See discussions, stats, and author profiles for this publication at: [Link]

net/publication/245356476

Pipe Flow With Radial Inflow: Experimental and Modeling Work

Article in Journal of Fluids Engineering · March 1999

DOI: 10.1115/1.2821989

CITATIONS READS
11 174

3 authors, including:

Trygve Rinde
Ranold AS
6 PUBLICATIONS 32 CITATIONS

SEE PROFILE

All content following this page was uploaded by Trygve Rinde on 30 September 2020.

The user has requested enhancement of the downloaded file.

 Pipe Flow With Radial Inflow:
R. M. S. M. Schulkes
Principal Scientist. Experimental and
Trygve Rinde Modeling Work
Process Engineer.
In this paper results are presented from experiments in which the pressure loss in
single-phase pipe flow is studied when radial inflow occurs. Experiments have been
Ole Harald Utvik carried out with pipes which have dijferent perforation geometries so as to be able
Process Engineer. to investigate the effect of perforation geometry on the pressure loss. Data analysis
of these experiments, as well as analysis of experiments carried out by other groups,
Norsl( Hydro ASA, yields a pressure loss model which accurately describes pressure losses in single-
Researcti Centre Porsgrunn, phase pipe flow with radial inflow through perforations in the pipe wall. The experi-
N-3091 Porsgrunn, Norway mental data is subsequently used to establish a numerical value of a parameter which
is used in a model description. This leads to the formulation of an effective friction
factor for pipe flow with radial inflow.

1 Introduction While the pressure drop in the case of fully-developed pipe

flow is well-known, the effect of radially in- or outflowing fluid
The pressure loss in perforated pipes where fluid flows into
on the pressure losses is less well understood. The reason for
the pipe through the perforations have a number of important
this is that with radial inflow, the axial pipe flow is under
engineering applications. One of these applications arises in the
continuous development: fully developed turbulent flow does
oil industry where oil may be extracted from a reservoir by
means of a perforated pipe which is inserted into the reservoir. not occur anywhere in the pipe when radial inflow occurs. The
In recent years, it has become common to drill long ( > 1 km), development of the velocity profile and the pressure variation
horizontal wells so as to optimize production from long, flat in a porous tube with uniform injection of air through the tube
(thickness 10-15 m) oil reservoirs. Such reservoirs could not wall have been studied by Olsen and Eckert (1966). The mea-
be exploited economically until recently. Problems with wells, surements of the friction factor showed that the friction factor
which have such a long production part, are partly related to decreased linearly as a function of the ratio of radial velocity
the fact that the pressure drop along the length of the well may to mean axial velocity. Experiments in the same vein (Aggarwal
be of the same order of magnitude as the under-pressure re- et al., 1972) showed that radial outflow leads to an increase in
quired to get oil out of the reservoir. This is in particular the the friction factor. More recently, Ouyang et al. (1996) analyzed
case when the permeability of the reservoir formation is large. data from large-scale experiments and found, in accordance
In such situations, the pressure difference between the well with Olsen and Eckert (1966), that the friction factor decreased
and the reservoir increases significantly as one moves from the for increasing radial flow velocities. Ouyang et al. (1996) found
closed-end to the production-end of the well. A consequence that the decrease in the friction factor was only dependent on
of this may be that the production profile can be severely skewed the Reynolds number of the flow through the perforations.
with significantly more oil being extracted from the reservoir Somewhat surprisingly, the decrease was not found to depend
at the production-end than at the closed-end of the well. This, on the axial flow conditions. Recent experiments (Yuan et al.,
in turn, can lead to the early breakthrough of water or gas with 1996) yielded a number of correlations for pipe flow with radial
the associated drop in revenues. It is clearly imperative to be inflow. These experiments proceed along the lines of the experi-
able to predict the pressure drop in a production well accurately ments presented in this paper but the data analysis differs on
so as to optimize production from a given reservoir. fundamental points. As will be shown in this paper, by properly
The pressure loss in single-phase pipe flow has been studied scaling the results, the data from different experiments show
extensively. The detailed experiments of Schlichting (1936), clear patterns which allows very simple and transparent correla-
among many others, lead to the well-known Moody-chart tions to be obtained. The experiments of Su and Gudmundsson
(Moody, 1947). More recent experiments (Perry et al, 1969) (1995) (see also Su, 1996) are similar to the experiments pre-
have studied effects of the wall roughness on the pressure gradi- sented in this paper. However, once again the data analysis
ent. These and numerous other experiments have yielded pres- obscures some of the patterns which can be obtained from the
sure drop correlations where the frictional pressure drop is ex- experimental data. The experiments presented in the present
pressed in terms of the Darcy-Weisbach equation paper are an extension of the results reported by Schulkes and
Utvik (1998). Differences between the present experiments and
those of Schulkes and Utvik (1998) pertain to the experimental
Ap = f pu , (1.1) setup and perforation geometry of the pipe.
In this paper, results are presented from experiments of pipe
in which / is the friction factor, L the pipe length, D the pipe flow with radial inflow. The aim of the experiments was to
diameter, p the fluid density, and u the mean fluid velocity. A determine the pressure loss in pipe flow with radial inflow and
large number of implicit and explicit expressions exist for the establish accurate and simple models which can be used to
friction factor /(Haaland, 1983). predict the pressure loss in such systems. The paper is organized
as follows. We start with theoretical considerations concerning
the effect on the pressure profile of radial inflow. A one-dimen-
Contributed by the Fluids Engineering Division for publication in the JOURNAL
OF FLUIDS ENGINEERING . Manuscript received by the Fluids Engineering Division
sional model is derived which takes into account radial inflow
January 20, 1998; revised manuscript received October 5, 1998. Associate Techni- effects. This is followed by a discussion of the experimental
cal Editor: D. R. Williams. setup. A detailed data analysis is then presented in which it is

106 / Vol. 121, MARCH 1999 Copyright © 1999 by ASME Transactions of the ASME
 Sx • Pressure cap Perforated pipe section
1 I / I I I H I I

Y "(X)

L1/2
I1M
U J i t | dpi ^8
dp1M
dp1/2
~/K /K A\ /IS /IS dp1/1
L1/1
V
(a)
Fig. 1 Schematic diagram of pipe flow with radial inflow

shown that results obtained under widely different experimental dp1/4

conditions have a similar structure. It is also shown that the L1/2
dp1/2
experimental results agree well with the anticipated theoretical dp1/1
L1/1
behaviour. Finally, an expression for an effective friction factor
is presented which accounts for radial inflow effects in pipe (b)
flow,
Fig. 2 Schematic diagram of the experimental setup In the cases of
experiments with Pipe 1 (a) and Pipe 2 (b)
2 Theoretical Considerations
Consider a pipe with an internal diameter equal to D. Fluid
with a density p and a viscosity /x flows along the axis of the
- (p«) + — ipu^) - - ^ - P«>J = - / - 77: pu\ (2.5)
pipe. It is assumed that the pipe wall is porous and that through at ax R ax 2D
the pores fluid with the same physical properties enters the pipe.
Details of the flow field near the porous walls are exceedingly in which D denotes the diameter of the pipe. As stated above,
complicated and our aim is not to model these local features. the parameter \ is difficult to quantify theoretically. The experi-
Instead, we take a one-dimensional view of the problem, aiming ments described below are essentially aimed at determining \ .
to parameterise the complicating three-dimensional flow struc-
tures close to the porous wall. Hence, let u(x) denote the mean 3
Experimental Setup
velocity in a pipe segment of length Sx as shown in Fig. 1. The experiments were performed with two different types
The rate of change of momentum in the segment (that is, the of commercially available pipes which are used in actual oil
momentum flowing out of the segment minus the momentum production. Pipe 1 (a HALLIBURTON liner) is a perforated
flowing into the segment) is equal to steel pipe (64 perforations per meter, diameter of each perfora-
tion is 9.5 mm) with an inner diameter of 0.16 m and a perfora-
— (Sm-u) = — (PU)SXTTR^ + TT (PU^)SXTTR^ + 6M„ (2.1) ted length of 6 m. Pipe 2 (a wire-wrapped CONSLOT screen)
dt at ox is a pipe geometry which is constructed by wrapping a triangular
in which 6M„ denotes the change in j;-momentum due to radi- wire around supporting rods which lie in a circular formation
ally in-flowing fluid. Let $ denote the fraction of the pipe wall along the axis of the pipe. The base of the triangular wire is
which is perforated. The total area of the perforations in the directed outwards and is typically in the order of 3-4 mm wide.
segment 6x is then equal to $ • & • 2irR. If the velocity of the The spacing between the wire-wrappings is 150 pm.. The inner
fluid which flows into the pipe through the pores is equal to v, diameter of this pipe is 0.15 m while the length of the tested
then the mass flow rate through the pores is equal to pipe was 8.15 m.
pv • ITTR^SX. This momentum is carried in the x-direction with The experimental setup is described elsewhere in some detail
a velocity which is a fraction of the mean axial velocity u, \u (Schulkes and Utvik 1998; Schulkes et al., 1997). The position
say, where \ <^ 1. An actual estimate of the parameter X. is of the differential pressure transducers differed somewhat be-
difficult to give since it is likely to depend on the local flow tween Pipe 1 and Pipe 2. Figure 2 shows the arrangements of
structure in/near the wall boundary layer as well as on the ratio the pressure transducers and in Table 1 the distances between
of the radial and axial velocities. The above considerations lead the pressure transducers are listed.
to the following expression for the change in j;-momentum due In all the experiments, the pressure cells were calibrated to
to radial inflow of fluid measure in different ranges so as to give the most accurate
result. The accuracy of the pressure cells is 0.2 percent of full
SM„ = -InRki^puvSx. (2.2) range. Water flow rates were controlled using manual regulating
valves and the accuracy with which the flow rates could be
The reason for the negative sign on the right-hand side of controlled was 0.5 percent of the volume flow rate.
(2.2) is the fact that there is only radial inflow and no radial
outflow of momentum. The change in x-momentum in the seg-
ment Sx is equal to the forces exerted on the segment Sx by the 4 Experimental Results
fluid, thus The experimental procedure consists of two parts: a first part
without radial inflow to establish the friction factor of the pipe
— (6m -u) = - -— TTR^SX 2-KRT„SX, (2.3) and a second part where radial inflow effects are studied. We
dt OX begin with a discussion of the experiments in which there was
in which p{x) denotes the pressure and T„ denotes the wall
shear stress. The wall shear stress can be expressed in terms of
the mean velocity in the pipe via Table 1 Distances between pressure transducers for the
two pipes used in the experiments
sP/"^ (2.4)
Pipe 1 Pipe 2
in which /denotes the friction factor which is a function of the DPl/8: L„8 = 0.55 m DPl/4: I„2 = 8,33 m
Reynolds number and the wall roughness. Combining ( 2 . 1 ) - DPl/4: L„4 = 1.15 m DPl/2: Z.1,2 = 8,33 m
(2.4) it follows that a force balance in the segment Sx yields DPl/2: L„2 = 7.52 m DPl/1: /.„, = 9.53 m
DPl/1: L„i = 9.60 m
the equation

Journal of Fluids Engineering MARCH 1999, Vol. 121 / 107

Table 2 Experimental axial and radial volume flow rates Fig. 3). However, when radial inflow occurs, the flow field in
for Pipe 1 and Pipe 2 the pipe is under continuous development and can not be as-
sumed to have a fully developed turbulent structure anywhere
Pipe 1 Pipe 2
in the pipe. Therefore, (4.2) must be regarded as a first approxi-
Axial volume Radial volume Axial volume Radial volume mation to the frictional pressure drop when radial inflow occurs.
flow Q flow q flow Q flow ^ As in some previous experiments (Su and Gudmundsson,
(m'hr"') (m^hr~'/m) (m'hr-') (m-'hr-'/m) 1995; Schulkes and Utvik, 1998), the total pressure drop in a
pipe with radial inflow is decomposed into three terms: a friction
50 0^3.9 25 0^2.8
75 0->3.9 50 0-*2.8
loss term (Apj) based on the assumption of fully developed
100 0->3.9 75 0-»3.4 turbulent flow (Eq. (4.2)), a pressure loss term associated with
150 0->3.9 100 0-»2.5 momentum increases (Ap„) and a term which is essentially a
175 0->3.9 125 0->3.4 correction to the friction term ([Link]). Thus,
200 0^3.9 150 0-»2.7
Ap = Apf + Ap„ + Ap, (4.3)
Note that the correction term Apc is intimately related to the
no radial inflow. The experimental set-up allowed us to deter- third term on the left hand side in Eq. (2.5). The exact relation
mine the pressure drop in the smooth pipe sections upstream between Ap,. and this term will be established later. The pressure
and down-stream from the perforated section as well as in the loss associated with accelerating the fluid in the pipe is simply
perforated section. The experiments show, as expected, a defi- given by
nite increase in the pressure drop per meter in the perforated
section. When the friction factor is calculated we find that the
Ap, = p(u(L)' - «(0)^), (4.4)
variation with the Reynolds number is as shown in figures 3(a, in which M(0) and u(L) denote the velocity at the start and end
b) for Pipes 1 and 2, respectively. In the same figure we have of the perforated section, respectively.
plotted the value of the friction factor as calculated using Haa- In order to determine the correction term Ap^, the friction and
land's equation (Haaland, 1983) acceleration pressure losses are subtracted from the measured
pressure drop {Ap„). The results is scaled with the friction
pressure loss Apf in order to be able to compare the results
/ = 1.8 log ( f + (4.1)
from different experiments. Plotting now ApJ Apf = (Ap,„ -
3.7D
Apf — Ap„)/Ap/versus the flux ratio qlQ,v^e, obtain the results
as shown in figures A{a, b) for Pipe 1 and Pipe 2 respectively.
with a roughness of e = 10"^ m for Pipe 1 and e = 1.5 X 10"^ In the figures all the data points from the experiments are used
m for Pipe 2. In the above equation, Re denotes the Reynolds and two things are clear. First of all, we observe that results
number of the axial flow, based on the pipe diameter D. We
observe that the experimental data follows the Haaland-curve
quite well. The calculated roughness for Pipe 1 is surprisingly
low given a perforation density of 64 holes per meter. As antici- 0.025 experimental
pated, the roughness of Pipe 2 is significantly larger than that values
of Pipe 1. The error analysis as presented in the appendix shows
that the errors in the friction factors were less than 2 percent
for both pipes.
We next turn our attention to the experiments with radial
inflow. In these experiments a given axial volume flow, Q, is I 0.015
specified and the radial volume flow, q, is increased stepwise.
Table 2 lists the axial and radial volume flow rates which were
investigated for Pipe 1 and Pipe 2. In Table 2, q denotes the 0,01 -+- -I- H- H
radial volume flow per unit length, that is ^ = qlL where L 50000 100000 150000 200000 250000 300000
denotes the length of the perforated pipe section. Reynolds number
The essential aim of the experiments presented here is to
determine a modification to the friction factor such that radial Fig. 3(a)
inflow effects can be taken into account. As the basis of the 0.045 n
friction pressure drop computations it is beneficial to be able • experiments
to use existing friction factors for pipe flow without radial in-
flow. In this way we naturally build on existing knowledge Haaland,
while extending it to take into account radial inflow effects. eps=0.0015
0.04-
Thus, we use Eq. (1.1) as the starting point for the frictional
•
pressure loss computation. The axial velocity u is constant when
no radial inflow occurs, so that, given the friction factor/, the r-—•— •
pressure drop can be computed using ( I . l ) . When radial inflow • " • -•
occurs, the axial velocity u is not constant so that (1.1) can not 0.035 •
be used directly. However, if we know the variation in « as a
function of the axial length, one can approximate the frictional
pressure drop via

0.03- 1 1 1— 1 1 1
^^•-Uf(u)u^dl. (4.2) 0 50000 100000 150000 200000 250000 300000
Reynolds number
It is, however, important to realize the restrictions of the
pressure loss estimate as in (4.2). The friction factor in the Fig. 3(b)
integral is the friction factor which is derived from fully devel- Fig. 3 A plot of the friction factor as a function of tlie Reynolds number
oped turbulent flow experiments without radial inflow (as in for Pipe 1 (a) and Pipe 2(b)

108 / Vol. 121, MARCH 1999 Transactions of the ASME

 45 •

-§" 0.15

1
40 • •

^u 0.1 35 •

0.05

•0.05 ^
0
Wta^^1 0.02
0.02 0.03
0.03 0.04
0.04 0.05 0.06
30 •

26 •

20
i I ^:
I

15 -
<>
•0,1 I
10 .
•0.15
6 •
-0.2
i 1 1 1 H 1

q/Q 0,000 0.005 0,010 0,016 0,020

Fig. 4(a)
0.15
Fig. 5 A plot of the parameter a as defined in Eq. (3.1.1) as a function
of perforated area fraction <&. The different symbols correspond to the
experiments of Schulkes and Utvik (1998) ( • ) , the experiments with Pipe
0.1
1 ( • ) , the experiments with Pipe 2 (A), and the experiments of Su and
Gudmundsson (1995) ( O ) . The error bars Indicate the experimental un-
certainties.
[Link]

IF
-+- -+- In Fig. 5 we have plotted the value of a as a function of the
0.04 0.06 0.08 0.1 0.12
perforated area fraction $ for the experiments of Schulkes and
•0.05 •• ^ Utvik (1998), the present experiments and for the experiments
• • • of Su and Gudmundsson (1995). The size of the error bars in
Fig. 5 is related to the envelope within which the data points
•0.1
as in Fig. 4 lie.
Figure 5 indicates that the correction term Ap,, is smaller in
-0.15 ^/Q the case of the pipes with a large wall roughness (like Pipe 2)
Fig. 4(6) than with the perforated smooth pipes. In Fig. 5 it is seen that
while there is a fair degree of scatter, it is interesting that the
Fig. 4 A curve of A p ^ / A p , versus the flux ratio ^/Q for the experiments quantity a is fairly constant for large ranges in the value of $.
with Pipe 1 (a) and Pipe 2 (b)
From the experiments we find that a = 26.1 ± 7.6 for the
perforated pipes. For the rather special geometry of Pipe 2, the
value of a is significantly lower. We can conclude that with
from different experiment collapse onto a well-defined line.
the experiments considered until now, the total pressure drop
Second, we see that the line passes through the origin. The fact
in a perforated pipe with radial inflow can be approximated by
that the lines in Fig. 4 passes through the origin is as anticipated:
in the limit q/Q-^ Owe have Ap„ -> 0 and the total measured
pressure loss Ap„, is given by the frictional pressure loss in Ap = Apy X ( 1 - aD ^ ) -h Ap„, (5.2)
fully developed turbulent flow, that is Apf.
The trends observed in the experiments are qualitatively the
same as those from the earlier experiments of Olsen and Eckert where Apf is as computed in (4.2).
(1966), Yuan et al. (1996), and Schulkes and Utvik (1998): At least some of the scatter in Fig. 5 will be due to the fact
for small values of the flux ratio the correction term Apc is that the values of a were obtained from experiments with vastly
negative. This implies that computation of the frictional pressure different geometrical properties as seen in Table 3. In addition,
drop on the basis of (4.2) leads to an over-estimate of the actual the experimental error in determining Ap,, may be significant,
frictional pressure drop. in particular for low values of the volume flux ratio q/Q. How-
ever, the error analysis as given in the appendix shows that for
the largest values of the volume flow ratio, the error in Ap is
5 Comparison Between Different Experiments of the order of 10 percent. It is in particular these large volume
It is interesting to see if it is possible to establish a relationship flow ratios which are used to determine the coefficient a.
between the present experimental results and similar experi-
ments carried out by Su and Gudmundsson (1995) and Schulkes
and Utvik (1998). Additional experimental investigations exist
6 Discussion
(e.g., Olsen and Eckert, 1966; Ouyang et al, 1996; Yuan et The experiments of pipe flow with radial inflow have lead to
al., 1996) but the data supplied in the papers describing these the pressure loss Eq. (5.2). In Section 2 we established the
experiments is insufficient to perform an analysis which is simi-
lar to that carried out in the previous section.
In the present experiments as well as those by Su and Gud- Table 3 Geometrical properties pipes in four different sets
mundsson (1995) and Schulkes and Utvik (1998) it was found of experiments
that Ap decreases with increasing values of the flux ratio. A
Pipe Pipe
detailed analysis of the data shows that in all the experiments diamete length
one can approximate the correction term via Experiment (m) (ra) *
Pipe 1 0.16 6 9.1 X IQ- 3
Ap, = -aDApfX^ (5.1) Pipe 2 0.16 8.15 2.1 X 10"
Schulkes and Utvik
in which a is an experimentally determined parameter and Apj (1998) 0.15 8.3 5.4 X 10" 4
is as calculated in Eq. (4.2). It turns out that the parameter a Su and Gudmundsson
(1995) 0.022 0.9 1.4 X 10" '-2.8 X 10-'
is fairly constant for a large range of different pipe geometries.

Journal of Fluids Engineering MARCH 1999, Vol 121 / 1 0 9

theoretical model given by Eq. (2.5) which contained an un- 4a$ _
known parameter \ which was related to the rate at which radial / = /X (6.7)
u
momentum was carried in the axial direction. Now it will be
shown what the relation between the parameters a and \ is.
Assuming steady state, it follows that upon integration, Eq. denotes the effective friction factor for the case of pipe flow
(2.5) becomes with radial inflow. Equation (6.7) shows that radial inflow leads
to a reduction in the friction factor. It should be stressed that
Eq. (6.7) is valid, provided the velocity ratio v/u < 1. Schulkes
f
Ja
dp „ ^ A A
-dl = Ap = Apf + Ap„
dl y yf y
4\fp

Q
f
Jo
vudl. (6.1) and Utvik (1998) showed experimentally that a rather abrupt
change from "lubrication" to "obstruction" occurs when v/u
« 1.
Equating the expressions for Ap in Eqs. (5.2) and (6.1), it
follows that It is instructive to estimate the volume flux ratio for which
the correction term Apc is significant. Using Eq. (5.2), it is
4 \v$p
$p r immediately clear that the correction term can not be neglected
Apf + Ap, vudl when
D Jo
aDq
= Ap^X (6.2) 1. (6.8)
1 -aD J ) + Ap,,
For a perforated section of length L in which a significant
from which one finds immediately that fraction of the liquid enters the pipe through the perforations,
we have q = qlL ?« QIL. Using this approximation together

f
aDq 4X$p with (6.8) shows that the correction term cannot be neglected
Apf vudl. (6.3) when
D
Jo
We now note that Q = •KR'^U(0) and q = 7rZ)$u where the (6.9)
radial velocity v is assumed to be constant. Substituting this L
into (6.3) and using the definition of Apf&s given in (4.2), the Equation (6.9) shows that the correction term Ap,, is im-
following expression for \ is readily found
portant only when the length of the perforated section is approx-
imately at), or less. This implies that only in cases where the
\ = J f{uWdl/\ udl. (6.4) axial flow increases significantly over a length of approximately
2M(0) 30 D, the correction term is expected to influence the frictional
pressure loss significantly.
Table 2 shows that the axial volume flow changes at most
10% due to radial inflow in the case for the experiments with 7 Conclusions
the lowest axial flow rates. For experiments with higher axial
flow rates the relative change of the axial flow rate is signifi- In this paper, results from experiments have been presented
cantly less than 10 percent. This observation allows us to sim- which show how the frictional pressure loss is influenced by
plify expression (6.4) by taking the velocity « to be constant radial inflow. Data analysis shows that the radial inflow leads
as a first approximation, that is u = a (0). It follows immediately to a reduction in the frictional pressure loss as compared with
that the case where frictional pressure losses are computed on the
basis of fully developed turbulent flow. The same trend in the
data was observed in experiments with large differences in per-
(6.5) foration geometry and experimental scale. The experimental
2^ results lead to a correction term to the frictional pressure loss
which takes into account changes due to radial inflow effects.
showing that the parameter \ is directly proportional to the
friction factor. While this was hard to anticipate without experi- The experimental results can be adequately explained by a
mental knowledge, the relation (6.4) is in line with physical one-dimensional model for pipe flow with radial inflow. The
expectations. Namely, that the rate at which radial momentum model contains one parameter which is related to the rate at
is carried in the axial direction should be proportional to the which the radial momentum is transferred in the axial direction.
friction factor, which in turn is related to the flow structure near This parameter is subsequentiy determined experimentally. Fi-
the porous pipe wall, is no real surprise. nally, a correction to the usual friction factor is established
which takes into account radial inflow effects.
For perforated pipe walls it was shown in the previous section
that a i=s 26.1. Together with variations in the friction factor as
shown in Fig. 3, it follows that X lies in the range 0.2 s X. s: Acknowledgment
0.25 for Reynolds numbers exceeding 5 X 1 0 " . For very rough The work presented in this paper was carried out as part of
pipe walls, like in the case of Pipe 2, it is found that X «=! 0.2. a co-operative research program on horizontal production wells
This suggests that the rate at which radial momentum is carried by Saga, Statoil and Norsk Hydro.
in the axial direction is fairly constant over a large range of
Reynolds numbers and for vastly different pipe geometries. It
is also important to note that X < 1 which is in agreement with References
the arguments leading to the introduction of X (see Section 2). Aggarwal, I. K., HoUingsworth, M. A., and Mayhew, Y. R., 1972, "Experimen-
Using (6.5) and (2.5) it is now possible to define an effective tal Friction Factors for Turbulent Flow with Suction in a Porous Tube," Interna-
tional Journal of Heat and Ma.<:s Transfer, Vol. 15, pp. 1585-1602.
friction factor for the case of pipe flow with radial inflow. Haaland, S. E., 1983, "Simple and Explicit Formulas for the Friction Factor is
Namely, Eq. (2.5) can be written as Turbulent Pipe Flow," ASME JOURNAL OF FLUIDS ENGINEERING, Vol. 105, pp.
89-90.
9 , ^ 9^2^ dp f 2 Moody, L. M., 1947, "An Approximate Formula for Pipe Friction Factors,"
(6.6) Trans. ASME, Vol. 69, p. 1005.
Olsen, R. M., and Eckert, E. R. G., 1966, "Experimental Studies of Turbulent
Flow in a Porous Circular Tube with Uniform Fluid Injection Through the Tube
in which Wall," ASME Journal of Applied Mechanics, pp. 7-17.

110 / Vol. 121, MARCH 1999 Transactions of the ASME

 Ouyang, L-B., Aiteibi, S., and Aziz, K., 1996, "General Wellbore Flow Model
for Horizontal, Vertical and Slanted Well Completions," Paper 36608, Society (A.l)
of Petroleum Engineers Conference, CO, Oct., pp. 1-13. f Q
Perry, A. E., Schofield, W. H., and Jouhert, P. N., 1969, "Rough Wall Turbu-
lent Boundary Layers," Journal of Fluid Mechanics Vol. 37(2). The accuracy of the pressure transducers and volume flow
Schlichting, H., 1936, "Experimental Investigation of the Problem of Surface control is such that 8ApJAp,„ = 0.002 and dQIQ = 0.005.
Roughness," NACA TM-823.
Schulkes, R., and Utvik, O. H., 1998, "Pressure Drop in Perforated Pipe with These values, together with (A.I) imply that the uncertainty in
Radial Inflow; Single-Phase Flow," Society of Petroleum Eengineers Journal, the friction factors is equal to 6f = 0.012/.
Mar., pp. 77-85. For the experiments with radial inflow (Figs. 4(a, b)) the
Schulkes, R., Utvik, O. H., and Rinde, T., 1997, "Pressure Drop in Horizontal relative error in the correction pressure term Ap^ is given by
Wells: Multiphase Flow with Radial Inflow," 8th International Conference Multi-
phase '97, A. P. Burns, ed., pp. 4.5-55. Mechanical Engineering Publications
Limited, London. 5Ap, SAp,„ + SApf + 8Ap„
(A.2)
Su, Z., and Gudmundsson, J. S, 1995, "Pressure Drop in Partly Perforated Ap, Ap,
Pipes," Report of Department of Petroleum Engineering and Applied Geophysics,
University of Trondheim—The Norwegian Institute of Technology (NTH).
Su, Z., 1996, "Pressure Drop in Perforated Pipes for Horizontal Wells," PhD Defining the flux ratio </> = qlQ, we find that the error in
thesis University of Trondheim—The Norwegian Institute of Technology (NTH). volume flow control implies 64>I4> = 0.005. In Schulkes and
Yuan, H. Sarica, C , and Brill, J. P., 1996, "Effect of Perforation Density on Utvik (1998) it was shown that <5Ap„, &Apf « pu^-8<f)l^ so
Single [Link] Liquid Flow Behaviour in Horizontal Wells," Paper 37109 Society that the magnitude of the numerator in (A.2) is approximately
of Petroleum Engineers Conference, Calgary, Nov., pp. 603-612.
equal to 0.002 Ap,„ + 2 X 0.005p«^ f« [Link]^. Using this
result in (A.2) and using the expression for Ap, as given in
APPENDIX
Eq. (5.1), it follows that
Error Analysis
In the experiments without radial inflow (Figs. 3(a, b)) the dAp, 0.02 Q
(A.3)
friction factors were determined experimentally on the basis of Ap, aLf
the measured pressure drop. Equation (1.1) was used to evaluate
the friction factor, given Ap and u. Standard error analysis where Ap/has been estimated using (1.1). Clearly, as the flux
methods show that the uncertainty 6/in the friction factor / i s ratio qlQ decreases, the realtive error in Ap, increases. For
related to the uncertainties in the measured pressure drop and example, di qlQ = 0.05 we have 6Ap, = 0.1 Ap, while at q/
the volume flow rates via Q = 0.01 the relative error has increased to SAp, = 0.5 Ap,.

Change of Address Form for (he Journal of Fluids Engineering

If you are planning Present Address - Affix Label or Copy Information from Label
To Move, Please
Notify The
ASME-Order Dep't
22 Law Drive
Box 2300
Fairfield, N.J. 07007-2300
Don't Wait! Print New Address Below
Don't Miss An Issue!
Allow Ample Time To Name
Effect Change.
Attention.
Address _
City . State or Country. -Zip.

Journal of Fluids Engineering MARCH 1999, Vol. 121 / 111

View publication stats