Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

Journal of Dispersion Science and Technology

ISSN: 0193-2691 (Print) 1532-2351 (Online) Journal homepage: http://www.tandfonline.com/loi/ldis20

Modeling liquid-liquid interface level in a


horizontal three-phase separator with a bucket
and weir

Chaojun Deng, Weixing Huang, Haoyuan Wang, Hongqiang Lan & Boya Xu

To cite this article: Chaojun Deng, Weixing Huang, Haoyuan Wang, Hongqiang Lan & Boya Xu
(2018): Modeling liquid-liquid interface level in a horizontal three-phase separator with a bucket and
weir, Journal of Dispersion Science and Technology, DOI: 10.1080/01932691.2018.1452755

To link to this article: https://doi.org/10.1080/01932691.2018.1452755

Published online: 02 Apr 2018.

Submit your article to this journal

Article views: 13

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=ldis20
JOURNAL OF DISPERSION SCIENCE AND TECHNOLOGY
https://doi.org/10.1080/01932691.2018.1452755

none defined

Modeling liquid-liquid interface level in a horizontal three-phase separator


with a bucket and weir
Chaojun Denga, Weixing Huanga, Haoyuan Wanga, Hongqiang Lanb, and Boya Xua
a
School of Chemical Engineering, Sichuan University, Chengdu, P. R. China; bChina Petroleum Engineering Co., Ltd Southwest Branch, Chengdu,
P. R. China

ABSTRACT ARTICLE HISTORY


The flow behavior in a three-phase separator with a bucket and weir was analyzed, and a theoretical Received 26 November 2017
equation for calculating the liquid-liquid interface level (HH) in gravity separation zone was derived. The Accepted 24 February 2018
analysis indicates that the HH increases as the flow rate and the density ratio of heavy to light liquid KEYWORDS
increase, and decreases linearly with increasing height difference between heavy and light liquid overflow Liquid-liquid interface level;
weirs. The calculated HH under different operating conditions is in good agreement with the liquid-liquid separation;
experimental. With the proposed equation, the thicknesses of heavy and light liquid layers can be three-phase separator; weir
calculated, and then the minimum lengths of two phase layers required for separation can be determined height difference
separately. From the minimum lengths it can be clearly indicated that the governing step of liquid-liquid
separation is in heavy or light liquid layers, hence the sizing of the separator can be optimized.
GRAPHICAL ABSTRACT

1. Introduction residence time; the rising time of oil droplets must be greater
than water residence time; the settling time of water
Three-phase gravity separator is widely used as primary
droplets must be greater than oil residence time.[8–10] The
process equipment in the oil and gas production.[1–4] The
specifications recommend some residence time for liquid-
horizontal separators with a bucket and weir are preferred
liquid separation since uncertain liquid-liquid interface level
when there are heavy oils or large amounts of emulsion or
in the separator leads to the difficulty in predicting
non-stable feed mixtures which leads the liquid-liquid
droplet settling or rising time.[8,9,11] However, the residence
interface level control to become difficult.[5–7] Figure 1 show
time method for sizing liquid-liquid separation zone has
a cutaway view of a three-phase separator with a bucket and
some limitations. The selection of the residence time was
weir, which can be divided into three parts: entrance zone,
so reliant on empirical data. This rules-of-thumb design
gravity separation zone and overflow zone. The size of gravity
may cause various design results even for the same operating
separation zone has a great influence on the performance of
condition.[2,9] The residence time method does not take into
gas-liquid-liquid separation.
account the amount of feed mixture, and the proportion of
API 12J and GPSA both point out fundamental principles
each component in the mixture. Obviously, these factors
for sizing of gravity three-phase separators, i.e., the settling
have impact on the separator performance.[9] Besides, a
time of oil and water droplets must be greater than gas

CONTACT Weixing Huang hwx@scu.edu.cn College of Chemical Engineering, Sichuan University, Chengdu 610065, China.
Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/ldis.
© 2018 Taylor & Francis
2 C. DENG ET AL.

Figure 1. Schematic of a horizontal three-phase separator with a bucket and weir.

three-phase separator equipped with coalescing plate pack is inviscid, and incompressible between points ‘a’ (at the
able to operate at two or three times as much as the gas-liquid interface) and ‘b’ (at the liquid-liquid interface) as
capacity of the same-sized conventional separator since the depicted in Figure 2, the energy equation is written in head
plate pack greatly decreases droplet settling distance.[3,12] form,
However, the residence time method is inadaptable to
accurate design of three-phase separators with coalescing Pa u2 � Pb
þ L þ HLw HH þ hL;ow ¼ ½1�
plate pack. qL g 2g qL g
Arnold & Stewart and Monnery & Svrcek considered that
where Pa, Pb are the pressures at the gas-liquid and
the separator with a bucket-and-weir design has advantages
liquid-liquid interfaces, Pa, respectively. ρL is the
of controlling the liquid level by adjusting the height difference
density of light liquid, kg/m3. HLw stands for the height of
of weirs.[5,6] They used the hydrostatic equation to calculate
light liquid weir, m. HH means the liquid-liquid
the liquid-liquid interface level (HH) which provided the
interface level (i.e., the thickness of heavy liquid layer) in
possibility of calculating droplet settling and rising time, and
gravity separation zone, m. uL is the overflow velocity of
further sized the separator by droplet settling theory. However,
light liquid, m/s.
the HH may be not accurate enough since the effect of liquid
flowing over the weirs on the HH is ignored in the hydrostatic QL
equation. uL ¼ ½2�
BLw hL;ow
In this work, liquid-liquid flow behavior in a bucket-and-
weir separator was analyzed, and a theoretical equation for where QL is volumetric flow rate of light liquid, m3/s; BLw is
liquid-liquid interface level was derived. The calculated the length of light liquid weir, m. hL,ow denotes the height of
results show good agreement with the experimental light liquid crest over the weir in m, using Eq. (3) which is
data. Based on the proposed equation, the thicknesses of derived by the Bernoulli equation.
gas, light and heavy liquid layers can be calculated, thus � �
the minimum lengths of gas, light and heavy liquid layers QL 3 1 2=3
hL;ow ¼ pffiffiffiffiffi ½3�
required for separation can be determined separately. The CBLw 2 2g
design result can clearly indicate the separator size is
governed by which separation zone, gas, light liquid or where C is an overflow coefficient, depending on the weir
heavy liquid layers. As a result, the sizing of separator can configuration.
be optimized. Taking into account the local resistance loss due to
the contraction flow, the energy equation for the
steady incompressible flow between the points ‘b’ and
‘c’ (at the surface of heavy liquid over the weir)
2. Liquid-liquid interface level in the
becomes,
bucket-and-weir separator
2.1. Liquid flow analysis Pb Pc u2 �
þ HH ¼ þ H þ HHw þ hH;ow þ hf ½4�
qH g qH g 2g
Although the flow field in the entrance zone is highly
turbulent when the mixture enters the three-phase separator where ρH is the density of heavy liquid, kg/m3. Pc is
and hits an inlet diverter which provides primary separation the pressure at the free liquid surface, which is
between the gas and liquid, the liquid flow become smooth approximately equal to Pa. HHw is the height of heavy
and little turbulent via a perforated baffle, and then there will liquid weir, m. uH denotes overflow velocity of heavy liquid,
be a clear liquid-liquid interface in the downstream of gravity m/s. hH,ow means the height of heavy liquid crest over the
separation zone.[13] Assume the flow is steady, isothermal, weir, m.
JOURNAL OF DISPERSION SCIENCE AND TECHNOLOGY 3

Figure 2. Flow behavior in a horizontal bucket-and-weir separator.

The local resistance of heavy liquid flowing through the Rearrange Eq. (6) to obtain the liquid-liquid interface level,
passage under the light liquid bucket, hf, which can be H H,
calculated by Eq. (5). � �
c
HH ¼ HLw DH
u2B c 1
hf ¼ n ½5� � �
2g chH;ow hL;ow cu2H u2L
þ þ ½7�
c 1 2g ðc 1Þ
where ξ is the local resistance coefficient due to the contraction
cnu2B
and expansion flow, ξ ¼ 1.5. uB means average velocity of þ ¼ hH1 þ hH2 þ hH3
heavy liquid passing through the passage. 2g ðc 1Þ
Combining Eqs. (1), (4) and (5) lead to
where γ is defined as the density ratio, γ ¼ ρH/ρL. DH is the
u2L � height difference between heavy and light liquid weirs, m.
qL þ qL HLw HH þ hL;ow þ qH HH From Eq. (7), HH is determined by three parts: hH1 repre-
2g
½6� sents the contribution of fluid static pressure, hH2 reflects
u2H � u2
¼ qH þ qH HHw þ hH;ow þ qH n B the effect of two-phase flow over the weir, hH3 means the effect
2g 2g of local resistance loss. This equation provides an estimation of

Figure 3. Experiment setup; (1) oil-water separator, (2) calming baffle (3) oil tank, (4) water tank, (5),(6) centrifugal pump, (7) oval wheel flowmeter,
(8),(9) rotor flowmeter, (10) hydraulic ejector, (11) SK static mixer, (12) globe valve.
4 C. DENG ET AL.

Table 1. Operating condition of oil-water separation experiment. behavior and separation process. The length, width and height
Fluid Density (kg/m3) Viscosity (mPa · s) Flow rate (L/h) of the separator were 1500, 305, 350 mm, respectively, and the
Water 997 0.903 250 ∼ 1050 heights of oil and water weirs were 290 mm and 278 mm,
White oil 889 15.0 49 ∼ 307 respectively. The oil-water interface levels in the separator
were extensively measured at different flow rates of oil and
water phases at 25(�2)°C. Table 1 lists the operating condition
the liquid-liquid interface level in the separator when light and
of the experiment.
heavy liquid flow rates and the height difference between the
two weirs are given.
3. Results and discussion
3.1. Measured oil-water interface levels at different
2.2. Experimental study
flow rates
In order to validate the derived equation, experiments were
Figure 4 shows the comparison between calculated oil-water
performed in an oil-water separator with a bucket and weir.
interface levels by Eq. (7) and the measured ones at different
As illustrated in Figure 3, white oil and water phases were
flow rates. The average deviation between the calculated and
delivered by two centrifugal pumps, respectively. The two
fluids were merged and premixed in a hydraulic ejector, and
well mixed via a SK static mixer. Then the oil-water mixture
was delivered in a rectangular bucket-and-weir separator made
in Plexiglas which was visualized for the observation of flow

Figure 6. Contribution of the hHi on oil-water interface level HH at different flow


Figure 4. Comparison of the calculated and measured oil-water interface level. rates.

Figure 5. Relationship of oil-water interface level HH with water and oil flow Figure 7. Effect of the height difference of two weirs and flow rate ratio on
rates. oil-water interface level HH.
JOURNAL OF DISPERSION SCIENCE AND TECHNOLOGY 5

case, the water phase is likely to flow over the oil weir into
the oil bucket, reducing the separation efficiency. The HH
can reduce from 281.2 mm to 59.7 mm just by increasing from
6 mm to 30 mm. Thus, the DH should be regarded as an input
adjustable design parameter to control the liquid-liquid
interface at a proper level for normal and safe operation at
different flow rates. In addition, when the total flow rate is
fixed at 1200 L/h, the HH becomes increased with increasing
the QH/QL for a given DH.

3.3. Effect of the density ratio on the HH


Figure 8 shows the effect of density ratio γ of water to oil on
the HH. At a fixed DH, the HH increases logarithmically with
increasing the γ. For heavy oils, it is very easy to precisely
adjust the oil-water interface via varying the DH. For
Figure 8. Effect of the density ratio on oil-water interface level HH. example, when the γ is 1.08, the HH reduces from
251.8 mm to 9.6 mm only by increasing the DH from
6 mm to 23 mm. While for light oils, when the γ is 2.9, the
experimental ones is only 2.6%, and the maximum deviation is HH only varies 9.9% when increasing the DH from 6 mm
less than 5.9%. As shown in Figure 5, the calculated HH to 23 mm. Hence, the bucket-and-weir type design is
increases with the increase of the flow rate, which are in good preferred when liquid-liquid interface level control may be
agreement with the experimental ones, while the calculated difficult, typically for heavy oils or two liquids with low
hH1 by the hydrostatic equation cannot reflect this variation density differences.
trend. Hence, the derived equation is validated for estimating
liquid-liquid interface level in the bucket-and-weir separator.
Figure 6 shows the relative contribution of each part hHi on
the HH at different flow rates. Although the hH1 caused by fluid 4. Optimization of the sizing of a three-phase
static pressure is dominated, the other factors become more separator
and more important with the increasing flow rates. For The proposed equation can be used to calculate liquid-liquid
instance, under the operating condition of QH ¼ 1050 L/h interface level (HH) if the flow rates and height difference
and QL ¼ 300 L/h, the sum of hH2 and hH3 exceeds 19% of between the two weirs (DH) are given. For sizing a bucket-
the HH. Hence, the effect of two-phase flow on the HH cannot and-weir separator, the DH shall be an input adjustable para-
be neglected. meter to control the HH under different operating conditions.
Further the settling and rising time of heavy and light liquid
droplets in gas and liquid layers can be predicted by droplet
3.2. Effect of the height difference of two weirs and
settling theory.[5,7,10,14] The separation criterion is that the
flow rate ratio on the HH
residence time of heavy and light liquids must be greater than
Figure 7 shows the influence of the height difference between or equal to the settling and rising time of droplets in the gas
the two weirs DH on the HH at different flow rate ratios. For a and liquid. Hence the minimum lengths of gas, light and heavy
given QH and QL, the HH always decreases linearly with the liquid layers required for gas-liquid-liquid separation can be
increase of the DH. When the QH and QL are 1000 L/h and determined separately. Through the design result, we can
200 L/h, respectively, and the DH is set only 6 mm, the HH clearly judge whether the governing step of three-phase separ-
reaches 281.2 mm which is close to oil-weir height. In this ation is in gas, heavy liquid or light liquid layers, further the

Table 2. The sizing of gravity separation zone.


Gas layer Heavy liquid layer Light liquid layer
Thickness HG ¼ (0.2q
∼ffiffi0.3)D
ffiffiffiffiffiffiffi HH, by Eq. (7) HL ¼ D − HG − HH
2 2
Terminal velocity UTG ¼ K qLq qG UTH ¼
ðqH qL ÞgdPL
UTL ¼
ðqH qL ÞgdPH
G 18lH 18lL
Axial velocity UAG ¼ QAGG UAH ¼ QAHH UAL ¼ QALL
Settling time tTG ¼ HUGT tTH ¼ UHTHH tTL ¼ UHTLL
Retention time tr ¼ ULAGG tH ¼ ULAHH tL ¼ ULALL
The minimumrequired length qffiffiffiffiffiffiffiffiffi
LG ¼ QKA
G HG qG LH ¼ ðq 18lqH QÞgd
H HH
2 A LL ¼ ðq 18lqL QÞgd
L HL
2 A
G q q L G H L PL H H L PH L
Length of gravity separation zone Repeat the calculation and adjust DHor HG to make LG, LH and LLget close.L1 ¼ maxfLG ; LH ; LL g
UTG, UTH and UHL are terminal velocities of light and heavy liquid droplets in the gas, heavy liquid and light liquid layers, m/s, respectively. K is the settling velocity
coefficient in m/s. UAG, UAH and UAL are axial velocities of gas, heavy liquid and light liquid, m/s, respectively. tTG, tTH and tTL are settling time of liquid droplets in the
gas, heavy liquid and light liquid layers, s, respectively. dPH and dPL are the diameters of heavy liquid and light liquid droplets, m, respectively. µH and µL are dynamic
viscosities of heavy and light liquids, Pa · s, respectively.
6 C. DENG ET AL.

sizing of separator can be optimized. The design flowchart is AL cross-sectional area of light liquid layer in
given in Table 2. gravity separation zone [m2]
As depicted in Figure 1, the length of gravity separation B weir length [m]
zone (L1) is determined by the maximum value of gas layer C overflow coefficient in Equation (3) [-]
length (LG), light liquid layer length (LL) and heavy liquid dPH heavy liquid droplet diameter [m]
layer length (LH). For optimizing the separator size, the above dPL light liquid droplet diameter [m]
three lengths should be equal or close by appropriately D separator diameter [m]
adjusting the vessel diameter (D), the gas space height (HG) hf local resistance loss [m]
or DH. hH,ow heavy liquid crest height over the weir [m]
▶ If LG >> max{LL, LH} (gas-liquid separation controls), hL,ow light liquid crest height over the weir [m]
then increase the HG or D and repeat the design. H height (thickness) [m]
▶ If LG << min{LL, LH} (liquid-liquid separation controls), DH height difference between heavy and light liquid
then decrease the HG or D and repeat the design. If weirs [m]
LL >> LH (heavy liquid droplets settling out of the light L seam-to-seam length of the separator [m]
liquid controls), then increase DH. If LL << LH (light L0 length of the entrance zone [m]
liquid droplets rising out of the heavy liquid controls), L1 length of gravity separation zone [m]
then decrease DH and repeat the design. L2 length of the overflow zone[m]
The sizing of the entrance and overflow zones are given by Q volumetric flow rate [m3/s]
Monnery and Svrcek.[6] The total length of the separator is the u overflow velocity [m/s]
sum of L0, L1 and L2. Finally, check the aspect ratio (L/D). UT terminal velocity of liquid droplet [m/s]
Arnold and Stewart suggested the L/D should be between 3
and 5.[5] Greek symbols
γ density ratio of the heavy to light liquid [-]
μ dynamic viscosity [Pa · s]
5. Conclusions ξ local resistance coefficient in Equation (5) [-]
The derived equation provides a good prediction for the ρ density [kg/m3]
liquid-liquid interface level (HH) in a horizontal three-phase
separator with a bucket and weir. The calculated HH is in Subscripts
good agreement with the experimental. Further analysis G gas
reveals that the HH increases with the increase of the flow rate L light liquid
and density ratio, and decreases as the height difference H heavy liquid
between the two weirs DH increases. This suggests that the w overflow weir
DH shall be regarded as an input adjustable design parameter
to control the liquid-liquid interface level under different
operating conditions in the design of this type of separators. References
On this basis, the thicknesses of the space occupied by gas,
[1] Fossen, M.; Arntzen, R.; Hemmingsen, P. V.; Sjöblom, J.; Jakobsson,
light and heavy liquids can be calculated, respectively, and J. J. Dispersion Sci. Technol. 2006, 38(10), 1509–1515.
then the rising or settling time of light and heavy liquid dro- [2] Laleh, A. P.; Svrcek, W. Y.; Monnery, W. D. Can. J. Chem. Eng.
plets can also be determined by droplet settling theory. 2012, 90(6), 1547–1561.
Accordingly, the minimum lengths of gas, light and heavy [3] Jaworski, A. J.; Meng, G. J. Pet. Sci. Eng. 2009, 68(1–2), 47–59.
[4] Wang, X.; Yan, Y.; Xu, Z. J. Dispersion Sci. Technol. 2016, 38(10),
liquid layers required for gas-liquid-liquid separation can
1509–1515.
be calculated separately. The design results can clearly show [5] Arnold, K.; Stewart, M. Surface Production Operations, Vol 1: Design
the separator size is governed by the size of gas, light liquid of Oil-Handling Systems and Facilities. 3rd ed.; Gulf Professional
layer or heavy liquid layers, so that the sizing of the separator Publishing: Burlington, 2008.
can be optimized. [6] Monnery, W. D.; Svrcek, W. Y. Chem. Eng. Prog. 1994, 90(9), 29–40.
[7] Wiencke, B. Int. J. Refrig. 2011, 34(8), 2092–2108.
[8] API Specification 12J. Specification for Oil and Gas Separators. 8th
ed.; American Petroleum Institute: Washington, DC, 2008.
Acknowledgments [9] Gas Processors Suppliers Association (GPSA). GPSA Engineering Data
The authors would like to acknowledge China Petroleum Engineering Book. 12th ed.; Gas Processors Suppliers Association: Tulsa, 2008.
Co., Ltd Southwest Company for supporting this work. [10] Grødal, E. O.; Realff, M. J. Optimal Design of Two and Three Phase
Separators: A Mathematical Programming Formulation. SPE Annual
Technical Conference and Exhibition, Houston, Oct 3–6, 1999.
[11] Arnold, K. E.; Koszela, P. J. SPE Prod. Eng. 1990, 5(1), 59–64.
Nomenclature [12] Sarkar, S.; Kamilya, D.; Mal, B. C. Water Res. 2007, 41(5), 993–1000.
[13] Wilkinson, D.; Waldie, B.; Nor, M. I. M.; Lee, H. Y. Chem. Eng. J.
A cross-sectional area of the separator [m2] 2000, 77(3), 221–226.
AH cross-sectional area of heavy liquid layer in [14] Hernández-Suárez, R.; Puebla, H.; Aguilar-López, R. Ind. Eng.
gravity separation zone [m2] Chem. Res. 2007, 46(21), 7008–7017.

You might also like