Int. J. Multiphase Flow Vol. 13, No. I, pp. 13-21, 1987 0301-9322/87 $3.00+0.

00
Printed in Great Britain. All rights reserved Copyright © 1987 Pergamon Journals/Elsevier

DETECTION OF SLUG FLOW

FROM PRESSURE MEASUREMENTS

P. Y. LIN and T. J. HANRATTY

Department of Chemical Engineering, Universityof I11inois,Urbana, IL 61801, U.S.A.

(Received 17 January 1985; in revised form 23 June 1986)

AlUraet--A slug pattern for gas-liquid flowin horizontalpipelinesis observed whenslugs of liquid block
the whole pipeline and move as a coherent mass downstream at a velocityapproximatelyequal to the
gas velocity.At low gas velocitiesthey are easily observed in a transparent pipe. However, at high gas
velocitiesit is difficultto differentiateslugs from large amplitudewaveswhich momentarilyclose off the
pipe cross section.This paper showshow the pressuremeasurementsat two locations,separatedfrom each
other in the flow direction, can be used to detect slugs.

1. I N T R O D U C T I O N

Gas and liquid flowing in a horizontal pipeline show a number of interfacial configurations, called
flow patterns, which have received considerable attention in the literature. One of these is slug flow,
in which liquid slugs, filling the whole pipe cross section, are intermittently propelled down the pipe.
This paper describes how a pair of pressure transducers can be used to detect the presence of these
slugs.
The motivation for this work came from a recent study in which the flow regimes were defined
for air and water flowing in horizontal 2.54 and 9.53 cm i.d. transparent pipes. At low gas velocities
a transition to slug flow was quite easy to detect visually. At high gas velocities this was not the
case. The slugs became much shorter and highly aerated, so it was not possible to detect whether
they filled the whole pipe cross section or whether they were durable. Therefore, it was necessary
to use an instrumental rather than a visual detection technique.
A summary of the different types of flow-regime detection techniques has been given by Hewitt
(1978) and by Jones (1979). The X-ray absorption measurements, used by Jones & Zuber (1974),
have the disadvantages of high cost and the need for safeguards. Barnea et al. (1980), Reimann
et al. (1981), and others have used conductance probes which detect the presence of a particular
phase by making use of the large difference in electrical conductivity between the gas phase and
the liquid phase. Jones & Zuber (1978) made use of a hot-film anemometer, but its fragility and
short lifetime have limited its acceptance. Hubbard & Dukler (1966), who first recognized the value
of instrumental techniques, used the power spectra of the fluctuations in the local pressure to
characterize different flow patterns. However. it is not at all certain that the frequency domain is
an appropriate basis to distinguish between slugs and large amplitude waves that close off the pipe
momentarily. Weisman et al. (1979) used the real-time differential pressure signal. They suggested
a scheme that uses the amplitude and frequency of the signal to characterize a particular pattern.
Tutu (1982) proposed using the probability density of the fluctuations in the pressure difference
for vertical systems.
We initially explored the use of conductance techniques (Lin 1985) to detect the slug-annular
transition. Because the slugs are highly aerated it was difficult, with this technique, to differentiate
a slug from a large-amplitude wave which does not block the pipe cross section. We, therefore,
decided to explore the possible use of a pair of pressure transducers 1.3-5 m apart in the flow
direction as a detection device. The technique is different from that used by Weisman et al. (1979)
in that the local pressures at two different locations were measured instead of the differential
pressure drop over a short distance. It is felt that the analysis of differential pressure drop is more
suited for flow patterns which are steady with respect to time rather than for intermittent flows.
The interpretation of the measurements is also quite different from that of Weisman et al.
13
14 P Y LIN and T J. H A N R A T T Y

The picture of a slug used in this analysis is that it blocks the pipe cross section, that it keeps
its identity until it reaches the exit of the pipe and that it moves approximately at the gas velocity.
Large disturbances, which from visual observations look like a slug, that do not have all of the
above characteristics are called pseudo-slugs. Dukler & Hubbard (1975) reported that the pressure
drop in horizontal slug flow takes the form of a steep jump when a slug is between the two taps.
The pressure in front of the slug is low and increases sharply to a much higher value behind the
slug. The pressure gradient behind the slug is small, but the pressure remains high. This, then, is
the type of pressure behavior that was used to identify slugs. The upstream pressure measurement
detected blockage from the characteristics of the pressure signal. The downstream pressure
measurement was used to determine whether the slug remained coherent.

2. D E S C R I P T I O N OF THE E X P E R I M E N T S

The experiments were conducted in two horizontal pipelines of 2.54 and 9.53 cm i.d. The
phase-mixing section were simple tees. The air was introduced in the branch and the water in the
run of the tee.
The experimental setup used to detect slugs is sketched in figure I. Two Viatran Model 218 strain
gauge pressure transducers with a flat frequency response up to 2 kHz and a range of 0-1.03 b were
used. The reference pressure to these transducers is adjustable to allow for increased pipeline
pressures at high flow rates. The transducers were located 200 pipe diameters away from the entry
so as to allow the slugs to develop. They were connected to pressure taps at the bottom of the pipe
located 1.35-5 m apart, depending on the flow conditions. The lines to the transducers were filled
with liquid and frequently purged of air bubbles.
Two sets of parallel wire conductance probes, that extended over the pipe cross section were used
to measure the heights and velocities of waves and pseudo-slugs. These provided a better
understanding of the flow but are not necessary for the interpretation of the pressure signals.

Parallel
wire

~;I Reference
u~r~
~ ~i Upstream
Oo.os,ream I I
pressure
transducer

c,q~
Compressed
Pressure o)r
gouge

ent

Figure 1. The experimental setup.

 DETECTION OF SLUO FLOW VI./~ ~K~SURE MEASUREMENTS 15

9 53 cm p)pe
0.6" VSG. 152m/s
VSL" 0 62 m IS

0.4'
r~
L ~ dowfl~ttoam

02 ~dk

O"
4-
-3

0 I upstream
x rllslure Q.

¢. W
bJ
rr
o3
bJ
O~
b.J n
rr
{1.

-0

3 25 425 ~5 6~5 725 8 25

TIME[ (s)

Figure 2. Wave-heightand pressure-fluctuationmeasurementsin slug flow.

Signals from the measuring devices were recorded simultaneously in two different ways: a
Honeywell Model 1508 C Visicorder with a frequency response of I kHz was used to print out a
trace representing the signal. This recorder produces the traces by deflecting high-intensity light
beams that are focused on a moving light-sensitive chart. The signals were also digitized and stored
in a microcomputer for later analysis.

3. RESULTS

(a) Pressure signals for slugs

Figure 2 shows a typical set of recorded pressure measurements and the signal from the upstream
pair of parallel wire probes, located 3.9 cm downstream from the downstream pressure transducer,
which gives the ratio of the height of the liquid to the pipe diameter, hiD. It can be seen that at
about t = 4.05 s a slug arrived at the upstream tap. This causes the pressure at that point to increase
sharply because the cross section of the pipe is plugged. The slug has not reached the downstram
pressure tap, as evidenced by a low h/D at that location. The downstream pressure does not
experience any increase in pressure at this time because the upstream slug blocks the high pressure
air behind it. When the slug reaches the downstream pressure tap the pressure and hid
simultaneously experience a sudden increase. A comparison of the h/D and downstream pressure
signal shows that the pressure increase commences at the front of the slug and reaches a maximum
at the tail of the slug (indicated by the drop in h/D) and then remains almost constant.
The pressures at both taps remain high until the slug leaves the pipe, at which point the two
pressures drop simultaneously. This behavior indicates the slug sustains a competent blockage of
the pipe throughout its existence.
Another distinctive feature of slug flow is that slugs travel approximately at the gas velocity, as
shown by Hubbard (1965), Nicholson et al. (1977) and others. Pseudo-slugs and large waves, on
the other hand, do not maintain a pipe blockage and arc expected to travel at a speed much lower
16 P. Y. LIN and T. J. H A N R A T I ' Y

9.5:3 cm pipe
VSG- 23 2 m/s
0 4"] VsL-O 22 m/s wove hmQl~l

21 upsfrlom

O
x

I I i I i i

ko
Iz
to
w

0. downstream

1-

i b i i i I
0 I ,5 30 45 60 7'5

TIME (s)

Figure 3. Wave-height and pressure-fluctuation measurements for pseudo-slugs.

than the gas velocity. The delay time between the arrivals of the two pressure signals gives a means
of measuring the slug velocity. For the signals shown in figure 2, the delays for the two slugs are
0.1625 and 0.1500 s. The pressure taps were 2.25 m apart, so slug velocities of 13.8 and 14.9m/s
are to be compared with a superficial gas velocity, VsG, of 15.2 m/s.

(b ) Pressure signals for pseudo-slugs

Figure 3 shows pressure signals observed in the pseudo-slug regime. Large peaks in the h/D signal
correspond to the observation of pseudo-slugs moving along the pipe. However, no large pressure
pulses, of the type in figure 2, are observed. The reason is that the gas phase can flow through the
pseudo-slug so that a huge pressure build up behind it is not possible. A measurement of the time
delay in the wave-height signals shows that the pseudo-slug velocity is 2.83 m/s compared to a value
of VSG of 23.2 m/s.
For most cases slugs can be distinguished from pseudo-slugs by examining the amplitude of the
pressure fluctuation at a single location. However, in some cases, especially when both the gas and
liquid throughputs are high, the distinction is not so clear-cut. In these cases, pseudo-slugs plug
the pipe momentarily and cause considerable pressure fluctuations. Since the blockage is only
momentary, the pressure is not sustained for the entire duration of the existence of the pseudo-slug,
as would have been the case for slugs. The second pressure transducer distinguishes whether the
blockage of the pipe remains coherent through a long section of the pipe. Furthermore, the use
of dual pressure transducers gives the direction of travel of the pressure wave, hence eliminating
ambiguities caused by possible pressure reflections from downstream.
Figures 4a,b show the pressure signals obtained for annular and wavy stratified flows. The
pressure fluctuations are much smaller than those observed for slugs and pseudo-slugs.

(c ) Cross-correlation functions
A convenient way to represent the pressure signals for slug flows is to compute the cross-
correlation function of the two pressure signals.
 DETECTION OF SLUG FLOW VIA PRESSURE MEASUREMENTS 17

4" (0) annular

VSG-45.t m/s
Vsu,0 22 rn/s

ud

03
cn
uJ
(b) svrot,hea
VSG• r 06 m/s
VsL-O 166 rn/s

O~o - -I2 ~, ~, ~ ,b
T I M E (s}

Figure 4. Pressure signals in (a) annular and (b) stratified flow in a 9.53 em pipe.

If the pressure records as a function of time t at the upstream and d__~ownstream locations are
represented by Pu(t) and PD(t), respectively, then the means Pu and PD are given by

P----u= lim 1 F r
r ~ T .]o Pu(t)dt

and

= r-Do
lim T1 for PD(t) dt.

The cross-covariance CuD of the two signals at delay time t is then given as

CUD(t) = rlim
~ - T1 fo r [Pu(t)-- -Pu][PD(t + t ) - - -PD]dt.

The autocorrelations at time delay zero, Ru(0) and RD(0), of the two signals are given by

Ru(0) = lim 1
T~cc "T
for [Pu(t) -- Pu] 2dt
and
T -

RD(0) = lim [PD(/) -- PD] 2 dt.

The cross-correlation function, Rub(t), of the two pressure signals, Pu(t) and PD(t), is obtained
by normalizing CUD(Z) with Ru(0) and RD(0):

Rub(t) = CuD(t)
Ru (O)RD (0)"

The cross-correlation function is hence dimensionless and has a value of between - 1 and + I.
Figure 5a is a cross-correlation of the pressure signals in slug flow, a portion of which is presented
in figure 2. A characteristic peak at a time delay of 0.145 s can be obsorved in the cross-correlation.
This represents an average of the time delays of all the slugs in the record, including those shown

M F I~I--B
18 P. Y. LIN end T. J. HANRATTY

10 ~ L ~ - O 145s ( O ) 9 53cm pipe

I ~ VSG= I 5 2 m / s
0.7'
VSL.0 62 m / s

04

01

a: - o 2
t~
o
0 -0 ! i t I I I I I I

( b ) 9 53 cm p ~
oe VSG"23 2 m/s
VSL'0 2 2 m / s
06

04

02

-02 I l I I I I I I
0 05 10 15 20

T I M E DELAY (s)

Figure 5

9 53cm pipe
7- ( a ) VSL. I 0 1 m / s

9 53cm pipe
( C ) slug
5- 5-- VSL I 0 46m/s

O
x
o
v
~J
0:
D
3-

3- l
(/~
UJ
rv I
0.
W
t~ [ i 1
D 22.50 24 0 0 25 50 2700 28 50 3O0O

rr
tL ( d ) pNuao - slug

3--

2-

, i t T ~ ~ T ~ ,
925 ~25 11 25 t2'25 1325 1425 I 5 t6 t 7 18 19 20

TIME (S) TIME (s)

Figures 6a-d. Pressure signals m slug flow at VsG = 33 m/s.

 DETECTION OF SLUG FLOW VIA PRESSURE MEASUREMENTS 19

100 . ~ (G)V~ - t . 0 1 m / s 1.O0

O 75 ~ 0.77
0 50 0,53

0.25 0.30
0 0.07

"~
-025-- - 0.17I
r~ -0.5C ' I , I I I I I I - 0.40 I I I I I I I I
o 0.31 063 0.94 125 0 050 100 150 2,00
(J
(fl
(/)
0
n~
091 (d) VsL-O31m/s
0.7

O~

0-~

0.1

-0.1 - 0.t
--05 I I 1 l I I I I -0 I
0 0.5 1.0 t .5 2.0 0 05 1.0 1.5 2.0

T I M E DELAY(sl

Figure 7. Cross-correlation in slug flow at Vs6 = 33 m/s.

in figure 2. As discussed previously, this time delay compares well with the superficial gas velocity
at the measurement points.
Figure 5b is a cross-correlation of the pressure signals in the pseudo-slug flow represented in
figure 3. This cross-correlation function does not show a high degree of correlation except at zero
time delay and, hence, is distinctively different from that for slug flow.
Figures 6a-d are portions of the pressure signals obtained at a constant VsG of 33 m/s in the
9.53 cm pipe that illustrate the transition from slug to pseudo-slug flow. The corresponding
cross-correlations of these measurements are presented in figures 7a-d. The transition from slug
to pseudo-slug flow is distinctive in both the pressure signals and the cross-correlations.

) ups?reom Dressure
2

(b)

,¢
o
0
1

,,
n.,
w 03
0
¢J o.1
O?

.•" • 054s

u)
n clownstre(]m pressure
~ 0,3
05 / r ' r i i ! ~

0 05 1.0 15 20
i
TIME DELAY (s)

0 1'" Cross Correloflon

12 1 , 116 ~.....
14 18
2 5 4 c m pipe
TIME (s)
Vso = 7 1 m / s
VSL" 0 . 3 4 m / s

Figure 8. Cross-correlation of pressure signals for slug flow at low gas velocities.
20 P. Y. LIN and T. J. HANRATTY

80

70

60

E 50
>-
t-
G 40
q
w
> 3¢

-- 0
_J
m 2C

10

0 ~ I I I I I I I
10 20 30 40 50 60 70 80

LOCAL SUPERFICIAL GAS VELOCITY ( m / s )

Figure 9. Slug velocity vs Vs~ for the 2.54cm pipe,

The shapes of the cross-correlation functions presented in figures 7a-d are typical of most of
the slug flows observed in the 2.54 and 9.53 cm pipes. However, at low gas velocities the
cross-correlation function may have a shape such as that shown in figure 8b. Figure 8a shows a
portion of the pressure signals from which figure 8b was computed. The reason for this behavior
is that the pressure signals are not "true" time-delay signals. Once the slug passes through both
pressure taps, the pressures at both taps are almost identical. Provided the pressure signals on both
sides of the pressure plateau for each probe are identical, as is the case in figure 8a, the signals
for the two pressure taps will give the same degree of cross-correlation for all time delays from
zero up to that which corresponds to the slug velocity. The pressure signals for slug flows at higher
gas flow rates are not so symmetrical about the pressure plateau as for the case in figure 8a and
hence do not generate a cross-correlation of the shape shown in figure 8b.

(d) Slug and pseudo-slug velocities

The slug velocities obtained from the cross-correlation of the pressure signals at various flow
conditions are summarized in figure 9 for the 2.54 cm pipe and in figure 10 for the 9.53 cm pipe.

30
' ' ' 19 53 ' ' '

28
A
i¢1
x. 2 2
E

ID

Ld 14 - 0 0 . 0 0
>
(.9

271 I I I i I I i I
0 2 6 ~0 ~4 18 22 26 30 34

LOCAL SUPERFICIAL GAS VELOCITY ( m / s )

Figure 10. Slug velooty vs YSG for the 9.53 ¢rn pipe.
 DETECTION OF SLUGFLOWVIA PRESSUREMEASUREMENTS 21

20 l' , l = i J

18-- V$L , m / s
tn
46 O 0122
E
>-
I--
14
2 54 ¢m
pipe IA 0,061
tO
o 12 9 53¢m •{. 0 2 1 6
..J pipe
LIJ
>
10
(.9
_J
(/') 8

6
LU
(/3
n 4
+
2

0 I I
0 4 8 12 16 20 24 28 32 36

SUPERFICIAL GAS VELOCITY (m/s }

Figure 11. Pseudo-slug velocity vs VSG.

In both cases, the slug velocity is plotted against the superficial gas velocity at the pressure taps.
The agreement between the slug velocity and VSGis quite good for both cases. Figure 11 presents
some pseudo-slug velocities measured from the cross-correlation of the parallel-wire conductance
probes. The velocities of pseudo-slugs are much lower than the gas velocities, as expected.
Acknowledgement--This work has been supported by the AIChE Design Institute for Multiphase Processing
and by the Shell Companies Foundation.

REFERENCES
BARNEA,D., SHOHAM, O. • TAITEL, Y. 1980 Flow pattern characterization in two phase flow by
electrical conductance probe. Int. J. Multiphase Flow 6, 38%397.
DUKLER,A. E. & HUBBARD,M. G. 1975 A model for gas-liquid slug flow in horizontal and near
horizontal tubes. Ind. Engng Chem. Fundam. 14(4), 33%347.
HEwrrr, G. F. 1978 Measurement of Two Phase Flow Parameters. Academic Press, London.
HUBBARD,M. G. 1965 An analysis of horizontal gas-liquid slug flow. Ph.D. Thesis, Univ. of
Houston, Houston, Tex.
HUBBARD,M. G. & DUKLER,A. E. 1966 The characterization of flow regimes for horizontal
two-phase flow: I. Statistical analysis of wall pressure fluctuations. In Proc. 1966 Heat Transfer
and Fluid Mechanics Institute (Edited by SAAD& MILLER).Stanford Univ. Press, Stanford, Calif.
JONES, O. C. 1979 Two-phase flow measurement techniques in gas-liquid systems. Presented at the
Minnesota Short Course on Fluid Mechanics Measurements.
JONES, O. C. E¢ ZUBER, N. 1974 Statistical methods for measurement and analysis in two-phase flow.
5th Int. Heat Transfer Conf. Sci. Coun. Jap. 4.
JONES, O. C. & ZUBER, N. 1978 Use of a hot-film anemometer for measurement of two-phase void
and volume flux profiles in a narrow rectangular channel. AIChE Syrup. Ser. 174, 191-204.
LIN, P. Y. 1985 Flow regime transitions in horizontal gas-liquid flow. Ph.D. Dissertation, Univ,
of Illinois, Urhana, Ill.
NICHOLSON, M. K., AzIz, K. & GREGORY, G. A. 1977 Intermittent two-phase flow in horizontal
pipes: predictive models. Presented at 27th Can. Chem. Engng Conf., Calgary, Alberta.
REIMANN, J., JOHN, H. & SEEGER, W. 1981 Experiments on the transition from slug to annular flow
in horizontal air-water and steam-water flow. Presented at the Mtg Fur. Two-phase Flow Up.
Ttrru, N. K. 1982 Pressure fluctuations and flow pattern recognition in vertical two phase
gas-liquid flows. Int. J. Multiphase Flow 8, 443-447.
WEISMAN, J., DUNCAN, D., GIBSON, J. & CRAWFORD, T. 1979 Effects of fluid properties and pipe
diameter on two-phase flow patterns in horizontal lines. Int. J. Multiphase Flow 5, 437-462.