Sunteți pe pagina 1din 11

International Pipeline Conference Volume 2

ASME 1996

IPC1996-1882
PIPELINE PURGING PRINCIPLES AND PRACTICE

James E. Johnson, Steven J. Svedeman, Christopher A. Kuhl


Southwest Research Institute, San Antonio, Texas

John G. Gregor
Gas Research Institute, Chicago, Illinois

Alan K. Lambeth
Texas Eastern Transmission Corporation, Houston, Texas

ABSTR ACT Nb - Number o f pipe bends


Gas purging, a process o f displacing one gas by another p - Fluid density [lbm/ft3]
gas, occurs on a routine basis in the natural gas industry when Re - Reynolds Number
pipelines are purged into and out o f service. In a project t - Purge time [sec]
sponsored by the Gas Research Institute and in cooperation V - Velocity [fit/sec]
with the American Gas Association (A.G.A.) the purging x Distance from pipe inlet [ft]
practices as outlined in the A .G A .s P urging P rinciples a n d
P ractices manual are being reviewed because many o f todays INTRODUCTION
pipeline purging operations occur under conditions not ad The Operating Section o f the American Gas Association
dressed directly in the manual. The program focus is on the (A.G.A.) has become involved in the task of reviewing and,
purging procedures outlined in Chapter 8 o f the manual enti as necessary, revising its Report, Purging P rinciples and
tled G as Transm ission an d D istribution P ipes. P ractice, Catalogue No. XK0775. While this report is in its
third printing (1990), it is in its second edition (197S) and
The technical objective o f the project is to develop an reflects the recommended purging guidelines appropriate
understanding o f the scientific principles upon which safe, twenty years ago. This report is commonly referred to as the
practical purging practices can be based. Direct displacement Purging Manual and this terminology will be used in the
and inert gas slug purging operations are explained in terms remainder o f this paper.
of dispersion and mixing parameters and their relationship to
the gas velocity. Field data is compared to the results of an The A.G.A. Pipeline Committee has concentrated its ef
analytical mixing model. Computer software for planning forts on Chapter 8, G a s Transmission a n d D istribution P ipes,
safe and cost effective pipeline purges has been developed. o f the Purging Manual and has identified the need for numer
Finally, recommendations for revising Chapter 8 o f the ous revisions. In particular, the gas industry believes that
A.G.A. manual are presented. Table 8.1, entitled Purging Data for Inlet Control Proce
dure," may not be entirely sufficient for todays operating
NOMENCLATURE conditions and blow down configurations.

Ai, A2 - Constants The Gas Research Institute (GRI) is supporting a re


C - Gas concentration [mole %] search program initiative in cooperation with the A.G.A.
d - Pipe diameter [ft] Pipeline Committee to develop improved procedures for
Dc Turbulent dispersion coefficient fft2/sec] pipeline purging operations. Southwest Research Institute
g - Gravitational acceleration [ft/sec2] (SwRI) was commissioned to conduct the research for provid
g' - effective gravity, defined in Eq. (2) [ft/sec2] ing the necessary technical basis for revisions to pipeline
L - Pipe length [ft] purging operations. The research program was initiated in
Lmi, - Length o f the mixed gas region [ft] November o f 1994 and is scheduled for completion in the

Copyright 1996 by ASME

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


summer of 1996. The intent of this paper is to provide an liquids, odors, and obtaining supply gas (like nitrogen if
overview of an ongoing purging program and to the extent the inert slug method is used).
possible, make preliminary recommendations for revising
Chapter 8 of the Purging Manual. Nearly all of the company procedures incorporated Table
8.1 o f the A.G.A. Purging Manual as part o f their stan
BACKGROUND dard purging procedures.
The goal of the A.G.A. Pipeline Committee, GRI, and the
gas industry is to improve industry guidelines for pipeline Numerous comments were received about Table 8.1.
purging practices (focus of Chapter 8). The GRI/SwRI re Generally, the respondents were interested in the techni
search effort includes pipeline simulations and field studies. cal basis o f the table and what to do when required to
Specific objectives of the research are: purge systems outside o f the range o f pipe sizes and
lengths shown in the table.
Gain a better technical understanding of direct and inert
slug purging processes, LITERATURE SURVEY
Gas mixing processes and purging fluid mechanics de
Provide recommendations for improved purging opera scribed in Chapter 8 o f the Purging Manual are based to a
tions supported by analysis and field studies, great extent on the work by Henderson (1941) and Cortelyou
and Curtis (1964). Hendersons (1941) paper introduces the
Develop user-friendly personal computer (PC) software concepts o f gas stratification and turbulent mixing and the
for planning purge operations, and report addresses factors involved in the formulation of a
moving barrier or slug of inert gas between the air and natural
Recommend revisions to Chapter 8 of the A.G.A. Purg gas. It is this paper where the purging velocity criteria o f 100
ing Manual. feet per minute seems to originate which is embedded
throughout Chapter 8. Henderson (1941) recommends that
SURVEY OF CURRENT PRACTICE when exhaust gas or nitrogen is to be used for the inert slug,
To better understand current purging practices, a nine the velocity of the purge gases must not be below 100 feet per
teen question survey was sent to 68 gas industry groups re minute and preferably be 200 feet per minute or higher. The
sponsible for purging operations. Thirty-six organizations velocity should not be allowed to fall below 300 feet per mi
responded (53% return) and twenty-five company purging nute when carbon dioxide is the inert gas used. This paper
procedures were provided. While it is impractical to present also discusses the effects o f turbulence, delays during purg
all details found within the survey in this paper, key findings ing, and pipeline static pressure on the retention o f the slug.
are summarized below:
The Cortelyou and Curtis (1964) paper extends the work
Pipe sizes involved in purging range from 0.5 inch to 48 o f Henderson (1941). In this paper, an empirical model o f the
inches in diameter. (The smallest pipe size considered in slug shortening process is developed based upon experimental
the present study is 4 inch diameter.) data obtained from tests with 12, 20, and 34 inch diameter
pipe. The curves found in the Purging Manual for slug short
The typical length of pipe purged at one time ranged ening have been extracted, from this paper. Conclusions
from 600 feet to 40 miles. (In this paper, results are drawn by Cortelyou and Curtis (1964) are (1) purge velocities
shown for purges of 75 and 50 mile runs.) less than 100 feet per minute in large pipelines allow stratifi
cation, (2) Reynolds number is not a satisfactory criterion for
The most common method for purging into service is the purge velocity because large pipelines require greater purge
direct method. velocities than smaller pipelines, and (3) changes in horizon
tal or vertical direction because o f ells or return bends do not
The air mover method is the most common method for destroy the nitrogen slug.
removing gas from the pipeline, especially for field work
involving welding operations. (Normally, a blowdown Marshall, et al. (1992) address the explosion hazard as
will take place first to vent the majority of the gas. Con- sociated with the direct purging method. Explosion pressure
cons exist when venting in congested areas.) generated on ignition o f a mixing zone was related to the ratio
of flammable volume as a fraction o f the pipeline swept vol
The most common areas of concern are: vent noise (not ume. While detailed explosion pressure data was not pro
addressed in this paper), predicting time of completion of vided, the authors comment that pressures developed on igni
the purge, purging of branches (not addressed in this pa tion o f the mixing zones formed under stratified flow condi
per), controlling pressure, and actually knowing when the tions were considerably higher (by approximately a factor of
purge is complete. Other areas of concern were pipeline 10) than those generated by ignition o f the mixing zone
foimed during wedge flow. Generally, for near atmos

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


pheric pressure purges, the overpressures generated are in the length of the flammable region for direct purges and
significant when compared to the strength of most piping for predicting slug shortening for inert slug purging.
systems designed for transmission and distribution pressures.
This would imply that the control pressures recommended in Generally, the literature addresses the fluid mechanics of
Table 8.1 (typically less than SO psig) for transmission piping pipeline mixing in terms o f a key control variable which
are conservative with respect to potential pipe rupture with is gas velocity. Unfortunately, when purges are per
respect to current British Standards, the authors have con formed in the field, a direct measure o f gas velocity is
cluded that (1) current restrictions on the use of direct purging usually not available. Pipeline velocity can be inferred
are over conservative and unnecessary for purging operations from pressure drop calculations, but the pipeline geome
involving natural gas, and (2) the use o f direct purging can be try must be known. In effect, the purge pressures rec
extended to longer pipe lengths and larger diameter pipes than ommended in Table 8.1 of the Purging Manual are suffi
currently permitted without compromising safety, provided ciently high so that turbulent flow exists and therefore the
that the purge velocity is equal to, or greater than, the critical purge meets an acceptable velocity criteria.
velocity required to prevent stratified flow. While the above
comments are directed toward the British purging standards, By supplying the upstream purging pressure suggested in
the methodology does provide the basis for specific velocity the Purging Manual the resulting flow velocities, for the
criteria. Simply stated, one should be flowing at a velocity majority of the piping configuration, will be well above
above the stratification velocity and if overpressure is a con that needed to minimize the effects o f axial dispersion,
cern, one should not let the volume of mixed gas become such as stratification.
large enough to support detonation.
MODELING CONSIDERATIONS
Perkins and Euchner (1988) discuss calculation methods Presently, a unified mathematical model is being devel
for sizing an inert slug. Dispersion in turbulent flow in con oped for calculating stratification velocity, combustible gas
junction with flammability limits are discussed. Major con mixture length, and purge time. The types o f purges ad
clusions and recommendations are: (1) slug sizing equations dressed by the model include direct purging and slug purging.
should be derived for the flammability envelope for the par In this particular paper, only the topics o f minimum purging
ticular natural gas o f interest. In the absence o f detailed velocity (to prevent stratification) and mixture length as ap
flammability data, the slug-sizing function for ethane can be plied to direct purging methods will be discussed.
used as a conservative approximation for most natural gases,
and (2) to minimize dangers associated with flammable gases, STRATIFICATION VELOCITY
the flowing pressure should be kept as low as possible. When the density of the purged fluid (the gas to be
Perkins and Euchner (1988) also point out in their paper that removed from the pipeline) is different from that o f the
the air and gas mixture zone should be less than about 60 pipe purging fluid (the gas that replaces the purged fluid), the
diameters to eliminate possible detonation which can generate purging flow velocity must be larger than some minimum
considerable overpressures due to the generation o f shock value for good results. This requirement can be understood
waves. As will be discussed later, software developed for by examining the displacement o f one fluid by another in a
planning purges incorporates calculation procedures dis pipeline, as sketched in Fig. 1. Assume that the two fluids are
cussed in this paper. initially separated and stationary, such that the lighter fluid is
on the right The heavier fluid attempts to flow under the
Woodhead, et al. (1971) provide information on the tur lighter fluid, and the lighter fluid attempts to flow over the
bulent gas dispersion process which is important for predict heavier fluid. These flows are driven by the difference in
ing slug shortening. The slug shortening process occurs when hydrostatic heads between the fluids. The velocity V o f this
an inert slug (used as a buffer between natural gas and air) is gravity wave depends on the pipe diameter d and the den
dispersed on each end o f the slug. Consequently, the air mi sity difference pr P2. Experiments (Linden and Simpson,
grates through the inert slug toward the gas and on the other 1986) have found that:
end of the slug, gas migrates toward the air. Given sufficient
time, the air will meet with the natural gas. This process is
termed slug shortening.

SIGNIFICANCE OF LITERATURE REVIEWED


Based upon the literature review and upon our prelimi
nary analysis, the following observations were made.
The numerical factor depends on the geometry o f the pipe
Common purging practice employs both direct purge and
cross-section and it should be noted that the shape o f the
a slug purge method. A single dispersion model, ad
wave is more complicated than that shown in Fig. 1. The
justed for gas specific properties, can be used to predict
interface is spread over several diameters o f the pipe.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


G ra vity w a v e - sta gn an t fluids

W a v e suppression - turbulent flow

Figure 2. Minimum purging velocity - direct purging.

wave, since it is an instability that cannot be completely sup


pressed; each fluid will always penetrate the other to some
extent even if the velocity is constant over the cross-section.)
The effects o f this leaking can be minimized by requiring the
flow to be turbulent. For one thing, a turbulent velocity pro
file more nearly approximates the desired uniform velocity
W a v e suppression - lam inar flow
than does a laminar velocity profile. But more importantly,
Figure 1. Suppression of a gravity wave in a pipeline. when the flow is turbulent, any fluid that leaks into the other
fluid will be rapidly mixed with it; since this makes the inter
When there is an overall flow in the pipe, the gravity face less distinct, there is less driving force to propagate a
wave interacts with the mean flow. If, as shown in the middle gravity wave. So for good purging, the purging velocity
sketch o f Fig. 1, the flow is to the right with a velocity equal Vpurge must be at least as large as the gravity wave velocity
to V over the entire pipe cross-section, fluid 1 continuously V, but it m ust also be large enough to ensure a Reynolds
pushes the gravity wave in the upper half (fluid 2) back Num ber greater than 2500 in pipe flow.
towards the interface. Since fluid 2 is prevented from flowing
over fluid 1, there is no volume made available for fluid 1 to There is evidence from purging experiments to confirm
flow under fluid 2 (conservation o f mass or volume). Conse these two conclusions. Figure 2 shows the results of a series
quently, propagation o f the gravity wave is suppressed by the o f direct purging experiments by Marshall, et al. (1992) to
flow in a pipe (and stratification or channeling o f the fluids is determine the minimum purging velocity that will prevent the
prevented), just as a shock wave in a compressible fluid can occurrence o f channeling (lighter fluid flowing over heavier
be brought to rest by an oncoming flow that has a velocity fluid). The bars in this plot indicate that the minimum veloc
equal to the shock velocity. When the gravity wave is sup ity was determined for each o f a series o f pipe sizes, and this
pressed, the purging fluid displaces the purged fluid by push velocity was then recommended for a range o f bounding pipe
ing it bodily towards the right. This explains why a minimum sizes. The figure also shows correlations o f a gravity wave,
purging velocity is needed to prevent channeling and to ob Eq. (1), derived from the data so as to bracket the recom
tain good purging. mended velocities. It is evident that the required velocity is
proportional to Vd, as Eq. (1) implies. The mid-point o f the
Unfortunately, it is not possible to have a uniform veloc bracketing correlations can be used as a purging specification:
ity V over the entire cross-section since the velocity must de
crease to zero near the wall. Hence, near the wall, the gravity
wave speed is greater than the flow velocity, and some o f the
Pg as
heavier and lighter fluids will leak past each other. (In
addition, a gravity wave is not exactly analogous to a shock
W = 0-7Vi7^" g ' = 2g ( 2)
air + P y

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Equation (2) accurately represents the data for d > 5
inches. For d < 5 inches, the experimentally-determined
minimum Vpurge was, however, substantially greater than that
predicted by Eq. (2). The reason for this is shown by the
Reynolds Number Re plot in Fig. 2 (Re was computed for
natural gas, which has a slightly larger kinematic viscosity
than air). For d < 3 inches, Re based on the Vpurge computed
by Eq. (2) is less than 1500, so the flow would not be turbu
lent In an experiment a larger Vpurge would have been found
to be necessary. For 3 < d < 5 inches, Re based on Eq. (2)
yields a mildly turbulent flow, but nonetheless Marshall, et al.
(1992) found that a somewhat larger Vpurge is also needed for
these pipe sizes. In fact, it appears that the minimum test ve
locity used in their work was never less than 100 ft/min.

With these results, a recommended minimum value o f


direct purging velocity Vpurge >s calculated from Eq. (2). If the
purge velocity is greater than 100 ft/min and if Re > 2000 the
answer is acceptable; otherwise, choose Vpurge so Re > 2000,
but if this value is less than 100 ft/min; then choose 100
ft/min.
Pipe Diameter, in
The values o f computed from Eq. (2) are somewhat
larger than the minimum purging velocities found in another Figure 4. Length of mixed region.
test program (Cortelyou and Curtis, (1964)). Also the test
results shown in Fig. 2 may be conservative, because it is not A fraction o f the mixed region is combustible, and if this
known if the data represents the true minimum velocity fraction exceeds a critical length (depending on pressure
needed to prevent stratification. The 0.7 factor used in Eq. and temperature), a low-pressure deflagration can build
(2) could be reduced to account for a lower stratification ve up to a high-pressure detonation. (An ignition source
locity however, a better coefficient value is not known at this must be present.)
time.
It is known from the turbulent mixing theory that the
MIXED REGION LENGTH length o f the mixed region increases more rapidly in time as
When the purging fluid displaces the purged fluid, there the flow velocity increases. The increase is less than linear,
is a region where, because o f turbulence, the fluids become however, so for a given pipe length, the total length of mixed
mixed, as shown schematically in Fig. 3 by the backward-S- fluids decreases as velocity increases - because the available
shaped concentration profile. The length o f this region in time decreases linearly with 1/velocity. Figure 4 (taken from
creases as the purging fluid travels down the pipeline. The Marshal, et al. (1992)) shows this effect. The mixed length
details o f how the mixed length increases in time are o f inter approaches the total available length between the inlet and the
est for several reasons: measurement location (210 feet) for the smallest velocity used
in these experiments (which is just about the smallest allowed
When the length o f the mixed region approaches the velocity from Eq. (2)), but for larger velocities the fraction of
length o f the pipe, purging is by dilution rather than by the available length occupied by the mixed region decreases.
displacement, and this is an inefficient type o f purging. Consequently, the length o f the mixed region can be con
trolled by increasing the purging velocity.

These observations can be quantified by the turbulent


mixing theory. Turbulence is affected when the mixing fluids
have different densities (because the small eddies are acted on
by buoyancy forces as well as by inertial forces), but simple
relations are not available for this situation so mixing rela
tions for fluids having the same density are used here.
Perkins and Euchner (1988) show that the backward-S-shaped
concentration profile o f Fig. 3 is described mathematically
by:
Figure 3. Concentration profile for direct purging.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


(* Spurge )
0 .5
T T T

^pur^c 2
2y f J c
o I lengtn
(3)
xV,purge (x + Spurge ) o 0 .4 pupe htafoce locellen
o
+TexP erfc

u
T Purge Velocity
\ 2 ^
r O 1 9 7 H/min

0 .3
where erfc is the complimentary error function.
2D>
a
Equation (3) is a complicated expression, but it can be
simplified substantially when the value of De/xV purge is small o> 0.2
c
(as it usually is, except right at the inlet where x is smaller).
Therefore, die entire second part of the expression on the JD
n
right o f Eq. (3) can then be dropped as being negligibly small.
E 0.1
W ith this simplification, the mathematical relation between x E
and t where the concentration remains constant at any desired
LL
specific value is found from Eq. (3) to be:

0.0 L
0 50 100 150 200 250
Location of Purging Interface Down the Pipe, ft
----- O f1
' = constant (4)
Figure 5. Flammable length for direct purging.

Thus, the length of the combustible region increases as the


Here constant (= 2 X argument of erfc) is a positive number square root of time.
that depends on the concentration of interest; the sign is
used for x locations behind the interface (jc < tVpurgf) where Since the interface is located at a distance down the pipe
the concentration is greater than 0.5, and the + sign is used equal to t V ^ ^ , the ratio of the combustible length given by
for locations ahead of the interface (x > r V ^ , ) . Thus, the Eq. ( 6) to the total length o f pipe occupied by purging gas at
location of any specific value of purging gas concentration any time t is therefore proportional to 1N t. Figure 5 shows
travels down the pipe according to: how well this prediction agrees with experimental data ac
quired for a 22.6 inch diameter pipe (Marshall, et al. (1992)).
The plots give the ratio o f the flammable length to the dis
tance from the inlet to the interface location, for various
x = tVpurge constant 7 5 7 (5)
purging velocities as a function o f downstream location of the
interface. The plot indicated by the heavy line is a prediction
of this ratio (which, as stated above is proportional to 1N t)
For example, a location where the concentration of the purg for a velocity of 472 ft/min, with the proportionality constant
ing fluid is 0.9 lb/lb-mixture requires the argument of erfc to chosen to match the experiment at a point halfway along the
be = - 0.905; thus constant = - 2 X 0.905 = - 1.91; and pipe. As can be seen the prediction agrees with the experi
similarly, for a location where the purging gas concentration ment fairly well. (The prediction is m ost in error for locations
is 0.1 lb/lb-mixture, constant = 1.91. near the inlet. The discrepancy is at least partly due to the
fact that the experimental data points correspond to the length
If >41is the value of constant where the concentration is of mixture determined as the mixture passed by the measure
at the lower combustible limit and A2 the value where the ment location; the measurements thus represent mixed lengths
concentration is at the upper combustible limit, the length of over a time interval rather than at a specific discrete time, so
the combustible region included between these two limits the error is most evident near the inlet where the time interval
is found by using Eq. (5) twice: is the greatest fraction o f the total time.)

The experimental data also show that the flammable


length increases at a rate considerably greater than Vr for the
Lmix=X2 - *1 = (tVpurge + A2 y f f y ) ~ smallest velocity o f 197 ft/min (the light solid curve). This
result is not surprising since the minimum velocity required
[ ( 'W - a V * v ) ] = iA2 + A

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


D^ = o . ( 1- 3 0 x ) +450l''"fiV> (8)
Nh = number o f bends

As an example, for a 250 feet pipe with d - 24 in, the


minimum Vpurgc is 250 ft/min. Hence, the interface reaches
the exit 60 seconds after purging starts. The turbulent diff
usion coefficient (Fig. 6) is about 1.75 ft2/sec for this d and
Vpurgt A\ + A2 = 2.5 to 3.0 for flammable concentrations, so
from Eq. ( 6) the length of the flammable region when the
interface reaches the exit is 26 to 31 feet. The flammable
length thus occupies 10% to 15% o f the pipe length, which
agrees fairly well with the data shown in Fig. 5.

FIELD TESTING METHODOLOGY & INSTRUMENTATION


Field tests were vital for obtaining a better understanding
o f the flow processes in a pipeline purge. Results obtained
from this data are used to verify and improve upon a com
puter model simulation which is under development. As will
be shown, the particular field tests conducted are useful for
confirming information on concentration profiles (Eq. 3).
However, all field observations to date were for velocity
Figure 6. Turbulent dispersion coefficient for conditions well above the stratification velocity and thus no
straight pipes. experimental data has been obtained from this study for an
accurate determination of the coefficient used in Eq. 2.
from Eq (2) to prevent stratification is 2S2 ft/min for a pipe of
this diameter. Since the fluids are probably stratified for a Field tests involve collecting pressure, velocity, and gas
velocity o f 197 ft/min, the mixing occurs over a much greater, concentration data from three locations along the pipeline
interface length than ju st the pipe diameter, and Eq. ( 6) would being purged. At the upstream inlet, a pressure gage is used
not be applicable. to acquire an accurate record o f the inlet pressure throughout
the purge. A t an intermediate location down the pipeline, a
To use Eq. ( 6), for computing the length of the flamma probe is inserted into the pipe through an existing tap (pipe
ble region, the upper and lower combustible limits must be line nozzle or standoff pipe with a flange) with a block valve
known so that A\ and A2 can be computed, and De must also attached to the tap. A blind flange, modified with a 1
be known as a function o f pipe diameter, purging velocity, Swagelok fitting, is attached to the block valve and the probe
viscosity, pipe roughness, etc. Figure 6 can be used to de is inserted through the fitting into the pipeline. The Swagelok
termine De. For CH 4, air, and N2, the Schmidt Number is ap fitting is tightened and the sample lines are attached to the
proximately one, and for Pe = 1, the data shown in Fig. 6 can probe. The sample lines are fed into a instrumentation pack
be correlated by the expression: age, which reads static and dynamic pressure, temperature,
and concentrations o f oxygen and natural gas from two verti
cal locations inside the pipe. The gas concentrations are
measured by two GMI Gasurveyor 11B sensors and the data
io^ . / w ,- .0 3 ) r l8( Re -1 0 0 0 Y
from these sensors are logged to a laptop computer. The gas
0.1 1000 J sensor system which normally operates below 1 psig was
(7)
modified to operate at purging pressures up to approximately
e x p [-(lo g fle )u ] Re > 2 0 0 0 100 psig. A t the blowdown vent, exit velocity and concentra
tion measurements are recorded by a pitot tube and a third gas
sensor. The concentration vs. time profile at the blowoff is
To be conservative, the value o f De should be computed using used to determine the amount o f slug shortening or growth of
the gas with the larger viscosity, since this will give the larg the mixed gas region.
est De. Bends in the pipe increase the turbulent mixing; their
effect can be accounted for approximately by B ischoffs HELD STUDY
(1964) method to give an effective Deeff: Table 1 and 2 are a summary o f the pipeline purges that
have been surveyed. A unique constraint on the field studies

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 1. Summary of pipeline geometry from surveyed purges.

Pipe Pipe Probe Blowoff Blowoff Blowoff Valve


Diameter Length Location From Diameter Length Diameter
# Date an) (Ft) Inlet (Ft) an) (Ft) an)
1 5/19/95 30 74659 N/A 10-6 55
2 7/13/95 20 43296 N/A 8 10 8
3 7/19/95 8 6800 N/A 2 10 2
4 8/8/95 10 26928 N/A 4 6 -

5 9/29/95 30 87806 4171 10 8 8


6 11/11/95 26 17200 13705 6 25 -

7 12/19/95 30 399168 N/A 8 7 6.8


8 12/20/95 30 284592 N/A 8 7 6.8

Table 2. Summary of data obtained from pipeline purging observations.


Mixed Length Mixed Length
Inlet Pressure Average Velocity At Probe At Vent Exit
r (PSD (Ft/Min) (Ft) (Ft)
i 50 1821 N/A 2730
2 27 3522 N/A 1110
3 20 1344 N/A 232
4 21 4062 N/A 846
5 50 4182 1901 2955
6 25 2100 224 289
7 2022 N/A 1889
8 2238 N/A 1865

is that tests were planned around ongoing purging operations Aside from the average velocity and mixed gas length,
conducted by a number of gas companies. The field team was other significant information was gathered from the purge
not involved with the planning of these purges and they were observations. Figure 8 shows the inlet and exit pressures and
observers only; although, permission was obtained to use field exit velocity for Purge #8. Note, that there was a steady in
instrumentation. crease in each of these values throughout the entire purging
process. The exit pressure and velocity maintained a steady
The important information obtained from these purges is increase to a point where an abrupt change was observed,
average velocity and the length of the mixed region of natural which was caused by the change in properties of the passing
gas and air. The average velocity is determined using the fluid upon the arrival of the gas/air interface.
arrival time of the gas/air interface at the vent exit. The
mixed length of gas is calculated by taking the time for the Purge #8 is an example of a d irect purging operation
natural gas concentration to increase from 2% to 98%, and where air and natural gas were in direct contact The alterna
multiplying this time by the average velocity in the pipeline. tive to direct purging is to use a slug of inert gas to act as a
buffer between the air and natural gas, called slug purging.
Example calculations are provided from Purge #8. The Purge #6 is a typical example of a slug purging operation.
concentration profiles (Fig. 7) show that the gas/air interface Figure 9 shows the concentration profiles obtained by the
arrived at the vent exit in 2:07:08 (7628 seconds) after the probe from an intermediate location along the pipeline. An
start of the purge. Dividing this time into the total pipe length important point to note is that the oxygen and natural gas
resulted in an average velocity of 37.3 ft/sec (2238 ft/min). concentration lines do not intersect, contrary to the direct
The mixed gas/air region passed the vent exit in 50 seconds purge shown in Fig. 7. Consequently, no flammable mixture
(2% to 98% natural gas concentration). Using the average was present in the pipeline. An initial amount of nitrogen
velocity of 37.3 ft/sec, the mixed zone length at the pipeline (5472 ft3) was introduced into the pipeline and the amount of
exit was calculated to be 1,865 feet. slug shortening at the probe location was approximately 24%
(down to 4134 ft3 of pure, unmixed nitrogen). The mixed gas

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Vent Velocity (ftfeec)
2:05.00 2:00:00 3:07:00 2:00:00 2:00:00 2:10:00
00000 0* 1:30 0:43:12 1.04:48 140:24 1:4800 20900 201:12 202:48 3: 14*24 30000
71m*From8Url ol Purge(htammite) TimeFromStartol Purge(hh:mm:ee)
Figure 7. Concentration profiles of oxygen and natural gas for Purge Figure 8. Pressure and velocity curves taken at the vent exit during
#8 (Direct Purging). The two profiles Intersect, signifying an oxy- Purge #8. The abrupt change In the exit data Is evidence that the
gen/natural gas mixture within the pipeline. gas/alr Interface has passed.
Cj3

0 1 2 3 4 8 8 7 8 8 10
TVne(mln]

Figure 9. Concentration profiles of oxygen and natural gas for Purge #6 (Slug Purging).
The two profiles do not Intersect, due to the presence of an Inert nitrogen slug.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


Table 3. Example showing a portion of the Table 8-1 in
A.G.A/S Purging Principles and Practice
P urgin g D ata F o r In le t C on trol P ro ced u re
M inim um Inlet P ressu res Psig
_____________ (By Line Size)_____________

L ength O f 4 Blowoff Valve In le t 6 Blowoff Valve In let


Pipeline P re ssu re Psig Line Size P ressu re Psig L ine Size
(M iles) (Inches) (Inches)
6 8 10 12 12 16 18 20
1 8 8 11 18 6 11 14 21
2 12 11 13 20 8 12 15 22
3 16 13 15 21 10 13 16 23
4 19 16 17 22 11 14 17 23
5 22 18 18 23 13 15 18 24

Table 4. Example of a new pipeline purging table that included the gas flow rate,
average velocity, minimum purge pressure (to eliminate gas stratification), and
pressure that just causes choke flow in the blowoff vent pipe.
6 Blow off Valve
In let P re ssu re (Psig)
L ine Size (Inches)
L ength O f 12 P ipe 18 Pipe
Pipeline P ressu re Velocity Slug Vol. P ressure Velocity Slug Vol.
(M ile) (PSIG ) (Ft/Sec) (SCF) (PSIG ) (Ft/Sec) (SCF)
1 6 65.8 159 14 52.5 527
minimum 0.02 3.0 0.04 3.6
choked 43.6 93.8 31.2 53.9
2 8 58.2 240 15 50.4 760
minimum 0.03 3.0 0.05 3.6
choked 55.7 81.1 33.7 51.9

Figure 10. Pipeline geometry used in software for predicting purge time.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use


length for this situation is defined as the mixture o f natural lengths up to at least 20 miles to cover typical distance of
gas with nitrogen between 98% and 2% natural gas. purge in transmission piping.

CHANGES TO STANDARD PURGE TABLES 3. The mathematical model o f the S-shaped concentration
Guidelines for planning a pipeline purge are provided in profile is applicable for calculating slug shortening (inert
the A.G.A. Purging Manual. Table 8-1 in that document slug purge) as well as the length o f the flammable mixing
provides purge pressures based on the pipeline diameter and region when air and natural gas are mixed during a direct
length and the pipeline blow off valve diameter. The sug purging process.
gested purge pressures are intended to provide an average
purge velocity o f 2 minutes per mile (44 ft/sec). Table 3 4. Table 8.1 should be modified to show the minimum pres
shows a portion o f Table 8-1. sures needed to avoid stratification and perhaps a step-
by-step calculation procedure should be included as a
In addition to improving the accuracy o f Table 8-1, the supplement to the guideline.
research will result in more information for the revised purge
tables. Table 4 shows an example o f the type o f calculations 5. Software suitable for execution on a personal computer
that can be added to the new purging tables. This new table would be beneficial to those responsible for planning
includes the average velocity and volumetric flow rate that purge operations.
results from the recommended purge pressure when applied
to a typical piping geometry (inlet pipe, pipe to be purged, ACKNOWLEDGEMENTS
and the purge pressure that will cause choking in the blowoff The authors would like to acknowledge Dr. Frank Dodge
vent with or without a value). Two additional lines have been for his critical assessments o f the literature and for his many
added to indicate the minimum allowable purge velocity and technical contributions to the purging program.
the purge pressure that will cause choking in the blow off vent.
The minimum allowable purge pressure is based on the mini W e also want to acknowledge the respondents to the sur
mum velocity necessary to eliminate gas stratification. Pres vey, the gas companies that made the field testing possible,
sures higher than the minimum pressure will provide suffi and the A.G.A. Pipeline Committee.
cient driving force to bring about a turbulent flow necessary
for effective purging. REFERENCES
Bischoff, K.B., 1964, An Example o f the Use o f Com
SOFTWARE FOR PURGE PLANNING bined Models: Mixing in a Tabular Reactor with Return
Software is being developed to aid in planning pipeline Bends, A.LCh.E. Journal, Vol. 10, No. 4, pp. 584-598.
purges. The program calculates the purge time based on the
Cortelyou, J.T., and H.M. Curtis, 1964, Safe and Eco
users pipeline geometry and the purge pressure. A user en
nomical Purging Practices, A.G.A. Proceedings.
ters the information listed in Fig. 10 into the program to de
fine the pipeline geometry. The user also specifies the gases Marshall, M.R., Cleaver, R.P., and C.L. Hinsley, 1992,
used in the purge and the number and type of fittings in each Pipe Purging Operations, 1992 International Gas Research
section of the piping (inlet piping, pipeline, and blow off pip Conference
ing). The software then calculates the purge time, average
Henderson, E., 1941, Com bustible Gas M ixture in Pipe
velocity, and the gas flow rate for different purge pressures.
Lines, PCGA Proceedings, Vol. 32.
CONCLUSIONS AND RECOMMENDATIONS Linden, P.F., and J.E. Simpson, 1986, Gravity-driven
Based upon the work conducted to date, several conclu Flows in a Turbulent Fluid, Journal o f Fluid Mechanics, Vol.
sions have been drawn. 172.

1. By using the recommended control pressures indicated in Perkins, T.K., J.A. Euchner, 1988, Safe Purging of
Table 8.1 o f the current A.G.A. Purging Manual, the re Natural Gas Pipelines, SPE Production Engineering, pp.
sultant pipeline velocities will ensure that gas stratifica 663-668.
tion will not occur for the applicable piping configura W oodhead, J.R., White, E.T., and D. Yesberg, 1971,
tions (a straight pipe with inlet and outlet risers). How Prediction o f Effective Axial Dispersion in Transitional and
ever, the resultant velocities are generally higher than re Turbulent Flow, The Canadian Journal o f Chemical Engi
quired for efficient purging reasons and, in many cases, neering, Vol. 49, pp. 695-698.
the blow off vent will experience choked flow.

2. T h e range o f pipe sizes used in T able 8.1 should at least


include sizes up to 42 inches in diameter and piping

Downloaded From: https://proceedings.asmedigitalcollection.asme.org/ on 10/19/2017 Terms of Use: http://www.asme.org/about-asme/terms-of-use

S-ar putea să vă placă și