters, and fluid properties.
It was found that the li uid holdup is related prj=
marily to the following ?our dimensionless terms:
Liquid velocity numberw
LV

=1.938vs1

pL
~
()

1/4
. . . . . . .(15)

Gas velocity number,


GV

=1.938vsg

Pipe diameter number,

pL 1/4
~
()

. . . . . . .(16)

()g

acc.

Pm
~dz

d(vm2)

. . . . . . . . ..(22)

Beggs and Brill Correlation


This is the newestof the three correlations. Like t~
Orkiszewski correlation, it takes into account thesli
between liquid and vapor phases, and recognizes the
existence of different flow regimes. The correlation
contains a relationship to calculate liquid holdup at
all pipe angles. Beggs and Brill developed their correlation from flow experiments in 90-ft long acrylic
pipe, 1 in. and 1.5 in. in diameter,whichcould be
inclined at any angle.
This correlation assumes t,?flow in a horizontal pip~
to exist in one of three major flow patterns: segregated, intermittent,and distributed, as shown in
FIGURE 2. Even for non-horizontalflow, the method
Tirst calculates the flow regime that would exist if
the flow were horizontal. Separate equations are use(
to calculate the liquid holdup for different horizontal flow regimes; the liquid holdup is then corrected for pipe angle of deviation from horizontal.

COMPARISON OF CALCULATED AND OBSERVED PRESSURE DROPS


TN
-.. GEOTHERMAL
--- . ... ... . .- WELLS
------ PRODUCING
. .------ .- STEAM4JATER
- .-. .-. ... . . . . MIXTURES

rhe horizontalflow regime limits are defined by the


%ouds Number (NFR) and various functions of AL where
Vz
FR =
wrd

~L.r

s1
m

rhe horizontalliquid holdup is corrected for pipe


inclinationby u~ing a multiplier, which, for vertical
Flow, reduces tfl(1 +0.3C). Values of C differ with
the calculated horizontalflow regime, and are different fo uprril
1 and downhill flow. (Refer to Beggs
ind Brill$ for calculation of C.)
rhe frictional pressure gradient is given by
ft pn

(Q
Idz)f

VM2
. . . . . . . . . . . .(23)

2gcd

#here Pn pLAL + Pg(l - AL)


ind ftp = fn es

SPE
6766
-.
----

b. From the calculated pressure gradient P, and


pressure increment Ap, the depth interval AZ iv
calculated. Heat loss from the well-boreinto the
surrounding formation over the depth interval AZ is
now calculated using a transient heat transfer relationship. This takes into account the time for which
the well has been producing, and the overall heat
transfer coefficient for the wellbore. Energy losses
due to potential energy and kinetic energy changes are
calculated and combined with the heat loss calculated
.!bove. A new fluid enthalpy IS thus computed. Using
this new enthalpy, and pressure p + Ap, the steam
fraction at depthAZ is calculated. If the difference
between calculated and assumed steam fractions is
within a specified tolerance, calculation proceeds to
the next pressure increment. If not, the average of
the two is assumed to be the steam fraction for the
second iteration. This iterative procedure is continued until assumed and calculated steam fractions
converge. The pressure gradient calculated in the
last iteration is used to calculate the depth incremer
AZ corresponding to the pressure increment Ap. For
single-phaseflow, iterations are made to achieve flui
temperature convergence rather than steam fraction
convergence.
3.

rhe non-slip friction factor, f , is calculated from


the Moody diagram using the fol~owing definition of
?eynoldsNumber:
1488 On vmd
!?en=pLAL+p

(l-AL)

(24)

The exponent s is a function of input liquid fraction,


wrd the calculated liquid holdup, corrected for vertical flow.
~ressure gradient due to acceleration is given by
f~

[pLHL + Pg(l - ~1 mvsQ . ~

~dz)ace.

9C P

. . . @)

COMPUTER PROGRAM FOR


CALCULATINGPRESSURE PROFILES
The computer programwe used in this study follows the
procedure outlined below in calculating the pressure
profile in a flowing geothermal well:

The pressure is increased again by increment Ap,


and the correspondingAZ is ca?zulated following
the procedure outlined in step 2 above. This procedure is continued until the total depth of the well
is reached.

Properties of pure water and steam were used in our


calculations. Improvements in calculations can be
made by taktng into consideration the effects of salt
concentrationon the phase behavior of water, and alsc
the effect of non-condensablegas content.
For heat transfer calculations, ground temperature ane
a geothermal graditmt are used; this gradient is used
to calculate the rock temperature up to the to of thf
producing formatior. The temperature within t!e producing formation itself is assumed to have no vertical
gradient. This has been found to be generally true ir
our experience.
The computer program calculates heat transfer from thf
wellbore into the surrounding rock formation using the
following relationship:
UndAZ(Tf- Tr

Q=-mm+

(26

1. Wellhead flowing pressure (p), and steam and water


flow rates at wellhead conditions act as the start- Where
c
ing point. (Althoughbottomhole conditions can just as

- 0.29
f(t) = - In
easily be treated as the starting point, the informa2m
()
tion is usually available for wellhead conditions, and
needs to be calculated for bottomhole conditions.)
Ramey8 has found equation (26) to be valid for flow
periods in excess ofa week. Because of high flow
2. A small pressure increment (AP) is selected, and
rates encountered during our surveys, heat tra?isfer
the depth interval (AZ) over which this increment
between the wellbore and the surrounding rock did not
would occur is calculated following these steps:
affect results appreciably. We used a heat transfer
coefficient of 10 Btu/(hr. F ftz) in ourcalculations.
a. Anew steam fraction (i.e., fraction of total
mass flow rate that is steam) is assumed to exist at
p + Ap. Using an average pressure and average steam
OBSERVED PRESSURE PROFILES
fraction over the interval, the s ecified two-phase
flow correlation is used to calcuYate the pressure
Outflowing surveys ~ere conducted during tubular as
well as annular flow; the annular flow surveys will b[
gradient (Ap/AZ) over the interval. If flow is in
single phase, the pressure gradient is calculated by
discussed later. For tubular flow, the typical wellusing single-phasedensity and Reynolds Number, and
bore configuration co~sisted ofa 9 5/8-in. casing in

the Moodydiagram.

the upper halfof

the well, and a 7-in. slotted

liner

SPE 6766

R.

N.

UPAONYAY.

J.

D.

HARTZ.

B.

N. TOMKORIA.

M.

S.

GULATI

in the bottom half. A typical wellbore dfagram appears The annular surveys can be generally classified into
two groups: One wfth wellhead pressures close to 135
in FIGURE 3.
psfa,the other wfth wellhead pressures fn the vicinfty
of 220 psia. Wellbore and flow rate data pertfnent to
Flowing wellhead pressures in these surveys varfed between 39 and 160psfa; measured bottomhole pressures
these surveys are summarized in TABLE I (surveys 6
?nd 7). Two comparisons of observed and calculated
between 240 and 1100 psia. Mass flow rates ranged bepressure proffles, representing the two survey
groups,
tween 90,000 lb/brand 389,000 lb/hr. Steam fraction
in FIGURES 9 and 10.
appear
at wellhead conditions varfed from 0.15 to 0.46 by
wefght. The shallowest survey was 1600 ft deep; the
deepest, 5000 ft. Wellbore and flow rate data for
In calculating the annular flow pressure proffle, we
used the hydraulic dfa!neter,deffned below as the pipe
these surveys are swmnarized in TABLE I (surveys 1
dfameter.
through 5).

The total dissolved solfds In the produced fluids


varfed from a low of 6100 ppm to a high of 9700 ppm.
Non-condensablegases, composed almost entfrely of
carbon dioxide, ranged from a low of 0.2% to a hfgh of
4.3% by weight of steam.

Hydraulic diameter,
d
h

= 4 x cross-sectionalarea of flow
.
wetted perimeter

If the casing insfdedfameter is dl and tubing outsfde


dfameter fs dz,

COMPARISON OF OBSERVED AND


~LCULATED PRESSURE PROFILES

4(m/4)(d12 - d22) = d
h =

The comDarfson otobserved and commited Rressure mofiles appears in FIGURES 4 throu~h8. In FIGURE ;, it
can be seen that Kbserved
pressure pro~rts
devfating from the calculated values at-a depth of
1500ft; this devfation becomes considerable below 2001
ft. There fs a sfmple explanation for this, however.
The observed pressure gradfentin this well below 2000
ft fs 0.376 psi/ft, and temperature surveys indicate
the reservofr temperature to be 383F. From s earn
tables, water densftyat 383F fs 54.76 lb/ft,J which
results in a statfc pressure gradfent of 0.380 psi/ft.
This shows that the observed pressure gradfent below
2000ft fs a statfc gradfent, and that there fs no
flow enterfng the wellbore below that depth. Between
1500 ftand 2000 ft depth, fluid enters the well at
different points, and therefore there fs divergence
between the calculated and the observed pressures.

- d
1

.(27)

For a 9 5/8-in. nomfnal casfng and 2 3/8-in. nominal


tubfng, dh is calculated to be 0.545 ft.
The hydraulic diameter was used fn calculating Reynold
Number and pipe relative roughness. For calculating
fluid velocftfes, we used the actual cross-sectfonal
area of the annulus.
An examination of FIGURES 9 and 10 shows that no clear
cut chofce can be made as to the best correlation for
annular flow. Thfs statement can bemade for all the
annular flow pressure surveys conducted by us. We
find that whfle Orkiszewski and Hagedorn and Brown
tend to under-predfct the pressure drop, the Beggs and
Brfll correlation has a tendency to over-predict ft.
However, since all the annular flow surveys were run
in a sfnglewell, we donotknow ifour ffndfngwfll
be generally true. We ffnd ourselves in agreement
wfth the conclusion of Sanchez9 that no sufficiently
accurate and precfse method exists for annular flow.

An examination of FIGURES 4 through 8 frmnedfatelyreveals that the Orkiszewskf correlation calculates pres
sures that sireclosest to the observed pressures; the
In
Hagedorn and Brown method fs a close second.
FIGURES 6 and 7, both Orkiszewskf and Hagedorn and
Brotinpredfct approximately the same pressure profile;
howe)~er,fn FIGURE 6, it is noteworthy that below 4500
ftthe Orkfszewskf correlation follows a change in the
observed pressure gradient, whereas the Ha edorn and
Brown correlation does not. In our work, tf e Beggs
and Brtll correlation did not do a satisfactory job
of predicting pressure proffles in goethennal wells
producing steam-watermixtures.

CONCLUSIONS
Based upon the limfted number of flowing pressure surveys conducted, we come to the following conclusions:
1.

For predicting flowing pressure proffles fn geothenoal wells producing steam and hotwatermfxtures, the drkiszewski correlation does a satisfactory
~~@n~he Hagedorn and Brown correlation comes a close
.

It should be noted that in our calculatfonof pressure


proffles, we assumed that all the fluid flows through
the inside of the slotted liner, i.e., there fs no
flow through the open hole-liner annulus.
ANNULAR FLOW

~(dl + dz)

2.
I

We ran several pressure surveys fn swell fn whfch flow


took place through the 9 5/8-fn. casing2 3/8-fn.
tubing annulus. The tubing was closed at the top and
open at the bottom; ft had several perforations at
dffferent intervals to allow pressure conmwnfcation between the casing and the tubing. A bottomhole pressure
recorder was moved up and down the tubing to conduct
the surveys. Although some flow obviously occurred
through the tubing because of the perforations,we feel
that ftwas not an appreciable part of the total flow,
and we assumed the total flow to be through the
annulus.

The Beggs and Brfll correlation did not adequately


predict pressure loss fn these flowing geothermal
wells.
For annular flow in geothermal wells, we cannot
make a definfte chofce of a correlation. Our resuits show tk~t whfle the Orkiszewski and the Hagedorn
and Brown correlations under-predict the pressure drop
the Beggs and Brfll correlation over-predicts ft.

4.

The Orkiszewski correlation has been used wfth


success in predicting geothermal well deliverabflfty under different wellbore desfgns. Together
with appropriate treatment of solubflity characterfstics of C02 in water, the Orkiszewski correlation

COMPARISON OF CALCULATED AND OBSERVED PRESSURE DROPS


IN GEOTHERMAL IIIELLS
PRODUCING STEAM-WATER MIXTURES

can be used to approximate the depth at which scale


deposits will take place in swell as a function of
wellbore diameter.

b
m
s

5.

Our calculations did not include the effectof


dissolved solids in the water, nor the noncondensable gas content of the steam. However, the
wellbore fluids in our wells were not ligh in dissolved solids content.

Sg

Parameters used to calculate bubble rise


velocity in the Orkiszewski correlation

6 = Liquid distribution coefficient


AL = Input liquid fraction, or no-slip liquid
holdup
Vg = Gas viscosity, cp

= Hydraulic diameter, ft

~L =
=
m
=
9
pL =
pm=

Absolute pipe roughness, ft

f = Darcy-Weisbachor Moody friction factor


fn = Non-slip friction factor
f(t) = A function of time, t (days), since well was
open to flow

$=

9C = Conversion constant, 32.2 lbm-ft/lbf-sec2


HL = Liquid holdup, fraction
= Bubble-slug boundary term

B
LM = Transition-mistboundary tem
LS = Slug-transitionboundary term
ND = Pipe diameter n!imber
= Froude Number

iv l~~~u~l~~~~s~~b~~mber
L
= Liquid Velocity Number
LV
= Reynolds Number
Re
= Bubble Reynolds Number
Reb
= Liquid Reynolds Number
Rel
= Non-slip Reynolds Number
Ren
P = Pressure, Psf
d
= Acceleration pressure gradient, psf/ft
(3)
z acc

()

U = Overall heat transfer coefficient,

Btu/(hr-ft2-

F)

Secondary correlation factor

1.

Orkiszewski, J.: PredictingTwo-Phase Pressure


Drop in Vertical Pipe, J. Pet. Tech. (June,
1967) 829-838.

2.

Hagedorn, A.R, and Brown, K.E.: Experimental


Study of Pressure Gradients Occurring During Continuous Two-Phase Flow in Small Diameter Vertical
Conduits, J. Pet. Tech. (April, 1965) 475-484.

3.

Beggs, H.D. and Brill, J.P.: AStudyofTwoPhase Flow in Inclined Pipes, J. Pet. Tech. (May,
1973) 607-617.

4.

Argonne National Laboratory: STEAM67--A


Subroutine package incorporating1967 ASME Steam
Tables.

5.

Griffith, P. and Wallis, G.B.: TwoPhaseS lug


F#w~oJ. Heat Transfer; Trans., AIME (Aug., 1961)
-.

6.

Duns, H., Jr. and Ros, N.C.J.: Vertical Flow of


Gas and Liquid Mixtures from Borehole, ProC.,
Sixth World Pet. Congress, Frankfurt (June 19-26,
1963) Section II, Paper 22-PD6.

t = Time since well was open for flow, days


Tr = Rock temperature surrounding a segmentof
wellbore, F

Mixture density, lbm/ft3

REFERENCES

rc = Outer radius of casing, ft

Tf = Fluid temperature in a segmentofwellbore, F

Liquid density, lbm/ft3

We thank the Management of Union Oil Company of California for permission to present this paper. The computer program used in our work was originally written
at the Petroleum Engineering Department of the University of Tulsa under a contract from Union Oil
Company of California.

s = Exponent used in relating no-slip friction


factor to two-phase friction factor

Mixture viscosity, cp
Gas density, 1bm;ft3

ACKNOWLEDGMENTS

= Frictional pressure, gradient, psf/ft


Q = Heat lost to surroundings,Btu/lb

Liquid viscosity, Cp

= No-slip mixture density, 1bm/ft3


n
a= Interracial tension, dynes/cm

= Two-phase friction factor


ftp
9 = Acceleration due to gravity, 32.2 ft/sec2

= Superficial gas velocity, ft/sec

a = Thermal diffusivity of earth, ft2/day

= Casing inside diameter, ft


1
d2 = Tubing outside diameter, ft

FR

= Slip velocity (differencebetween average gas


fthec
and liquid velocities),

Greek Symbols
.

d = Pipe inside diameter, ft

e =

= Mixture velocity, ft/sec

Z = vertical length, ft

CNL = A function of liquid viscosity number, NL

= Bubblerise velocity, ft/sec

= Superficial liquid velocity, ft/sec


s1
W= Mass flow rate, lbm/hr

~OMENCLATURE

c,, C2 =

SPE- 6766
-.
- . --

SPE 6766

R.

N.

UPADHYAY.

J.

D. HARTZ.

7. Moody, L.F.: Friction Factors in Pipe Flow,


Trans., ASME (1944) 66, 671-684.
8.

E.

9.

Ramey, H. J., Jr.: Wellbore Heat Transmission


J. Pet. Tech. (April, 1962) 427-435.

N.

TOMKORIA.

M. s

fMUATT

Sanchez, M. J.: Comparisonof Correlations for


Predicting Pressure Losses in Vertical Multiphase Annular Flow, M.S. Thesis, The University
of Tulsa, 1972.

TABLE I
WELLBORE AND FLOW RATE DATA FOR PRESSURE SURVEYS

Survey

Tubular Flow
3

Annular Flow

~
138

Flowing Wellhead Pressure, psia

160

129

80

39

151

Total Mass Flow Rate, lb/hr

281,174

271,3$5

89,608

120,475

389,000

149,700

133,700

Steam Fraction at Wellhead

0.162

0.157

0.463

0.149

0.211

0.643

0.530

Casing Inside Diameter, inches

8.921

8.921

8.801

8.758

8.921

8.921

8.921

219

Casing Depth, feet

1621

1621

2290

1783

1621

2371

2371

Liner Inside Diameter, inches

6.276

6.276

6.276

6.276

6.276

2.375a

2.375a

Liner Depth, feet

4976

4976

6060

4805

4976

2371b

2371b

Formation Temperature, F

540

540

540

383

540

525

525

a Tubing outside diameter. Flow takes place in the annulus between casing and tubing,
b

Tubing de ;h.

BUBBLE

SLUG

TRANSITION

ANN-ULARMIST

Fig. 1 - Flow reg~mes for the Orkiszewski correlation.

WELL BORE DIAGRAM

SEGREGATED FLOW

DEPTH
.

373

WAVY

INTERMITTENT FLOW

PLUG

1645
[

2290

25254

DISTRIBUTED FLOW

BUBBLE

MIST

2- FLOW REGIMES

FOR THE
BEGGSANDBRILL CORRELATION.
FIG,

6060
FIG,

3-

A TYPICAL

K/
GEOTHERMAL

WELLBORE

DIAGRAN,

PRESSURE,
.200

PSIA

PRESSURE,

400

600

200

OBSERVED
ORKISZEWSKI
HAGEDORN
BEGGS

PSIA
400

600

OBSERVED

x ORKISZEWSKI
b HAGEDORN

&BROWN

o BEGGS

& BRILL

& BROWN

& BRiLL

500 -

1000 1=
&
.
*
&
~
1500 -

An
x

2000

A
.nAO
x4ifJ

L ,;
A

XAO

xAo

XAoo

F19.4- Comparison of calculated

2500
Fig. 5- Crmparlson of calculated
for survay 2.

and observed preesura profi Ies

for survey 1,

and obsarved preesure profiles

PRESSURE, PSIA
200

600

OBSERVED
x ORKISZEWSKI

& HAGEDORN
1

200

O* ~

P
10

PRESSURE, PSIA

400

& BROWN

u BEGGS & BRILL

50CJ -

400

600

OBSERVED
x ORKISZEWSKI

A HAGEDORN
& BROWN
o BEGGS & BRILL

%.

All

~.

0
x

An

20
G
n
.
~

1000 -

+
&
. .
z
1&
a

*R
0
n

%
Xn
Aa

B
30

AO

0
A

1500 -

x
.0

ho

~n

AO

&o o

40

50
Fig. 6- Ikqarison
for survBy 3.

of calculated

and obssrved pressure profilae

Fig. 7- ikqarieon
for survay 4.

of calculated

and obsarved pres.wra profiles

PRESSURE, PSIA
*O

.200

400

OBSERVED
x ORKISZEWSKI

= BEGGS & BRILL


A HAGEDORN
& BROWN

<0

500 -

600

dAn
AQ

*AD
*D
A

1000 *
u.

so

~tl

F
a.

Xo
A

XJ

1500 -

%AA :

2000

Fig. 8 -Comparison
for aurvay 5.

of calculated

andobsarvad

preasure profiles

PRESSURE, PSIA

200

o+-

400

2oo-

400

%
a

HAGEDORN

BEGGS

600
I

OBSERVED
x ORKISZEWSKi

500

PR~SSURE, PSiA

400

OBSERVED

x ORKISZEWSKI
& BROWN

Q BEGGS & BRILL


A HAGEDORN
& BROWI

& BRIL1
50

A *D

.0
0
A

Xo

o
Ax

100
!4
U.

.J
Xo

AO
AO

x
150

AO

x
A

Xo
A

200C

200

x
Ax

.0
AX

Fig. 9- kqarison
for aurvdy 8.

of calculated

and observed praaaura prfiles

Fig. 10- kwparison


for survey 7.

of oalculatad

and obsarvad prassura profilas

You might also like