SPE 48937 Effect of Completion Geomety and Phasing On Single-Phase Horizontal Wells Liquid Flow Behavior in

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

rI

SPE 48937 Effect of Completion Geomety Horizontal Wells


*Nowwith Dwights **NowwithThe Pennsylvania State University
Copyright 1998, Swlety of Petroleum Engtneers, lnc This paper was prepared for presentation at the 1998 SPE Annual Technical Conference and Exhibition held in New Orleans, Lou!sianat 27-?0 September 19%

Societyof PetroleumEngineers

and Phasing on Single-Phase

Liquid Flow Behavior in

Hong Yuan*, Cem Saricati, and James P. Brill, SPE, The University of Tulsa,

This paper ws sefected for presentation by an SPE Program Committee follmng review of tnfomation contained in an abstract submitted by the author(s) Contenfs of the paper, as presented, have not been revrewed by the Society of petroleum En9inee~ and are subie~0 correction by the author(s) The material, as presented. does not necessarily reflect any position of the Society of Petroleum Engineew i~ o~~rs or membe~ papers presented at SpE meebngs are subject to publication review by Editorial Committ=s of the Sfxiety of Petroleum Engineers Sfectronic reproduction, distribution, or storage of any pat of this paper for commercial purposes without the written consent of the Socie~ of Petroleum Engineers is prohibited Pemlsion to reproduce !n print is restricted to an abstract of not more than 300 wordq dlustrations may not be copied The abstract must contain conspicuous acknuwfedgment of where and by whom the paper was presented Write Librarian, SPE, P O Box 833836, Richardson, TX 75083-3836, U S A, fax 01.972-%2-9435

agreement was found between the friction factor correlations and experimental data. A field example is presented to show the importance of using a proper friction factor correlation to calculate the pressure drop in a horizontal well. --Introduction The flow behavior in a horizontal well, and its interaction with

Abstract

Horizontal wells can have very complex flow geometries, in part due to interaction between the main flOWstre~ ~d the influxes along the wellbore, and aIso due to completion type. An experimental facility was used to investigate the effects of completion geometries and the density and phasing of injection openings in horizontal wells. Three test sections with perforation densities of 5, 10 and 20 shots per foot and. phasings of 360, 180 and 90 were. Four test sections were used for slotted liners, including one single slot case and three multiple slot cases. The numbers of slots for m~tiple slot cases were 18, 18 and 36 on 4-ft, long sections with slot phasings of 360, 180 and 90, respectively. A total of 1,257 tests were conducted for no fluid injection, no main flow at the test section inlet, and wfitilfluid injection for Reynolds numhers ranging from 4,000 to 60,000 and for influx to main flowrate ratios ranging from 1/5 to 1/2000. Experimental results show that completion geometry, phasing and density in addition to Reynolds number and influx to main flow rate ratio have dramatic effects on the pressure behavior and therefore the production behavior of horizontal wells. A general friction factor expression for horizontal wells with multiple injection openings was developed based on the conservation of mass and momentum. A commercial Computational Fluid Dynamics (CFD) computer progrm was used to determine the length of the flow developing region in a horizontal well after the flow is disturbed due to radial influx. Applying experimental data and the result of CFD simulation to the general friction factor expression resulted in new simple correlations for horizontal well friction factors. Very good

the reservoir, has been recognized as one of the most imporWt problern.s.in petroleumengineering. The flow behavior in a horizontal wellbore differs from flow in a regular pipe, The rou~.ess of a horiz.oritd.well Cm bg rnvch w~t!r th~ that of a regular pipe because of perforations and slots. Influxes along the welIbore, which cause changes in momentum transfer and Iaminar sub-layer flow behavior, can change the pressure di~tibutjon. .~erefore, re@ar pipe friction factor correlations should not be used to predict pressure drop in a horizontal well. in a horizontal well_,..dependigg upon the completion method, fluid may enter the wellbore at various locations along the well len~. The distance between perforations or slots may not be stilcient to achieve a stabilized velocity profile pressure behavior than for ..so .and .. .. this ...=. may lead~o @_fferent fully developed flow. The pressure distribution in a horizontal well cw .i@uegce .We well com~letion and well profile design, as well as having an impact on the production behavior of the .well. ~erefore, .both.the pressure drop vs. flow behavior..along the weII Wd the relationship betweenthe pressure drop along the well and the influx from the resefioir should be @vestigated. A Literaturesufie~ shows tkt the flow behavior of porous pipes with fluid injection from the wall and the flow behavior in manifolds, which are similar to a horizontal wellbore flow cofil ration, have been investigated by several investigaP. tors- 1m other disciplines. The results of these studies can not readily be applied to the horizontal wellbore flow cotilgurations since they are- focused on porous walls, transpired bound~ layers and low injection velocities. The petroleum industry started to investigate horizontal wellbore hydradics in the late 1980s, Several investigators2-32 conducted analytical or experimental studies to investigate different aspects of horizontal well flow behavior, Ouyang et al.24 investigated single phase flow behavior in horizontal by Yuans, A wells using an approach similu to hat proposed

H. YUAN,C. SARICA,J.P. BRILL

SPE 48937

wellbore model was proposed and it was claimed that the model is readily applicable to different wellbore perforation patterns and well completions. In their study, a local friction factor correlation was developed based on eight experimental tests for the case of fully developed main flow before the perforated section. It was claimed that the correlation can be used before the perforation point for wellbore flow rates of either zero or non zero. According to Olson and Eckert8, the phenomenon is quite different for the case of zero velocity at the beginning of a porous pipe. Ouyang et al. also stated that for turbulent flow, inflow reduces the wall friction, while other investigators (Kato et al.s, Yuan2932, Yuan et al.30-3) found that inflow could either reduce or increase wall friction depending on influx rates. For a solitary perforation horizontal well, Yuan29 developed a general friction factor expression and a friction factor correlation by applying experimental data to the general friction factor expression to predict the pressure behavior. The literature survey reveals that experimental data are limited and theoretical studies are inconclusive for predicting pressure loss in horizontal wells. In this study, the effect of completion geometries, such as the shape of injection openings (slots and perforations) and the density and distribution of the injection openings, on the flow behavior in horizontal wells are investigated. A general apparent friction factor expression is developed from the conservation of mass and momentum. New correlations are developed for the prediction of apparent friction factors in horizontal wellbores based on theoretical analysis, CFD simulation and experimental data.
Model Development

~= 11(AV2

)J

U2& A

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

where V is the average velocity in cross section A, and u is the local velocity. In the last term of Eq. 1, continuity gives V,=VP. ~P is the momentum correction factor for the influx stream and is defined as: P,= 1/(Apvrv)jAuruxfl,
..............................................(3)

where, u, and u. are the radial and axial velocity components in dA, respectively, and V. will be discussed later. For multiple injection openings, it is convenient to use average properties. The average velocity over Ax is Z and is defined as follows z=(E, +E2)/2 ..........................................# ......................(4)

where Z1 and Z2 are average velocities at the inlet and exit of the control volume, respectively, A mass balance for the control volume is given by:
iilA + nVPAP = F2A ...........................................................(5)

For uniform influx from each opening, the total volumetric influx rate can be given as: Qi~= n~)AP ........................................................................(6)
The average friction factor over a length Ax is defined as:
fT=-([P2-P,]/h)/(pti2 /2/d) .................................(7)

In this study, the approach developed by Yuan29is adopted to predict horizontal well friction factors for different completion geometries. Consider an incompressible fluid flowing Isothermally along a uniformly perforated or slotted pipe as illustrated in Fig. 1. Fluid is uniformly injected through each injection opening area of AP into the main flow stream. When the injected fluid enters the main flow stream through the injection openings, the streamlines change directions. Each local mean velocity is tangent to the streamlines and can be divided into two components, V, and VX,as shown in Fig. 1. The dotted lines define a control volume in which fluid is transported across surfaces A 1, Az, and Aps. Assume there are n injection openings along Ax. The momentum balance for the control volume in the axial direction is:
plA1 p2A2 Zwdb = .........................(1)

The wall friction factor is expressed as: jW = 8rW/(P~2) .................................................................(8) Let

fp=vxlz
p=nl

.............................!..............................................(9)

Ax ....,,,,.,,,,,, ........................................................... (lo)

where q is perforation density. An expression for the apparent friction factor can then be found by substituting Eqs. 4-10 into Eq. 1, rearranging and simpli~ing: ~~ ~~ 2d[~z ~1]/h+2dC~~qifl where c~ = PI +~2
-@#P +n/l(p2 -pl)qjn /D . . . . . . . . . . . . . . . . . . . ..(12)

/~ ....................(l 1)

P2P2~22A2- PIPI~12AI- pPv~v~PPAPn For the three terms on the Iefi-hand side of Eq. 1, it is assumed that average properties completely define the flow field. The first two terms on the right-hand side of the equation use average velocities by introducing momentum correction factors, ~ 1 and ~2, which are defined by the following equation:

The first term on the right hand side of Eq. 11 is the wall friction factor. The wall friction factor definition is due to no momentum change in the control volume. Although the wall friction factor in Eq. 8 can be quite different from that for regular pipe flow, the same form of wall friction factor is adopted here for simplicity, The wall shear stress in Eq. 8 is defined as TW= -P(du / d~)r=d,2. Therefore, f~ reflects the

94

SPE 48937

EffectofCompletion Geomet~and Phasingon Single-PhaseLiquid FlowBehaviorin HorizontalWell

velocity profile change in the radial direction near the pipe wall. Fluid injection will influence velocity profiles directly; therefore, it will influence the value of f~. At small injection rates, the boundary layer will be lubricated and f~ will decrease to a value that is less than a friction factor for the no fluid injection case. The second term on the right hand side of Eq. 11 is caused by a change in the velocity profile in the axial direction. For small injection rates, this term is negligible since the injected fluid will not affect the veloci~ field significantly, except in the near wall region, If velocities at both cross sections are fully developed, this term is also negligible. Then, Eq. 11 becomes,
fT = fw + 2cnd~in 1 ~ """"" """"" """"" """"" """"" -"'"" """"" """"" """"" """:"(13)

metering section is composed of three turbine meters, which can be used to measure different ranges of flow rate, and a temperature transducer for measuring fluid temperature. The water flows through the test section and is then circulated back to the tank. This results in a constant Iiquid level in the tank without using a constant level regulator, permitting long test times.
Test Sections. Various completion methods are used for horizontal wells. In this study, the flow behavior in perforated horizontal wells and horizontal wells completed with slotted liners are investigated. Slotted Liner Test Sections. Various slot sizes and slot patterns are used. The minimum slot width is 0.05 in., which is specified to minimize slot plugging. In the construction of the test section, a slot size 2-in. long and I/16-in. wide is used. The single-slot-test section is a 10-R long, 1-in. diameter PVC pipe with one slot in the pipe wall. Liquid feeds the slot from a 3-in. diameter casing (Fig. 3), There are six pressure ports along the test section. The distance between the test section starting point and pressure port No. 1 is greater than 2 R, which allows the flow to fully develop. The pressure differences along the test section were measured by U-tube manometers. Locations of pressure ports and the injection point are shown in Fig. 3. Three multiple-slot-testsections were used, They are 10-ft long, approximately 1-in. diameter PVC pipes with 4-R long slotted sections. The total numbers of slots are 18, 18 and 36 with phasings of 360, 180, and 90. The test sections are called the 18a slot case, the 18 slot case and the 36 slot case, respectively (Fig. 4). The slot dimensions are the same as for the single-slot case. Each slotted section was covered with about 50 layers of cloth to ensure uniform influx from slots since the pressure loss through the cloth will be much greater than the pressure loss inside the 1-in. diameter pipe or in the annulus through which liquid feeds the slots. The multipleslot-test section and U-tube manometer connections are shown in Fig. 5. Perforation Test Sections. Three multiple-perforation-test sections were used. They are three 10-ft long, approximate 1in. diameter plastic pipes with 4-ft long perforated sections. The perforation densities are 5, 10 and 20 shots per foot, respectively and the perforation diameters are all the same, equal to l/8-in. This is geometrically similar to a 6-in. diameter casing perforated 1, 2, and 4 shots per foot with a perforation diameter equal to 3/4 in. The perforations are uniformly distributed along the pipes with phasings of 360, 180, and 90, respectively. Each perforated pipe is covered with about 50 layers of cloth to ensure uniform influx. The same pressure measurement devices and similar arrangement were used as for the multiple-slot cases. Range of hperiments. Experiments were conducted for steady-state flow with Reynolds numbers ranging from approximately 4,000 to 60,000 and influx to main flow rate ratios ranging from 1/100 to 1/2000 for multiple-perforation cases, 1/50 to 1/2000 for multiple-slot cases and 1/5 to 1/100 for the single-slot case. The influx to main flow rate ratios at

The last term in Eq. 11 is due to fluid acceleration and velocity profile change caused by the injected fluid. C is given by Eq. 12 and can be simplified for special cases as described below. For small influx to main flow rate ratios, or no severe velocity profile changes, C. =PI+~Z-@~P. If the injected fluid enters the main flow with no momentum in the axial direction, If i.e. V.=0, then $~P in Eq. 12 is equal to zero, and C.=131+p2. we further assume that ~l=~z= 1, then C becomes 2. If the injected fluid is considered to have a velocity of ~ once it enters the main flow stream, i.e. V.= fi, then the $ value in Eq. 12 becomes 1, and therefore Cn ~l+~z-BP. If we further assume that ~l=Bz=~P=1, then Cn=l. For the special case of no main flow at the pipe inlet, Eq. 11 can be simplified to:
fT = fw +(4d/~)(2P2
-@Pp) .............( 14)(l4)

For the case of one injection opening in the pipe wall, the apparent friction factor is given by:
fT=fw+2d/~(#2-#,)+

2d qio ~
QI + gin [

2p, @p,, PI *

.....................(15)

When the second term is not important, Eq. 15 can be simplified to:
fT = fw + c(zdi~)qin

/[QI+ qin].

.........OOOOO. 000..(16)

c = 2B1 OPp - PIgin /[QI

+ gin] .................................. (17)

The detailed derivation for a single injection opening can be found in Yuan29.
Experimental Program

A small-scale test facility was designed and constructed to simulate the flow behavior in horizontal wells (see Fig. 2). Water is used as the testing fluid and is stored in a 200-gal steel tank. A centrifugal pump and a screw pump are used to supply the main flow and side flow rates, respectively. Each

95

H. YUAN,C. SARICA,J.P. BRILL

SPE 48937

different Reynolds numbers were determined based on uniform influx and infinite conductivity cases.
CFD Simulation

fecting apparent friction factor and the applicability of the apparent friction factor correlations.
Data Analysis Procedure. The parameters measured are influx and main flow rates, influx and main flow temperatures, and pressure differences along the test section. For fully developed flow, regular pipe friction factor correlations can be used to calculate the friction factor. For the flow region affected by radial influx, the apparent friction factor f~ is calculated from the pressure drop, the distance, and the flow rate as given below.

A commercial Computational Fluid Dynamics (CFD) computer program Fluentm was used to simulate single-phase liquid flow in a horizontal well. The purpose of this simulation was to determine the length of the flow developing region in a horizontal well after the flow is disturbed due to radial influx. For most horizontal wells, the flow is in the turbulent region. Therefore, solutions to the Reynolds averaged Navier-Stokes equations are needed to simulate the flow. Description of the equations can be found in Yuan32. A two-dimensional, cylindrical pipe geometry is used for the simulation and the fluid is defined as water. The pipe diameter is 1 in. and the pipe length is 3.28 fi. The pipe geometry is setup in Fluentw directly and boundary fitted grids are used. Different turbulent models and near-wall treatments are evaluated and the simulation results for different turbulent models, near-wall treatments and grid numbers are compared by Yuan32. The K-& turbulence model with standard wall functions was used to find the flow-developing length when there is fluid injection from the pipe wall. The K&turbulence model with standard wall functions needs much fewer grid numbers to obtain reasonably good simulation results than the RNG K-E model with a two-layer zonal model. Convergent results can be reached easily by using the K-& turbulence model with standard wall functions. The simulation results and the experimental observations agree with each other. Nine Reynolds numbers, ranging from 5,000 to 60,000 at the pipe inlet, and 11 injection to main flow velocity ratios, ranging from 0.01 to 6.5 were used during the simulations, Based on simulation results, the following correlation was developed to predict the flow developing length at various Reynolds numbers and injection to main flow velocity ratios:
L~/d = a log(~n/V) + b. .............................................(l8)

~T = (@/~)

/(P~2 /2d) ...............................................(2l)

The Reynolds number is calculated by the following equation: N~e = piid I p , ...................,,............................................ (22) where Ax = Distance between one pipe diameter upstream of the slot and Lddownstream of the slot; Ap = Pressure drop over Ax; ii = Flow velocity calculated by ii= 4(Q1+ qti )/(ti2 ) for single slot, and ii = 4(Q1+ nqi. / 2)/(7af2 ) for multiple slots.
Single-Slot Case. A total of241 experimental tests were conducted for the single-slot case, Data were acquired for both no fluid injection and with fluid injection cases for Reynolds numbers ranging from 5,000 to 60,000. For the with fluid injection case, influx to main flow rate ratios vary from 1/5 to 1/100. In this section, experiments with no fluid injection and experiments with fluid injection are given. No Fluid Injection Case. Two flow configurations are investigated for the no fluid injection case, one with the slotted pipe placed inside the casing and influx rate equal to zero, and the other with the slot covered by tape. The friction factors for no fluid injection cases, fo, are calculated from the pressure drop over Axo,which is the distance between one pipe diameter upstream of the slot and one pipe diameter downstream of the slot. The CFD simulations show that the small opening causes an extra pressure drop; however, the effect is very small and the flow becomes fully developed again with respect to pressure within one pipe diameter downstream of the opening. The results are plotted as f. vs. NR, in Fig. 7. Friction factors calculated from the Blasius formula (f=O.3 16/NRe02s) and

where
a = 3.79 10-ON~e+5.21 10-6N~e+ 0.753. ....................(19)

b = &17 10-lON:e + 2.59 10-6N~, + 5.139. ....................(20)


Figure 6 shows that the developing length increases with increasing Reynolds number at certain injection to main flow velocity ratios and at high injection to main flow velocity ratios. The increase in flow-developing length with increasing injection to main flow velocity ratio is less than at low injection to main flow velocity ratio cases. Results and Discussion

Seven test sections were used during the experiments and a large amount of experimental data was acquired. In this section, the data analysis procedure and experimental data for each test section are given. Also given are the parameters af-

for the smooth section of the pipe are also plotted for comparison purposes. For both flow configurations, the friction factors are greater than smooth pipe friction factors and friction factors from the Blasius formula at the same Reynolds numbers, which means that the slot in the pipe wall introduced an extra pressure drop. For the case of a slot covered by tape, the higher friction factor is probably due to the higher roughness introduced by the slot. For the case of a slot inside a casing, the much higher friction factor might be due to the interaction

96

SPE 48937

Effectof CompletionGeometry and Phasing on Single-Phase LiquidFlowBehaviorin HorizontalWell

between the fluid in the pipe and fluid in the annulus. It is clear that placing the slotied pipe in a casing full of liquid, or covering the slot with tape makes a significant difference in the pressure behavior, and the former case is probably closer to the real case. Whether or not the casing size will affect the results remains unknown. Therefore, in studying horizontal well flow, neither of them can be taken as the reference. The smooth pipe friction factor calculated from the Blasius formula can be taken as a reference if a measured smooth pipe friction factor is not available. Fluid Injection Case. Figure 8 shows the variation in apparent friction factor with different influx to main flow rate ratios and Reynolds numbers. Apparent friction factor is plotted vs. Reynolds number, and different symbols represent experimental results at different influx to main flow rate ratios. The fT is greater than the smooth pipe friction factor and friction factor from the Blasius formula for all influx to main flow rate ratios. No lubrication effect was found, even at an influx to main flow rate ratio as low as 1/100. This is different from the single perforation case (see Yuan et a/.30). The reasons for this could be: 1) the area of a slot is almost 10 times the perforation area for the experiments, therefore introducing more roughness; and 2) the length of a slot is about 2 pipe diameters, resulting in the streamlines changing shape when they are passing through the slot, Also, the streamlines of the injected fluid interact with each other and with the main flow differently than for the single-perforation case. Experimental data show that the fT values decrease with decreasing influx to main flow rate ratio at a given Reynolds number. Also, for a given flow rate ratio, fT decreases with increasing Reynolds number. The experimental data were used to determine C in Eq. 17, which is a function of momentum correction factors, influx to main flow rate ratios, and the wall friction factor, This was achieved by plotting f~ given by Eq. 16 vs. NRC. Based on the theoretical analysis, C varies between 1 and 2 if the influx velocity is not too high. By trial and error, C was found to be 1.74, for which all the data collapse into one curve regardless of the influx to main flow rate ratios (Fig. 9). The following correlation was obtained by curve fitting the results in Fig. 9.
f~ =

o.221(iv~e)4203 + 3.48d/ h[qi~ /(Ql + qi.)] ........(23)

Figure 8 shows the comparisons between this expression and the experimental data. The lines designated in the legend by the letter M are plots of Eq. 23. Comparison of the lines with the data shows excellent results. Multiple-Slots Cases. A total of 626 experimental tests were conducted on three multiple-slot test sections. Data were acquired for no fluid injection, no main flow at the test section inlet, and with fluid injection for Reynolds numbers ranging from 5,000 to 50,000. For the fluid injection case, influx to main flow rate ratios vary from 1/50 to 1/2000. In this section, experiments with no fluid injection, experiments with no main flow at the inlet of test sections and experiments with fluid injection are given.

No Fluid Injection Case. Similar to the single-slot case, two flow configurations are investigated for the no fluid injection case. The results once again suggest that, in studying horizontal well flow, the smooth pipe friction factor or a friction factor from Blasius formula should be taken as a reference if the measured smooth pipe friction factor is not available. No Main Flow at Znlet of Test Sections. The apparent friction factors for the cases of no main flow at the test section inlet, fTi~,are calculated from the pressure drop over &in. which is the distance between one pipe diameter upstream of the slotted section and Ld downstream of the slotted section. Ld is calculated using Eq, 18 by assuming uniform influx along the slotted section, The experimental results are given in Fig. 10. The friction factors for cases of no main flow at the test section inlet are much higher than friction factors calculated by the equation for laminar flow, f=64~R,. Reasons for this could be the large velocity profile distortion caused by the fluid influxes and the higher roughness. Figure 10 shows that, at the same Reynolds number, as the slot number increases, the apparent friction factor decreases. Also, for the same slot number, apparent friction factors decrease as the phasing decreases. Fluid Znjection Case. Figures 11-13 show the variation in apparent friction factor with influx to main flow rate ratios and Reynolds numbers for the 18 slot case, the 18a slot case and the 36 slot case, respectively, Each figure is plotted as apparent friction factor vs. average Reynolds number, with different symbols representing experimental results at different influx to main flow rate ratios. fT is greater than the smooth pipe friction factor for all three slot distributions and for all influx to main flow rate ratios. No lubrication effect is observed for the same reason as discussed in the single-slot case. fT decreases with decreasing influx to main flow rate ratio, Also, for a certain flow rate ratio, fT decreases with increasing average Reynolds number. The experimental data were used to determine C. in Eq. 12, which is a function of momentum correction factors, influx to main flow velocity ratios, fw, the wall friction factor, and slot number and distribution. This was achieved by plotting fw given by Eq. 13 vs. NRC. By the method of regression, C. is found for each case at which all the data collapse into one curve, regardless of the influx to main flow rate ratios. Correlations for f~ are obtained for each c~e and can be written in one general form as shown below:
fw = a N;= ...................................................................... (24)

where a, b and Cn are given in Table 1. An apparent friction factor correlation was then obtained by substituting Eq. 24 and the appropriate ~ value into Eq. 13:
fT = a(N~e)b + 2Cnd~in /~

.........................................(25)

It is interesting to note that C. is greater than 2 for the 18a


and 36 slot cases. The reason for t~s could be that the last

term in Eq. 12, and the second term in Eq. 11 may not be negligible. Both of these terms include ~z-~l. The flow in cross

97

H. YUAN,C. SARICA,J.P. BRILL

SPE 48937

section- 1 is fi,dly developed; therefore, ~1 is a momentum correction factor for fully developed turbulent flow. The flow in cross section 2 is fully developed with respect to pressure only, and the velocity may not be fully developed. The nonuniformity of the velocity profile caused by the injected fluid results in a larger ~z value. Figures 11-13 also show the comparison between Eq. 25 and the experimental data for the 18 slot case, the 18a slot case and the 36 slot case, respectively.
Multiple-Perforation Cases. A total of 390 experimental tests were conducted on three multiple-perforation test sections. Data were acquired for no main flow at the test section inlet and with fluid injection case for Reynolds numbers ranging from 4,000 to 55,000. For the fluid injection case, influx to main flow rate ratio varies from 1/100 to 1/2000. In this section, experiments with no main flow at the inlet of test sections and experiments with fluid injection are given. No Main Flow at Inlet of Test Sections. Apparent friction factors for the case of no main flow at the test section inlet, fTin,are calculated from the pressure drop over Axin,which is the distance bemeen one pipe diameter upstream of the perforated section and Ld downstream of the perforated section, Ld is calculated using Eq. 18 by assuming uniform influx along the perforated section. The experimental results are given in Fig. 14. Friction factors for cases of no main flow at the test section inlet are much higher than friction factors calculated by the equation for Iaminar flow, &64/N~., This is similar to the multiple-slot cases. Figure 14 shows that, at the same Reynolds number, apparent friction factor for the 5 shots/ft case is the highest while the apparent friction factor for the 20 shots/fi case is the lowest. This is because the influx velocity from each perforation is small for the high perforation density case. FZuid Injection Cme. Figures 15-17 show the variation in apparent filction factor with influx to main flow rate ratios and Reynolds numbers for the 5 shots/fi case, the 10 shots/fi case and the 20 shots/fi case, respectively. Each figure is plotted as apparent friction factor vs. average Reynolds number, and different symbols represent experimental results at different influx to main flow rate ratios. fT is greater than the smooth pipe friction factor for the 5 shots/ft case and the 10 shots/fi case for all the influx to main flow rate r;tios. For the 20 shots/ft case, fT is smaller than the smooth pipe friction factor at influx to average main flow rate ratio equal to 1/2000, which means that fluid injection actually reduced the friction factor and the influx has a lubrication effect on the flow. This may have been caused by the slow moving boundary layer formed by the injected fluid. The experimental data were used to determine Cn in Eq, 12, by using the same procedure as for the multiple-slot cases, Correlations for fWare obtained for each case and can be expressed in one general equation as given in Eq. 24 with a, b and C. given in Table 2. Cn values for multiple-perforation cases are all greater than 2. This may be due to the same rea-

son as discussed for multiple-slot cases. Apparent friction factors for multiple-perforation cases can be calculated with Eq. 25 by using the coefficients given in Table 2. Figures 15-17 show comparisons between this expression and the experimental data. Experimental results for the 5 shots/ft case and the 18a slot case are compared in Fig. 18. The phasing for both cases is 360. Figure 18 shows that apparent friction factor for slotted pipe is lower than for perforated pipe, probably due to the smaller influx velocity since the area of a slot is almost 10 times that of a perforation area in the experiments. Slotted pipe gives a flatter curve for each influx to average main flow rate ratio, similar to the rough pipe behavior in a Moody chart. This is due to the higher roughness of the slotted pipe.
Parameters Affecting Apparent Friction Factor. The apparent friction factor expression is presented in terms of dimensionless parameters, which are functions of some independent variables. Each is discussed below. qiJQ, Influx to Main Flow Rate Ratio. This dimensionless variable is composed of two dimensionless variables, ViniV and A~A. For the same qi./Q, a decrease in Vi.N indicates an increase in A~A. The influx to main flow rate ratio contributes significantly to apparent friction factor. Each term in Eq. 11 is either explicitly or implicitly a function of the influx to main flow rate ratio. For the no fluid injection case, Eq. 11 has only two terms on the right hand side, which represent the wall friction and the distortion of the velocity profile. If the flow is fully developed, the velocity distortion term disappears, and the equation is identical to the regular friction factor correlation. fWis a function of influx to main flow rate ratio as well as Reynolds number and pipe roughness for the case of fluid injection. Therefore, a regular friction factor correlation cannot be used for the fluid injection case. For large influx to main flow rate ratios, the last two terms in Eq. 11 become large due mainly to changes in the velocity field. A+A can also affect fTthrough changing of pipe roughness. Vw, Ratio of Axial Velocity Component of Injected Flow and Main Flow Veloci@. This is an indication of the

axial momentum carried to the main flow by the injected fluid. This ratio will affect the C, value in Eq. 12 directly. N~@Reynolds /Vtifiber. This is the ratio of inertia forces to viscous forces. NR~will mainly affect the f~ term in Eq. 11. n, Number of Injection Openings. This will affect the last term in Eq. 11 directly, and will also affect the velocity profiles, that is the ~ value, therefore affecting other terms in Eq, 11 indirectly. 9, Phasing of Injection Openings. This will influence the velocity distribution, that is the ~ value, and therefore the fT.
Practical Application of Apparent Friction Factor Correlations. The new correlations can be used with confidence for

operating conditions dynamically similar to those of the experiments. The performance of the correlations needs to be tested against field data to validate the correlations and to determine if they can be extrapolated beyond the range of the
98

SPE 48937

Effectof CompletionGeomeby and Phasing on Single-Phase LiquidFlowBehaviorin HorizontalWell

data presented in this study. When the perforation or slot density is low, i.e., the distance between peflorations or slots is large enough to achieve fully developed flow; the results from this study suggest that the correlations for a single injection opening can be used. Regular pipe friction factor correlations can be used for the filly developed flow region. When the perforation or slot density is higher, i.e., the distance between perforations or slots is not large enough to establish fully developed flow, the main flow cannot recover from previous influx disturbances before reaching the next injection opening. In this situation, correlations for the multiple-slot or multiple-perforation cases can be used, In a complete reservoir-well system, there is a global dependence between pressure behavior and fluid flow in both the reservoir and the weIlbore. The current practice in the petroleum industry is to couple independent reservoir, near wellbore, and wellbore models, considering the continuity of pressure and flux at the wellbore/reservoir interface. Many simulators use different coupling schemes. The proposed apparent friction factor correlations can be used in any model which considers pressure drop in the wellbore, such as the models proposed by Landman21, Ozkan et a/25 ., Brekke et al. 14,Dikken*6, etc. Here, as an example, the apparent friction factor correlation is incorporated into Dikkens horizontal wellbore model. Details of the derivation are given by Yuan32. Figures 19-20 give the relationship between horizontal well pressure drop and total production rate for a typical field data set given in Table 3. The distance between two adjacent slots is 3.94-ft for the single-slot case. The distance of each perforation/slot interval is 16.4-fi including a 10-ft perforated/slotted section and a 6.4-ft smooth section for multiple slotiperforation cases. Using the same data set, Ozkan et al.25 estimated a 3 psi pressure drawdown for a flow rate of 2,700 bbl/D considering no pressure loss in the wellbore. Pressure drops through the well obtained by using different apparent friction factor correlations are given in Table 4. The pressure drop in the wellbore is comparable to the pressure drawdown in the reservoir, Therefore, the pressure drop along the horizontal well can not be ignored. Since the difference between the results from the proposed correlations and regular pipe correlations is large, the importance of using a proper friction factor correlation is clear.
Concluding Remarks

multiple-perforation cases. The Reynolds numbers for the experiments range from approximately 4,000 to 60,000. New correlations for apparent friction factor have been developed for different completion geometries. For horizontal wells completed with slotted liners, for the same slot number, the apparent friction factor is smaller when the slot phasing is lower. For perforated horizontal wells with high perforation density case, the apparent friction factor is higher for high local influx to main flow rate ratio and is smaller for low local influx to main flow rate ratio. A small amount of fluid injection for the 20 shots/ft case decreases the apparent friction factor, which means that the injected fluid has a lubrication effect. No lubrication effect was found for both single and multiple-slot cases. For perforated and slotted pipes with the same phasing, the apparent friction factor for slotted pipe is lower than for perforated pipe at the same influx to main flow rate ratio. Slotted pipe gives a flatter fr vs. NR, curve for each influx to main flow rate ratio, similar to the rough pipe behavior in a Moody chart.
Nomenclature

Ihe tlow behavior m horizontal wells ror two completion geometries, slotted pipe and perforated pipe, was investigated. For each completion geometry, various perforation or slot densities and distribution patterns were investigated. A large amount of experimental data was acquired for the no fluid injection case, the case with no main flow at the test section inlet, and the case with fluid injection. For the case with fluid injection, the influx to main flow rate ratios range from 1/5 to 1/100 for the single-slot case, from 1/50 to 1/2000 for multiple-slot cases, and from 1/100 to 1/2000 for

A = pipe cross sectional area (m*) C = coefficient as defined by Eq. 17 C. = coefficient as defined by Eq. 12 d = pipe diameter (m) f. = Moody friction factor for no fluid injection cases fT = apparentMoody friction factor ffin = apparent friction factor for no main flow L = pipe length (m) Ld = flow-developing length (m) n = number of openings along Ax NR, = Reynolds number P = pressure (pa) pi = initial pressure of reservoir (pa) pW= well pressure (pa) Q = main flow rate (m3/s) qin = volumetric flow rate from each opening (m3/s) y~ = total production rate (m3/s) U = average main flow velocity (m/s) VP = average influx velocity from each opening (m/s) V., V, = velocity components as shown in Fig. 1 AX = distance in Fig. I (m) or as defined in the content x = distance along a horizontal well (m) ~ = momentum correction factor (-) & = absolute pipe roughness (m) $ = defined by Eq. 10 ~ = the density of openings defined by Eq. 11 (l/m) y = dynamic viscosity (N s/m2) p = density (kg/ m2) rW = wall shear stress (N/m2)

99

8 Subscripts O = variables pertinent to no influx case

H. YUAN,C. SARICA,J.P. BRILL

SPE 48937

1 2 in p r w x

= = = = = = =

variabies pertinent to cross section I- I variabies pertinent to cross section 2-2 variables pertinent to influx stream variabies pertinent to injection opening variabies pertinent to radial direction variables pertinent to wall or well variables pertinent to axial direction

Acimowiedgements We thank the Tulsa University Fluid Flow Projects (TUFFP)

member companies for funding this project. Fluent Inc. is acknowledged for providing an academic license of their CFD sotiare.
References 1. Denn, M.M.: Process Fluid Mechanics, Prentice-Hall,Inc.,

1980. 2. Dries~ E.R.V.: On Turbulent Flow Near a Wall, Journal of the Aeronautical Sciences, Vol. 23, 1956, pp. 1007-1011. 3. Hombeck,R.W., Rouleau,W.T.,and Osterle, F.: Laminar Entry
Problem in Porous Tubes, The Physics of Flui&, vol. 6, number 11,NOV.1963, pp.1649-1654. 4. Jerornin, L.O.F.: The Status of Research in Turbulent Boundary Layers Whh Fluid Injection, Progress in Aeronautical Sciences, Vol. 10, pp. 65-189, 1970. 5. Kate, H., Fujii, Y., Yamaguchi, H. and Miyanaga, M.: Frictional Drag Reduction by Injecting High-Viscosity Fluid into Turbulent Boundary Layer, Journal of Fluids Engineering, June 1993, vol. 115, pp. 207-212. 6. Kinney, R.B.: FuIly Developed Frictional and Heat-transfer Characteristics of Laminar Flow in Porous Tubes: Int. J. Hear Mass Transfer, vol. 11, 1968, pp.1393-1401. 7. Moffat, R.J. and Kays, W.M.: A Review of TurbulentBoundary-Layer Heat Transfer Research at Stanford, 1958-1983, Advances in Heat Transjer, vol. 16, 1984, pp. 241-365. 8. Olson, R.M. and Eckert, E.R.G.: Experimental Studies of Turbulent Flow in a Porous Circular Tube with Uniform Fluid Injection Through the Tube Wall, Journal of Applied Mechanics, March, 1966, pp. 7-17. 9. Yuan, S.W. and Finkelstein, A.B.: Laminar Pipe Flow With Injection and Suction Through a Porous Wall, Trans. ASME, vol. 78, 1956, pp. 719-724. 10. White, F.M.: Viscous Fluid Flow; McGraw-Hill, Inc., Second Edition. 11. Schlichting, H.: Boundary Luyer Theory; McGraw-Hill, Inc., Seventh Edition. 12. Asheim, H., Kolnes, J., and Oudeman, P.: A Flow Resistance Comlation for Completed Wellbore, Journal of Petroleum Science and Engineering, 8 (1992), pp. 97-104. 13. Brekke, K.: New and Simple Completion Methods for Horizontal Wells Improve the Production Performance in High Permeability, Thin Oil Zones, paper SPE 24762, presented at SPE 67th Annual Technical Conference and Exhibition, Washington, D.C., oct. 4-7, 1992. 14. Brekke, K., Johansen, T.E., and Olufsen, R.: A New Modular Approach to Comprehensive Simulation of Horizontal Wells: paper SPE 26518, presented at SPE 68th Annual Technical Conference and Exhibition, Houston, Texas, Oct. 3-6, 1993. 15. Brice, B. W.: Production Impacts on AP Friction in Horizontal

Production Wells, paper SPE 23666, presented at SPE Second Latin American Petroleum Engineers Conference, 11LAPEC, Caracas, Venezuela, March 8-11, 1992. 16. Dikken, B.J.: Pressure Drop in Horizontal Wells and its Effect on Their Production Performance, paper SPE 19824, presented at SPE 64th Annual Technical Conference and Exhibition, San Antonio, TX, Oct. 8-11, 1989. 17. Ihara, M.: Two-Phase Flow in Horizontal Wellbores, MS. Thesis, The University of Tulsa, 1991. 18.Joshi, S. D.: Horizontal Well Technology; Pennwell Publishing co., 1991. 19. Kloster, J.: Experimental Research on Flow Resistance in Perforated Pipe, M.S. Thesis, Norwegian Institute of Technology, 1990. 20. Lacy, S., Ding, W., and Joshi, S.D.: Horizontal Well Applications and Parameters for Economic Success, paper SPE 23676, presented at SPE Second Latin America Petroleum Engineers Conference, 11LAPEC, Caracas, Venezuela, March 8-11, 1992. 21. Landman, M. J.: Analytic Modeling of Selectively Perforated Horizontal Wells, Journal of Petroieum Science and Engineering, 10 (1994), pp.179-188. 22. Norris, S. O., Hunt, J. L., and Soliman, M. Y.: Predicting Horizontal Well Performance: A Review of Current Technology, paper SPE 21793, presented at the Western Regional Meeting held in Long Beach, California, March 20-22, 1991. 23. Novy, R. A,: Pressure Drops in Horizontal Wells: When Can They Be Ignored?, paper SPE 24941, presented at SPE 67th Annual Technical Conference and Exhibition, Washington, D.C., Oct. 4-7, 1992. 24. Ouyang, L., Arbabi, S. and Aziz, K.: General Wellbore Flow Model for Horizontal, Verticai, and Slanted Well Completions: SPE 36608, presented at 1996 SPE Annual Technical Conference and Exhibition, Denver, Colorado, Oct. 6-9, 1996. 25. Ozkan, E, Sarica, C., Haciislamoglu, M. and Raghavan, R.: The Influence of Pressure Drop Along the Wellbore on Horizontal Well Productivity, paper SPE 25502, presented at SPE 1993 Production Operations Symposium, Oklahoma City, OK, March 21-23, 1993. 26. Seines, K., Aavatsmark, I., Lien, S.C., and Rushworth, P.: Important Reservoir Considerations for Horizontal Wells: A Field Example, paper SPE 21124, presented at SPE Latin American Petroleum Engineering Conference, Rio de Janerio, Brasil, Oct. 14-191990. 27. Su, Z. and Gudmundsson, J.S.: Friction Factor of Perforation Roughness in Pipes, paper SPE 26521, presented at SPE 68th Annual Technical Conference and Exhibition, Houston, Texas, Oct. 3-6, 1993. 28. Su, Z. and Gudmundsson, J.S.: Pressure Drop in Perforated Pipes: Experiments and Analysis; paper SPE 28800, presented at the SPE Asia Pacific Oil & Gas Conference , Melbourne, Australia, Nov. 7-10, 1994. 29. Yuan% H.: Investigation of Single Phase Liquid Flow Behavior in a Skgle Perforation Horizontal Well, M.S. Thesis, The University of Tulsa, 1994. 30. Yuan, H., Saric~ C. and Brill, J.P.: Effect of Perforation Density on Single Phase Liquid Flow Behavior in Horizontal Wells, SPE 37109, pre>ented at the 1996 International Conference on Horizontrd Well Technology, Calgary, Albe@ Canad~ Nov. 18-20. 31. Yuan, H., Saric% C., Misk% S. and Brill, J. P.: An Experimental and Analytical Study of Single Phase Liquid Flow in a Horizontal Well~ASME Journal of Ener~ Resources Technolo#, Vol. 119, March 1997, pp. 20-25. 32. Yuan, H.: Investigation of Single Phase Liquid Flow Behavior in

100

SPE 48937

Effect of CompletionGeometty and Phasing on Single-Phase LiquidFlowBehavior in Horizontal Well

Horizontal Wells, Ph.D. Dissertation, The University of Tulsa, 1997. Table 1- Coefficients for Multi Ie-Slot Cases

Pump I

Table 2- Coefficients for Multiple-Perforation Cases 10 Shots/ft 20 Shotslft 5 Shots/ft 1,297 0,363 a 0.641 -0.421 -0.266 b -0.312 2.2 2.2 c 2.2 Table 3- Reservoir and Fluid Data
Fig, 2- Schematic Description of Test Facility.
Mc@ring S.xlion Pump

I Oil viscositv. u Oil density, p Formation volume factor, B Horizontal well length, L Wellbore radius, rw

I 1.43 Cp 55 Ib/fiJ 1.16 bbl/STB 2625 ft 0.1042 ft

I
I

2.S4 in. 0=1 in. 2

. 1

24in.

3.875in

Table 4- Pressure Drop Through a Horizontal Well Friction Factor Pi-Pwf

,,,,

i
In*lion

I
I mint

Fig. 3I 20 Shots/fi Case 34.48

Single-Slot

Test Section

2 la,

r r&a

In _
*A

+@
M

Fig. 1-

Schematic of HorizontalWell
Uniformly Distributed

Control Volume for Injection Openings.

Fig. 4-

Slot Distribution

Along the Pipes

101

10

H. YUAN, C. SARICA, J.P. BRILL

SPE 48937

0.12

0.10

0.08

0.04

7
0.02 0.52

~...
M<mlation Prediction D-Expcrimenti Dam

1-

201X0

mm

Re
Fig. 5- Multiple-Slot Test Section and U-tuba Manometer Connections Fig. 8- Comparison between Correlation and Experimental Data (Single-Slot
U.U50

Predictions Case)

--q
. , -

.-.

---- >- ----- > -=-. -.--*------* -------. am..-.-.-*-.--

-------

0.W5

0.040
0.035

.+--a
A......b ~+

...* .............

A .............& ......."" ....h .............


_-_+-=-=
21

0.030

.
I

: ~ __ ~ ~=,o,w
-

.*

=-==s-2
*.$,W
*-20.000 M -30.OW Nn-40,DDa h =So,lmll m-60#o

cl

0.025

S-Simulation RCSIJIU
C.COmlatiOn Wolm

0~ As ills q q 9J 21

.. . . . . . . -... ------cl

C2 C4 a C6

0.020

0.015

0.010 2 VinN14 6

1
0

fw=0.221 NRe-0.203

r~0.91

C=l.74
-1

lm

2&

mm

Re
Fig. 6- Flow-Developing Length vs. Velocity Ratio Simulation Results and Correlation Predictions Fia. 9- Wall Friction Factor vs. Reynolds Number

(Single Slot Case) -

0.25

0.20

0.15

:Z 0.10

Frklion F=wr Cakulakd by d4 I NRe

, 0
b

Tllc M

slot *

Tbe 18 Slot Cax m I& Slol Cax

0.05

102W

2&

4odol

t$

O.w

MJ

Re
Fig. 7- Friction Factor for No Fluid Injection Cases (SingIe-Slot Case) Fig. 10- Experimental Data for No Main Flow at Test Section Inlet (Single Multiple Slot Case)

102

SPE 48937

Effect of Completion Geometry and Phasing on Single-Phase Liquid Flow Behavior in Horizontal Well

11

0.6
0 :% $0 -Q.m

D-. D~,ml~ D*-lrso2 Dw-1/lm

l/lM

------------

Mq+:-lllW MqW@, Mq+-1~ Mq~.1/lm 0,4 . 1/2M 0.5

5 ShorsPr fDDI 10 ShoLs pr foot 20Skls psr [001 by 64/~

n
u

q m

o 000 1

Ftition Fwlor Glc.lalsd

~+

0.3

-0.

--~

0.2

M - Correlation Wction D - Expcrimcnti Dam

0.1

0.0

Im

2m

302

1020

Iw

mzmo

3000

3300

al

Re
Fig. 11- Comparison between Correlation Predictions and Experimental Date (18 Slot Case)
..o
0 rs u u .. u --g.g q 0 x -.n.oa D~,.lN DWS=IIIM Dq-lti Dx-lm Dm.llIW D~s-1~ DSMCCUIX -K =1~ .. ... ... -... ------... --M@/QI= Me/Q., M~m, Mq8/Q., l/SO .l/lM 0,07 .1r201 .lm O.M Me@#.lrlWO Me@s=lm 0.05.

Re
Fig. 14- Experimental Data for No Main Flow at Test Section Inlet (Multiple Perforation Case)

0,08

........
---------

MqjQI . I/lMl

M+Gs.lm
M%~-1~

b4q.~.1/la10 Mq. &,-lm

----

=l?vmwnmo c1
60..

M - Comlation Prediction D - Expcrimcrdsl DaIS

0.04

0.03

M3
0,02

&D2 1
0.014 0
lm

0.0 I

NRe

Xxl

Io

Re
Fig. 15- Comparison between Correlation Predictions and Experimental Data (5 shots/ft Case)

Fig. 12- Comparison between Correlation Predictions and Experimental Data (18a Slot Case)

0.12s D 0 0.100 D~=li20 D~-1/lw . .. ... .. ---M~-1~ Me-1/102

. ..-

0.075

MI

-b
am

O.m

o,o,~ Moa

o.

!m

.m

Re
Fig. 13- Comparison between Correlation Predictions and Experimental Data (36 Slot Case)

Re
Fig. 16- Comparison betieen Correlation and Experimental Date (1 O shote/ft Predictions

Case)

103

12

H. YU~,,C.

SARICA, J.P. BRILL

SPE 48937

0.14
u

D_ D-1=1~ D-: D @v:= D~vs-1DS&PIW

1/103

.. .. .. .. ~*F ---M@~vFlm ~~r M ~ovp


Mq~vpI~

,,,~

0.12 *.9

0
q m

a lo

w --l/lCOO ------

m
I/Ire

0.0s .+ 0.06

OG~

0.04

aoo
Re Fig. 17- Comparison between Correlation Predictions and Experimental Data (20 shotsrft Case)
OKr/,

s)

1s00

m
@bbl/D)

3 K1

Fig. 20- Total Production rate VS. Pressure Drop in a Horizontal Well (Perforated Pipes)

o.06-

o.os ~*
o.04-

0.03-

Re Fig. 18Comparison of Experimental Data (18a Slot Case vs. 5 shots/ft Case)

40

30

10
e

. . . A*
.:.:.:.~

$
0

..- .0 . ..e --. . ... ..o-....


---e

#.e

/.-*A* a

.e

.,-

,s

,, /

,.
..- --- #

,.~

..-

-..0--

---n--- -F.-a .....--. ....*


..

. ... . ...*...--.

..*-..-.
4

Fig. 19- Total Production rate vs. Pressure Drop in a Horizontal Well (Perforated Pipes)

104

You might also like