You are on page 1of 7

A Modern Gunbarrel of Unique Design

M.L. Powers, SPE, Consultant

Summary Even today, the gunbarrel is often the preferred means of crude-oil
Although regarded by many as obsolete, gunbarrels (or wash-tanks) dehydration in warm climates and/or where produced fluid tempera-
are still used for primary oil treating in many areas. This paper de- tures are high, or where other circumstances make heating unneces-
scribes a modern gunbarrel constructed from an existing 5,000-bbl sary or inexpensive. Many current gunbarrels are of larger diameter
tank. Oil treating without the addition of heat was feasible because and have lower Lv /dv ratios than early ones, and have improved inter-
nals. This paper describes a modern gunbarrel design that incorpo-
of the combination of relatively warm produced fluid and lenient ba-
rates effective residence time for the water-continuous phase, as well
sic sediment and water (BS&W) limits. The subject vessel provides
as improved oil-continuous phase distribution, gas separation, and
gas separation and contains two spreaders that were designed to pro-
solids removal.
vide good oil- and water-phase retention and to facilitate solids sepa-
ration and removal. The oil-phase spreader has a diameter equal to
Construction Site
78% of that of the vessel. It incorporates a unique deep skirt, having
a pattern of restrictive exit ports that imposes uniform radial oil-phase The gunbarrel described here was developed by modifying an exist-
flow over a wide range of rates and is relatively insensitive to minor ing ineffective 5,000-bbl (38.7 ft 24 ft) bolted, cone-bottom set-
misleveling, which is not the case for common serrated-skirt spread- tling tank. Before vessel modification, the BS&W and solids con-
ers. The design of the unique vessel internals permitted assembly tent of the effluent 29.3°-API oil was essentially the same as the
without welding at the battery site. Differences in the contribution influent. After the retrofit, the BS&W content of effluent oil was
of the water-bath zone of heated and nonheated gunbarrels are dis- easily maintained below 2%. The tank was originally equipped
cussed, and it is shown that nonheated-vessel designs that increase with a 3.0-ft-diameter internal flume and central “crow’s nest” oil
water-phase residence time and facilitate convection in the water-bath collector, which were retained in the modification. The flume ex-
tension functions as a vertical oil/gas separator, with gas capacity
are the most effective, a result of conservation of intrinsic
being governed by maximum-allowable superficial-velocity con-
well-stream heat. It is also demonstrated that the optimum oil-blanket
siderations. The gas-capacity formula shown below was extracted
thickness is a compromise between oil-residence time and oil temper-
from Ref. 1, and would be appropriate for calculating maximum
ature in nonheated vessels that capture water-phase heat. Vessel in- instantaneous gas-flow rates. Appreciable liquid carryover from
ternals that entrap oil beneath water (such as the oil-phase spreader the separator is undesirable because it would result in water and wet
of the subject vessel) are subjected to a buoyant force in addition to oil being dumped on top of clean oil leaving the vessel. The ap-
the weight of water displaced by steel. This effect is discussed, and propriate value of K for this equation is the largest one that does not
design equations are developed to calculate the net buoyant force ex- interfere with meeting pipeline-oil specifications.
erted upon a specific spreader and the gauge steel from which a
0.5
spreader must be constructed to preclude floating. An example is 67, 858 K d 2F pT sƪǒò o * ò gǓńò gƫ
included that illustrates the application of these equations to the sub- qg + . . . . . . . . . . . . . . . (1)
p sTz
ject vessel.
Details of the new gunbarrel design are illustrated in Fig. 2. This
Introduction design employs large-diameter oil-phase and water-phase distribut-
ing spreaders. These were fabricated from bolted 3,000-bbl tank-
The gunbarrel (or wash-tank) was devised for field processing at the
deck segments and rafters because this construction method per-
infancy of the oil-producing industry. If crude-oil dehydration re-
mitted assembly without welding at the tank-battery site. The rafters
quired increased temperature, energy was added by heating the water
of both spreaders extend from the tank wall to the flume, where they
bath by means of an internal firetube, internal steam coils, or external are bolted to attachment rings welded to the flume. The flume was
thermosiphon loop and direct heater. An alternative method was pre- removed from the battery site for installation of these rings. The up-
heating the influent fluid. By either process, these vessels were con- per-spreader rafters were supported at the outer end by attachment to
siderably less energy efficient than modern heat-treaters. However, rolled-channel steel, which was bolted to the tank wall. The lower-
during the era of the heated gunbarrel, low-pressure gas had little or no spreader rafters were also attached to the tank wall. However, vertical
value. Typically, gunbarrels would have the same diameter as battery loading was supported by legs extending to the tank bottom. A vent
stock tanks, but would be somewhat taller to assure gravitational flow. pipe from near the apex of the lower spreader extends up into the
Thus, Lv /dv ratios generally exceeded 1.0. A generic gunbarrel is upper spreader, and a second one extends from near the apex of the
equipped with a gas separation/fluid inlet device, such as the internally upper spreader up into the flume dome. The upper spreader has the
installed “flume” (or gas boot), illustrated in Fig. 1a, or the external one normal 1:12 tank-deck pitch and a diameter of 30.0 ft. The lower
shown in Fig. 1b. With either configuration, the fluid stream is nor- spreader was constructed with an 18.8° slope and is 28.5 ft in diame-
mally discharged beneath a serrated spreader having a diameter be- ter. This was accomplished by using only 19 of the standard 20 tank-
tween one-fourth and one-half that of the tank, which provides a mea- deck segments. This increased slope, in conjunction with a jetting
sure of flow distribution of the oil-continuous phase. Normally, the system, prevents sand accumulation on top of the lower spreader.
water-continuous phase is free to short-circuit directly to the water out- A steel plate seals the bottom of the flume, which is supported by an
let, minimizing energy consumption in the case of a heated vessel. angle iron framework from the tank bottom. Incoming fluid exits
However, it results in the effluent water having approximately the same the flume through 16 equally spaced 2-in. round inlet ports located ra-
oil content as the influent water. The oil outlet is normally a pipe cou- dially around the flume. The bottoms of these holes are at the depth
pling installed in the side of the vessel through which oil overflows, of the bottom of the upper (oil-phase) spreader skirt.
maintaining a constant level. An internally or externally installed wa-
ter siphon controls the oil/water interface. Oil-Flow Regulation. Radial oil-phase flow is imposed within the
oil-phase spreader by outflow regulation, using 9/16-in.-diameter re-
Copyright 1996 Society of Petroleum Engineers strictive exit ports in the spreader skirt. Port flow rate is a function
of the interface depression (D), illustrated in Fig. 3, and may be cal-
Original SPE manuscript received for review Oct. 10, 1994. Revised manuscript received
July 19, 1995. Paper peer approved Aug. 15, 1995. Paper (SPE 28538) first presented at the
culated from Eq. 2. This equation was derived from Eq. B-4 by con-
1994 SPE Annual Technical Conference & Exhibition, New Orleans, LA, Sept. 25–28. verting rate to barrels per day.

54 SPE Production & Facilities, February 1996


Fig. 1—Typical gunbarrel configurations.

0.5 at the oil/water interface, which would effectively be a vertical projec-


q h + 116.6 d 2h ƪD (g w * g o)ńg oƫ . . . . . . . . . . . . . . . . . . . . (2)
tion of the spreader skirt. The water bath is very beneficial in the case
The 12-in. spreader skirt has three horizontal rows of ports, with 60 of a heated gunbarrel, such as the one described in Ref. 2, because it
equally spaced ports per row, located 2-, 4-, and 6 in. from the skirt serves as the heat-transfer medium. However, in the subject unheated
top (see Fig. 4). No fluid flows through a row of ports until the inter- gunbarrel, the principal valve of the water-bath is to conserve
nal interface is depressed to a lower depth because there is no differ- well-stream heat, as discussed in a later section. From the oil/water-in-
ential pressure. Fig. 5 is a plot of flow rate exiting the spreader vs. terface impact zone, oil flow would be inward and upward to the oil
interface depth (Z), measured from the skirt top. This figure is based collector in an undefined, generally radial flow pattern. Thus, there
on Eq. 2, and the average tank-battery oil gravity (29.3° API) and wa- are, in effect, two stages of gravity separation. We show in Appendix
ter specific gravity (1.02). Curves are presented for each row of ports A that separation capacity is proportional to horizontal cross-sectional
and for the sum of all rows. It can be seen that good rate regulation area (AH ) regardless of the direction of bulk flow. Thus, effective sepa-
(and thus radial flow) is imposed from no flow when Zt2 in., up to ration is accomplished by efficient use of vessel horizontal area and not
7,452 B/D when Z+12 in., at which point oil spillover at the skirt bot- directly by efficient use of vessel volume. Consequently, separation
tom is impending. Since qh is a square root function of D, slight mis- effectiveness will not necessarily track with observed residence-time
leveling of the spreader would have minimal effect on radial-flow dis- data when comparing dissimilar vessels because the latter is a measure
tribution up to the point of spillover. Spillover would likely occur of volumetric displacement. The functional relationship developed in
along a small arc of the spreader skirt unless it was precisely level. Appendix A between separation capacity and AH assumed plug flow
Field circumstances dictated the wide range of controlled flow de- and that AH does not vary with depth. Real flow differs from plug flow
signed into the oil-phase spreader, which would be impossible to because of short circuiting and turbulence, which impede separation.
The greater the departure from plug flow, the greater will be the de-
achieve with a conventional serrated spreader even if precisely level.
crease in separation effectiveness. It is reasonable to assume that the
controlled radial (horizontal) flow within the oil-phase spreader more
Oil Dehydration and Desanding closely approximates plug flow than the undefined flow regime in the
After the radial-flow regime within the oil-phase spreader, oil globules oil blanket. Therefore, any suspended contaminant of constant partial
exit the 9/16-in. ports in the spreader skirt and rise vertically through the size that is not removed from the oil while within the spreader would
water bath at a velocity of approximately 0.75 ft/sec to the impact zone not settle out of the oil phase if introduced at the top of the oil blanket.

Fig. 2—Cross-sectional view of vessel internals.

SPE Production & Facilities, February 1996 55


Fig. 3—Interface depression within oil spreader. Fig. 4—Oil-spreader skirt detail.

However, oil from the spreader enters the oil blanket at the bottom. q MW + 2.1603 10 6ǒd 2sw * d 2FǓǒŤg p * g wŤǓd 2pńm w, . . . . . . . . (5)
Therefore, the oil blanket may provide a second stage of separation
of indeterminate effectiveness for particles not removed from the oil 0.5
while within the spreader. The greatest contribution of the oil blan- and d p + 6.8037 10 –4ƪq MW m Wńǒd 2sw * d 2FǓǒŤg p * g wŤǓƫ . . (6)
ket, however, is providing residence time for coalescence of small
water droplets into larger, separable ones. Use of these equations for overall vessel water capacity would be
conservative because they do not account for the separation occur-
Spreader-Design Equations. The oil-phase spreader separation- ring beneath the lower spreader, or from water driven by convection
capacity equation (Eq. 3) was derived from Eq. C-2 by converting into the area above the oil-phase spreader. The flow regime beneath
rate to barrels per day. Eq. 4, the particle-design-diameter equation, the lower spreader will assume some effective thickness. Oil drop-
is a rearrangement of Eq. 3. lets smaller than the design diameter (defined by Eq. 6) may be re-
moved from water near the top of this flow stream, but not deep in
q MO + 2.1603 10 6ǒd 2s * d 2FǓǒg p * g oǓd 2pńm o , . . . . . . . . . (3) the flow stream. The reverse would be true of solid particles. It
is recommended that Eq. 5 be used to estimate vessel-water capac-
0.5 ity, realizing that the results will be somewhat conservative.
and d p + 6.8037 10 –4ƪq MO m ońǒd 2s * d 2FǓǒg p * g oǓƫ . . . . (4)
Oil Blanket
These related equations are useful for estimating separation capacity
(or design-particle diameter) for removal of either water or solids. Be- We previously mentioned that the primary contribution of the oil
cause they describe only spreader performance, they are conservative blanket was providing residence time for coalescence of small water
when used to calculate overall vessel performance because they do droplets into larger, separable ones. Therefore, it might seem that
not include the contribution of coalescence and additional separation a very low oil/water interface (thus a large oil-blanket volume and
occurring in the oil blanket, which cannot be expressed mathematically. small water-bath volume) would be desirable. However, oil-phase
retention time is not the only factor to be considered. Maximizing
oil temperature also aids coalescence. A significant water-bath
WaterĆContinuousĆPhase Retention
volume will enhance the previously mentioned convection process.
As previously mentioned, common gunbarrels allow the water con- The only energy available to maintain oil temperature above ambi-
tinuous phase to short-circuit from the point of liquid entry to the ent is that intrinsic to the well-stream. At any specific fluid temper-
water outlet, resulting in high-oil and -solids content of the effluent ature, vessel heat loss to the atmosphere can be estimated with Eq.
water. The lower spreader of the subject vessel prevents this. The 7, which was adapted from an equation presented in Ref. 3. Recom-
apparent water-phase flow path of the subject design is radially out- mended values of k are presented in Table 1.
ward from the flume entry ports to the perimeter of the lower (water-
phase) spreader, then radially inward to the water outlet. However, Q + k p d v L v ǒT L * T AǓ. . . . . . . . . . . . . . . . . . . . . . . . . . . (7)
convection will drive some of the warm incoming flow into the wa-
ter-bath above the oil-phase spreader, displacing cooler water from
that area and conserving intrinsic well-stream heat. The low fluid
velocity at the spreader perimeter would provide only moderate re-
sistance to this convective current. (In contrast to the subject de-
sign, the gunbarrel designs shown in Fig. 1 would result in wasting
most of the heat contained in the water phase.) From the outlet, wa-
ter enters an external siphon that controls the oil/water interface.
In effect, the spreader area is used twice for oil and solids removal
from the water phase. In the initial stage of separation, oil droplets
float up into the upper (oil-phase) spreader and solids settle to the
conical surface of the lower spreader. Once the water phase flows
around the perimeter of the lower spreader, separated oil droplets
would float up into the apex of the lower spreader and then through
the vent pipe into the upper spreader. Solids removed in this stage
settle to the tank bottom. The water-phase-capacity equation (Eq.
5) was derived by inspection from Eq. 3. Eq. 6 is a rearrangement
of Eq. 5 and defines the design-particle diameter as a function of
qMW. The absolute value signs within the specific-gravity term are
necessary because this term would otherwise be negative in the case
of oil separation. Fig. 5—Effect of interface depression on spreader-flow capactiy.

56 SPE Production & Facilities, February 1996


TABLE 1—VALUES OF HEAT LOSS CONSTANT (k) TABLE 2—SHEET STEEL DATA
Wind Velocity k—Bare Steel Thickness Wu
(mph) (BTU/hr ft2 °F) Gauge No. (in.) (lbm/ft2)
0 2.96 00 11/32 13.750
5 3.12 0 5/16 12.500
10 4.20 1 9/32 11.250
20 5.00 2 17/64 10.625
3 1/4 10.000
4 15/64 9.375
The magnitude of liquid-temperature distribution within a vessel
would be very small and the average temperature would be applica- 5 7/32 8.750
ble to Eq. 7, regardless of the relative volumes occupied by oil and 6 13/64 8.125
water. Heat will flow from the water into the oil blanket, and facili- 7 3/16 7.500
tating the convection process of diverting influent water into the wa- 8 11/64 6.875
ter-bath by raising the oil/water interface will increase oil-blanket 9 5/32 6.250
temperature, illustrated in the following example. 10 9/64 5.625
11 1/8 5.000
Heat-Loss Example. It is assumed that the subject gunbarrel is 12 7/64 4.375
uninsulated and that the inflow is 2,000 BO and 8,000 BWPD at a
temperature of 100°F. Ambient air temperature is 40°F and wind force on the spreader would be neutral when the oil/water interface
velocity is 10 mph (thus k+4.20 Btu/hr ft2 °F). In the first case, was located 8.1235 in. from the top of the skirt (Z+8.1235 in.) using
it is assumed that all produced water short-circuits the vessel. Thus, Eq. D-10. With the interface at this depth, all three rows of exit
the oil provides all of the atmospheric heat loss. Assuming an oil ports would be in operation, and a resultant oil-flow rate of 5,264
specific-heat capacity of 0.5 Btu/lbm°F, a 70.7°F average tank-liq- B/D may be calculated with Eq. 2 or observed from Fig. 5. Thus,
uid temperature is calculated with Eq. 7. In the second case, it is the net force on the spreader will be downward so long as the oil-
assumed that all of the water is diverted into the water-bath and that flow rate does not exceed this value.
water leaving the vessel is at the average tank-liquid temperature.
The resulting liquid temperature would be 94.8°F. Oil processed Conclusions
at 94.8°F would likely meet pipeline specifications and that pro-
1. Insufficient gas separation capacity could result in water and
cessed at 70.7°F would not. Thus, it is desirable to capture as much
wet oil being dumped on top of clean oil leaving the vessel.
heat as possible from the water by inducing convection in the wa-
2. An improved oil-phase spreader has been developed that pro-
ter-bath. An additional benefit of maintaining a high oil/water in- vides uniform radial distribution over a wide range of flow rates and
terface is that oil temperature is less affected by short duration de- is relatively insensitive to minor misleveling.
creases in ambient air temperature because more heat is stored. The 3. Effective separation is accomplished by efficient use of vessel
foregoing discussion demonstrates that the optimum oil-blanket horizontal area, and not directly by efficient use of vessel volume.
thickness is a compromise between oil-residence time and oil tem- Consequently, separation effectiveness will not necessarily track
perature in nonheated vessels that capture water-phase heat. with observed residence-time data when comparing dissimilar ves-
sels because the latter is a measure of volumetric displacement.
Buoyancy Considerations 4. The value of the water-bath in a nonheated gunbarrel is to con-
The subject vessel was initially water filled, and the lower spreader is serve intrinsic well-stream heat, whereas for a heated vessel it is the
properly vented and always totally submerged in water. Consequently, heat-transfer medium.
the buoyant force applied to the lower spreader will equal the weight 5. The oil blanket may provide a second stage of separation of in-
of the water displaced by steel (Wgw /gs ). This amounts to approxi- determinate effectiveness for particles (water or solids) not removed
mately 13.26% of spreader weight for 1.02 specific gravity brine. The from the oil while within the spreader.
upper spreader is also properly vented and submerged in water but en- 6. The greatest contribution of the oil blanket is providing resi-
traps oil. Therefore, it is subjected to a buoyant force in addition to dence time for coalescence of small water droplets into larger, sepa-
Wgw /gs , which is a function of the depth of the oil/water interface within rable ones.
the spreader and consequently a function of oil-flow rate. Eqs. D-8 7. Raising the oil/water interface will increase oil-blanket tem-
through D-10 are expressions for the buoyed weight of the top spreader perature in nonheated gunbarrels designed to conserve water-phase
(including the oil entrapment effect), the weight per square foot of sheet heat.
steel required to preclude an unrestrained spreader from floating in 8. Oil-blanket temperature is less affected by short duration de-
the event it becomes oil filled, and the oil/water-interface depth at the creases in ambient air temperature when a high oil/water interface
neutral point (if one exists), respectively. is maintained because more heat is stored.
The following example demonstrates how these equations were 9. Optimum oil-blanket thickness is a compromise between oil-
used in the design of the subject gunbarrel, and that buoyancy residence time and oil temperature in nonheated vessels that capture
should always be considered in the design of vessel internals that water-phase heat.
10. Buoyancy should always be considered in the design of vessel
can entrap oil beneath water. First, Eq. D-9 was used to determine
internals that can entrap oil beneath the oil/water interface.
the gauge of sheet steel required to prevent the spreader from float-
ing should it become completely oil-filled. Substituting values of
1.25-, 1.0-, 15.0-, and 1.5 ft for hc , hs , rs , and rF, respectively, re- Nomenclature
sulted in Wu +12.855 lbm/ft2. Table 2 (extracted from Ref. 4) AH +horizontal area, L2, ft2
shows that 00 gauge (11/32-in.) steel would be needed to meet this Ah +spreader skirt port area, L2, ft2
criteria. Sheet steel of this thickness would be difficult to get and As +oil-phase spreader surface area, L2, ft2
work with. For these practical reasons, it was decided to use ¼-in. C +discharge coefficient, dimensionless
steel weighing 10.00 lbm/ft2. Eq. D-8 was then used to determine dF +flume diameter, L, ft
the buoyed spreader weight under oil-filled conditions by setting dh +diameter of exit ports in spreader skirt, L, in.
Z+12 in. This calculation resulted in a net upward force of 1,974 dp +design diameter of oil, water, or solids particle, L,
lbf. Consequently, it was necessary to bolt the spreader to the raf- cm
ters (which were secured to the tank wall and flume) to preclude the ds +oil-phase spreader diameter, L, ft
possibility of spreader floating. It was determined that the resultant dv +vessel diameter, L, ft

SPE Production & Facilities, February 1996 57


Fig. 6— Effect of (a) radial flow and (b) vertical flow direction on separation.

dsw +water-phase spreader diameter, L, ft z +supercompressability factor, dimensionless


D +distance from an oil-phase-spreader port down to Z +distance from oil-phase-spreader skirt top to
the internal oil/water interface, L, in. internal oil/water interface, L, in.
Fbc +buoyant force resulting from water being displaced Zn +value of Z resulting in neutral force, L, in.
by oil in the conical portion of the oil-phase go +oil specific gravity, dimensionless
spreader, mL/t2, lbf gp +particle specific gravity, dimensionless
Fbs +buoyant force resulting from oil accumulation gs +steel specific gravity, dimensionless
within the oil-phase-spreader skirt, mL/t2, lbf gw +produced water specific gravity, dimensionless
g +acceleration of gravity, L/t2, 32.174 ft/sec2 mo +dynamic viscosity of oil continuous phase, m/Lt,
Dh +differential head, L, ft cp
hc +altitude of cone frustum portion of oil-phase mw +dynamic viscosity of water continuous phase,
spreader, L, ft m/Lt, cp
hs +height of oil-phase-spreader skirt, L, ft òg +gas density at separation conditions, m/L3, lbm/ft3
k +constant from heat-loss equation (Eq. 7), m/t3T, òo +oil density at separation conditions, m/L3, lbm/ft3
BTU/hr ft2 °F
K +separator performance constant, L/t, ft/sec Acknowledgment
L +height, L, ft
I thank Conoco for the opportunity to work on this project and Larry
Lv +vessel height, L, ft
Gaertner for his input and construction supervision. The contribu-
p +absolute operating pressure, m/Lt2, psia tion of Bob Adam with Tank Tec Inc. of working out various fabrica-
ps +absolute standard pressure, m/Lt2, psia tion details is gratefully acknowledged.
q +flow rate, L3/t, ft3/sec
qg +gas capacity, scf/D
References
qh +port flow rate, L3/t, B/D
qmo +oil capacity, L3/t, ft3/sec 1. Powers, M.L.: “New Perspective on Oil and Gas Separator Performance,”
qMO +oil capacity, L3/t, B/D SPEPF (May 1993) 77.
2. Williams, A.R.: “A Wash-Tank Design,” API, Drill & Prod. Prac. (1953)
qMW +water capacity, L3/t, B/D
272.
qo +oil-flow rate, L3/t, B/D 3. Spec. 12L, Vertical and Horizontal Emulsion Treaters, third edition, API,
Q +heat-loss rate, mL2/t3, BTU/Hr Washington, DC (1986) 20.
rF +flume radius, L, ft 4. Engineering Handbook of Conversion Factors, National Tank Co., Tulsa,
rs +oil-phase spreader radius, L, ft OK (1991) 156.
tr +residence time, t, sec 5. Powers, M.L.: “Analysis of Gravity Separation in Freewater Knockouts,”
ts +particle settling time, t, sec SPEPE (Feb. 1990) 52–58; Trans., AIME, 289.
T +absolute gas temperature, T, oR 6. Vennard, J.K.: Elementary Fluid Mechanics, third edition, John Wiley
TA +ambient air temperature, T, oF & Sons, Inc., New York City (1954) 304.
TL +liquid temperature, T, oF
Ts +absolute standard temperature, T, oR Appendix AĊEffect of Flow Direction on Separation
vb +bulk-flow velocity, L/t, ft/sec The following demonstrates that gravity separation capacity is inde-
vs +average fluid velocity at the oil-phase skirt, L/t, pendent of the direction of bulk flow in vessels for which AH does
ft/sec not vary with depth. Plug flow is assumed in all cases, and all velo-
vt +terminal velocity of a settling particle, L/t, ft/sec cities (regardless of direction) are considered positive.
Vc +volume of cone frustum portion of oil-phase
spreader, L3, ft3 Radial Flow. Fig. 6a depicts the radial (horizontal) flow regime
W +oil-phase spreader weight in air, mL/t2, lbf within the oil-phase spreader. For complete separation, the water
Wb +resultant buoyed weight of oil-phase spreader, droplet (or sand grain) shown at the top of the interior cylinder (rep-
mL/t2, lbf resenting the flume) must reach the base of the exterior cylinder
Wbw +buoyed weight of oil-phase spreader immersed in (representing the intersection of the oil/water interface and the
water, mL/t2, lbf spreader skirt), before the bulk flow exits through the spreader-skirt
Wu +unit mass of sheet steel, m/L2, lbm/ft2 ports. At capacity, the particle would take the path shown. The
Wun +value of Wu required to preclude an unrestrained, time required for the particle to settle to the interface is presented in
oil-filled spreader from floating, m/L2, lbm/ft2 Eq. A-1.

58 SPE Production & Facilities, February 1996


t s + Lńv t. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-1)
A s + 2pr s h s ) p(r s ) r F)ƪh 2c ) (r s * r F) ƫ
0.5
2
. . . . . . . . (D-1)
Flow rate may be expressed as Eq. A-2.
The equation for spreader weight in air (Eq. D-2) was derived by
q + p d s L v s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-2) multiplying the surface area (expressed in Eq. D-1) by Wu , the unit
weight of sheet steel in lbm/ft2. Table 1 presents the thickness and
Residence time equals volume divided by rate (Eq. A-3).
Wu for tank steel of various gauge numbers.
p ǒd 2s * d 2FǓ L
tr +
4p d s L v s
+
d 2s * d 2F
4 ds vs
. . . . . . . . . . . . . . . . . . . . (A-3) NJ
W + W u 2pr s h s ) p(r s ) r F)ƪh 2c ) (r s * r F)
2
ƫ
0.5
Nj. . . (D-2)
At capacity ts +tr. Equating Eqs. A-1 and A-3 yields Eq. A-4. Sub-
The buoyed weight of the spreader immersed in water is given by
stituting this expression for vs into Eq. A-2 yields the separation ca-
Eq. D-3, which follows from Eq. D-2.
pacity equation (Eq. A-5).

vs +
ǒd 2s * d 2FǓ
v t, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (A-4)
W bw + W u ǒ1–g wńg sǓ NJ 2pr h ) p(r ) r )
s s s F
4d s L

and q mo + pǒd 2s * d 2FǓv tń4 + A Hv t. . . . . . . . . . . . . . . . . . . (A-5) ƪh2c ) (r s * rF)2ƫ


0.5
Nj. . . . . . . . . . . . . . . . . . . . . . (D-3)

Ref. 5 shows that Eq. A-5 is also valid for linear horizontal flow. The volume of liquid contained within the oil-phase spreader is
analyzed as two components, (1) a frustum of a cone minus a small
Vertical Flow. Fig. 6b depicts vertical oil-phase flow in a separation diameter cylinder (which will always be oil filled), and (2) a spool
vessel. Separation can occur only if the water droplet (or sand grain) defined by the spreader skirt and flume (which will contain an o/w
shown within the cylinder has a terminal velocity exceeding the up- interface, the level of which being determined by qo ). The volume
ward bulk-flow velocity. Thus, at capacity, vb +vt . By inspection, of the first component is expressed in Eq. D-4.
q+AH vb . Therefore, qmo +AH vt . This expression is identical to Eq. V c + ǒ ph cń3 Ǔǒr 2s ) r s r F * 2r 2FǓ. . . . . . . . . . . . . . . . . . (D-4)
A-5, the capacity formula for horizontal flow. The foregoing illus-
trates that gravity separation capacity equals the product of horizontal The buoyant force resulting from water being displaced from the
cross-sectional area and particle terminal velocity in a plug-flow re- conical component by oil is shown in Eq. D-5, which follows from
gime, and is independent of bulk-flow direction. Eq. D-4.
F bc + 62.37(g w * g o)ǒ ph cń3 Ǔǒr 2s ) r s r F * 2r 2FǓ . . . . . . (D-5)
Appendix B Ċ Derivation of PortĆFlowĆRate Equation
Flow rate through the oil-phase-spreader skirt ports would obey the Eq. D-6 describes the buoyant force resulting from oil accumula-
common orifice flow equation (Eq. B-1), which was extracted from tion within the spreader skirt.
Ref. 6. F bs + 62.37p Z ǒr 2s * r 2FǓ(g w * g o)ń12. . . . . . . . . . . . . . (D-6)
0.5
q + CA h(2gDh) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (B-1) The resultant buoyed weight of the oil-phase spreader follows.
In this equation, Dh pertains to the flowing fluid (oil phase), and W b + W bw * F bc * F bs. . . . . . . . . . . . . . . . . . . . . . . . . . (D-7)
is developed by the interface depression, D, illustrated in Fig. 3.
The equation for Dh (Eq. B-2) is obvious from inspection of Fig. 3. Eq. D-8 was derived by substituting Eqs. D-3, D-5, and D-6 into
D-7.
Dh + D(g w * g o)ń12g o. . . . . . . . . . . . . . . . . . . . . . . . . . (B-2)
Eq. B-3 was derived by substituting Eq. B-2 into Eq. B-1. NJ
W b + W u ǒ1 * g wńg sǓp 2r s h s ) [r s ) r F]ƪh 2c ) (r s * r F)
2
ƫ
0.5
Nj
q + CA hƪgD(g w * g o)ń6g oƫ .
0.5
. . . . . . . . . . . . . . . . . . (B-3) –62.37(g w * g o)pƪǒ h cń3 Ǔǒr 2s ) r s r F * 2r 2FǓ ) Zǒr 2s * r 2FǓń12ƫ
. . . . . . . . . . . . . . . . . . . . . (D-8)
Eq. B-4 is the result of assigning values of 0.6 and 32.174 ft/sec2
to C and g, respectively, and substituting p dh 2/576 for Ah in Eq. B-3. In the event Wb becomes negative, the spreader will float unless
it is attached to the tank. The maximum buoyant effect occurs at
0.5
q + 7.578 10 –3d 2h ƪD(g w * g o)ńg oƫ . . . . . . . . . . . . . (B-4) high qo , when the spreader is oil filled and spillover is impending.
Eq. D-8 may be used to calculate Wb under this condition by setting
Z=12 hs .
Appendix CĊRelationship of OilĆPhaseĆSpreader Eq. D-9 was derived from Eq. D-8 by setting Wb +0 and Z=12 hs ,
Capacity and Particle Size and solving for Wu . The result is Wun .
Eq. C-1 (adapted from Eq. 1 of Ref. 5) expresses Stoke’s law for a
settling particle in oil. 62.37(g w * g o)ƪǒ h cń3 Ǔǒr 2s ) r s r F * 2r 2FǓ ) h sǒr 2s * r 2FǓƫ
W un +
v t + 178.74ǒg p * g oǓd 2pńm o. . . . . . . . . . . . . . . . . . . . . . (C-1) NJ
ǒ1 * g wńg sǓ 2r s h s ) [r s ) r F]ƪh 2c ) (r s * r F) 2ƫ
0.5
Nj
Substituting Eq. C-1 into Eq. A-5 reduces to the following expres-
. . . . . . . . . . . . . . . . . . . . . (D-9)
sion for qmo as a function of dp .
If application of Eq. D-8 indicates that a net upward force could
q mo + 140.38ǒd 2s * d 2FǓǒg p * g oǓd 2pńm o. . . . . . . . . . . . . . (C-2) occur, the value of Z resulting in a neutral force may be computed
with Eq. D-10. This equation was derived from Eq. D-8 by setting
Wb +0 and solving for Z.
Appendix DĊDerivation of Equations Relevant to
Buoyancy
The surface area of the oil-phase spreader consists of a cylinder and NJ
W u ǒ1 * g wńg sǓ 2r s h s ) [r s ) r F]ƪh 2c ) (r s * r F) ƫ
2
0.5
Nj
a frustum of a cone. Using well-known formulas, this area can be Zn +
expressed as follows. 5.1975(g w * g o)ǒr 2s * r 2FǓ

SPE Production & Facilities, February 1996 59


4h cǒr 2s ) r s r F * 2r 2FǓ
* . . . . . . . . . . . . . . . . . . . . . . . . (D-10)
r 2s * r 2F

SI Metric Conversion Factors


°API 141.5/(131.5)°API)+g/cm3
Btu 1.055 056 E)00 +kJ
cp 1.0* E*03 +Pa@s
ft 3.048* E*01 +m
ft2 9.290 304* E*02 +m2
ft3 2.831 685 E*02 +m3
°F (°F*32)/1.8 +°C
in. 2.54* E)00 +cm
lbf 4.448 222 E)00 +N
lbm 4.535 924 E*01 +kg
mile 1.609 344* E)00 +km
psi 6.894 757 E)00 +kPa
°R 5/9 +°K
*Conversion factor is exact. SPEPF

Maston L. Powers is a consultant in Oklahoma City. His interests


include waterflooding, EOR, artificial lift, and production faciliĆ
ties. Powers served as Program Chairman for four Production
Operations symposia, was on the Facilities Engineering CommitĆ
tee, received a 1993 Central Plains Region Service Award and
the 1995 SPE Production Engineering Award, and is currently a
SPEPF review chairman. Powers holds BS and MS degrees in
petroleum engineering from the U. of Oklahoma.

60 SPE Production & Facilities, February 1996

You might also like