Retention-Time
Theories for Sizing Oil/Water
Separator
Kenneth E. Arnold, SPE, and Paul J. Koszela, Paragon Engineering Services Inc.
Summary. This paper discusses two techniques for sizing oil/water separators. Droplet-settling theory, based on Stokes' law govern-
ing water-droplet movement through a continuous oil phase, is used to develop sizing equations for both horizontal and vertical separa-
tors. Retention-time theory, which calculates theoretical fluid-retention time within the separator by relating liquid flow rates to vessel
geometry, is also presented. Published separator data comparing vessel sizes to allowable capacities are analyzed using both sizing
techniques.
Introduction
The separation of oil and water phases is one of the most common Similarly, liquid droplets in the gas fall perpendicular to the bulk
and least understood processes in a production facility. As fluids flow of the gas phase.
flow into the bottom of the wellbore, up the tubing, and through Fig. 2 shows a horizontal FWKO where very little gas is expect-
surface chokes and equipment, the oil and water are mixed thorough- ed and the gas is recombined with the oil. The oil/water separation
ly. The liquid must eventually be routed to a vessel where it is sepa- mechanics are identical to those in Fig. 1.
rated into a continuous oil phase containing dispersed water droplets Oil/water separation can also occur in a vertical vessel, as shown
(sometimes referred to as an emulsion) and a continuous water phase in Fig. 3. The liquid is routed to below the oil/water interface by
containing dispersed oil droplets. These liquids are then routed to a downcomer. In a vertical vessel, the water droplets entrained in
oil- and water-treating systems, respectively. the oil settle countercurrent to the upward oil flow and the oil
The vessels that perform this separation are usually called three- droplets entrained in the water rise countercurrent to the downward
phase separators when a significant amount of gas must be sepa- water flow. Ref. 1 describes in more detail the different equipment
rated from the liquid in the same vessel, or freewater knockouts types and operating problems; Ref. 2 presents the droplet-settling
(FWKO's) when there is little or no gas. In some areas, FWKO theory as it applies to the settling of liquid droplets in the gas-
refers to a vessel where very little gas must be separated and the continuous phase.
separated gas is recombined with the oil and flows out the oil out-
let. Other names that describe equipment performing this initial sepa- 011 Treating vs. Water Treating
ration of the liquid phases are wash tanks, settling tanks, and gun It is intuitively obvious that a separator designed to treat oil will
barrels. have a different flow pattern and internals than one designed to treat
This paper discusses the validity of two different techniques water. Is the oil/water separator essentially an oil-treating or a water-
(retention-time and droplet-settling theory) for choosing vessel size. treating device? That is, besides its main function of separating the
Ref. 1 provides a detailed description of these theories and the deri- liquid into two phases, is the quality of the oil or water outlet of
vation of appropriate equations. overriding concern? Certainly, if the oil outlet is not treated fur-
A review of the pertinent literature by the SPE Reprint Series ther, as in a gun barrel and many light-oil and condensate facili-
Committee while it was developing a volume on surface produc- ties, then the oil quality would govern. If the water is not treated
tion equipment indicated that operators have not published actual further, as in a skim vessel, then the water quality IS a primary
data on flow rates, liquid properties, and vessel geometries so that concern. Most oil/water separators, however, have both oil and
field experience can be used to validate these theories. In this paper, water treating downstream. In such cases, we must consider the
we discuss data published by vendors on the capacities of standard physics of the situation before concluding that its function is primar-
low-flow-rate separators. Operators and engineering companies have ily oil or water treating.
used these data for several decades to size separators. If these data The oil viscosity is normally one or two orders of magnitude great-
can be shown to be consistent with one of the sizing theories de- er than the water viscosity. Thus, an oil droplet can rise through
scribed, that theory may present an appropriate scaling factor to the water much more easily than a water droplet can settle through
size separators for modem conditions of much higher flow rates. the oil. In addition, experience has shown that the natural emulsi-
This paper does not discuss the further complicating factor of fiers in the liquid tend to make much more stable water-in-oil emul-
chemical treatment. Conceptually, some optimum economic bal- sions than oil-in-water emulsions. The result is that a separator
ance between vessel size and chemical usage should exist. Research properly sized to treat the oil will provide a reasonable water qual-
is under way to better our understanding of this phenomenon. ity. The water-treating system may contain additional separators
designed primarily to ensure a low oil content in the outlet water.
Equipment Description Such devices are described in Ref. 1 and are not considered in this
Fig. 1 shows a typical horizontal three-phase separator. Fluid enters analysis.
the vessel and hits an inlet diverter, where the majority of the gas
is separated. The liquid falls to below an oil/water interface where Inlet Dlverter
the liquid is "water washed." The oil and its entrained water Some form of inlet diverter is required, even in the classic FWKO
droplets flow horizontally to the oil weir, where a level controller shown in Fig. 2. The flow into this vessel normally comes from
regulates the rate at which it leaves the vessel. The water-continuous a higher-pressure source. Gas is liberated as the liquid reaches its
phase and oil droplets entrained in it flow horizontally to the water new pressure and temperature conditions. If the gas is not separat-
outlet. The discharge rate is regulated by an interface controller. ed by the inlet diverter and is forced to rise through the liquid, it
As the oil- and water-continuous phases flow the length of the ves- will bring water droplets with it, increasing the outlet-oil water cut.
sel, gravity forces cause the water droplets to settle perpendicular
to the bulk flow in the oil-continuous phase and the oil droplets Water Washing
to rise perpendicular to the bulk flow in the water-continuous phase.
It is extremely important that the inlet fluid be water washed. Be-
cause of the severe turbulence and unsteady-state flow conditions
Copyright 1990 Offshore Technology Conference to which the inlet fluid is subjected, immediately downstream of
Water
~
Water
Outlet
Fig. 2-FWKO.
~ ,~
,..---{)iQ--- G as Out
Inlet
CHIMNEY ~
Q)
Emulsion .J:
~
Downcomer ---+---<+-I
,~
Oil Out
Spreader---~~~===r~~ 3
Water Water .J:
Level Control
Valves
~ ~,
%Water in sample
~--------(----
Time
Fig. 4-Water washing. Fig. 5-Growth of water layer with time.
These are overflow equations where the overflow rate is deter- cal results of this difference. For the given properties, retention-time
mined by the product of a constant derived from unit conversion theory would indicate that a diameter smaller than 50 in. [127 cm]
and geometry, and a constant, C, that is a function of the liquid is acceptable. Settling theory states that a diameter less than 50 in.
properties. That is, both Eqs. 4 and 6 are a form of overflow equa- [127 cm] is unacceptable for removing water droplets greater than
tion where the allowable overflow is fixed by the properties (i.e., 500 JLm from the oil.
difference in specific gravity and viscosity) of the liquids. It is un- A similar problem exists for horizontal vessels. Retention-time
clear at this point whether the parameter d m is a constant or a func- theory (Eq. 3) states that the product of diameter squared times
tion of the liquid properties or flow configuration in the vessel. In length is equal to a constant. Settling theory (Eq. 6) says that the
the remainder of this paper, we use the term "overflow velocity" product of diameter times length is equal to a constant. Fig. 8 il-
when fluid properties are unknown and we are describing the con- lustrates the practical consequences of this difference for vessels
cept that qolA is the correct scaling factor. We use the term 80% full of liquid.
"settling theory" to describe the more specific concept that the value
of overflow velocity can be derived from liquid properties by use Historical Basis for Sizing
of C. Very little has been written on which of the two theories more ac-
curately models the process in the vessel. It is instructive to review
Importance of Difference Between Retentlon- the sizing charts and concepts published and field-proven by
Time and Settling Theories manufacturers to see whether they can provide guidance.
In sizing a separator, we must know which of the two theories gov- Table 1 shows capacities published by Parkersburg in their cata-
erns. For given flow conditions, retention-time theory (Eq. 2) states log in the 1950's. Both the oil and water capacities are based on
that the diameter of a vertical vessel changes as the height of liquid 10 minutes of retention time. Table 2 shows similar data for verti-
volume changes. Settling theory (Eq. 4) states that the required di- cal FWKO's for Black, Sivalls and Bryson (1959). Except for the
ameter is independent of fluid height. Fig. 7 illustrates the practi- 12 x 12-ft [3.7 x3.7-m] vessel, the maximum fluid capacities list-
200
180
160
..
.<::
()
140 \ \
c: MINIMUM DIAMETER FOR
120 ~-------II\~-----~-r- SETTLING
\
( dm =500 MICRONS)
0
100
GIVEN: -r-
::> qo =qw =1000 bbl
0 (1')0 =(I,)w =10 Min.
::; 80 -r-
.<:: " ~=15cp
~ t.1 wo =0.15
~--------+-------
60 --
- - K , . / R E T E N T I O N TIME EQUATION
40
-----r----.. --il------II--------==----F::;:::::::::==t-
......
20 - 1 - - - - - 1 - - - - 1 - -
-
o -l--------t------~~L-----+_------1_-------r------_i--------t_
24 36 48 60 72 84 96 108
d ( Inches)
\
\
\
50+--------r--~--~~------+_------_r------~--------+_------_+
\
\
\
\
\
-40+--------r------~~------+_-------r------~--------1-------_+
~
II
II \ GIVEN:
\
,
'" "0",
\ qo -qw =5000 bbl
(1')0 =(t,)w. to Min.
_ 30 +--------r--""",:--~-------"---1-------_+--
1'=20 cp
~
.1'Ywo· 0.15
d m -500 Microns
20 ____ ~~~--~-------+--------L-------~
~~SETTLING EQUATION
~
'----_!-------T-------__
10+--------r------~--------+_-----·---~------_+~~--_4--------+
0+----~--r_~~--41--~1--_+--_+
24 36 46 60 72 84 96 108
d (Inches)
TABLE 1-PARKERSBURG FWKO DATA (1950's) TABLE 2-BLACK, SIVALLS AND BRYSON
FWKO DATA (1959)
Vertical Design
Capacity" Vertical Design
(BID) Overflow Retention
Shell
Free Diameter Shell Fluid Capacity Velocity' Time"
Diameter Length
(ft) (B/D-ft2) (minutes)
(ft) ~ -Oil- Water
--
J!!L (BID)
Vertical Design
N
Overflow Retention
Diameter Shell Water Capacity Velocity· Time·· a
0..
150
(B/D-ft2) en
(ft) J!!L (BID) (minutes)
4 10 300 to 1,300 24 to 103 52 to 12
...<w
0:
6 10 600 to 3,000 21 to 106 60 to 12 ;; 100
8 10 1,000 to 5,000 20 to 100 60 to 12 0 RANGE OF OVERFLOW
...I RATES RECOMMENDED
u.
10 10 1,500 to 8,000 19 to 102 64 to 12 0:
W
IN TABL.ES 11 & JJI