http://pubs.acs.

org/journal/acsodf Article

Dynamic Simulator for Three-Phase Gravity Separators in Oil

Production Facilities
Shangfei Song,* Xuanzhang Liu, Chenxuan Li, Zhe Li, Shijia Zhang, Wei Wu, Bohui Shi, Qi Kang,
Haihao Wu, and Jing Gong*

Cite This: ACS Omega 2023, 8, 6078−6089 Read Online

ACCESS
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Metrics & More Article Recommendations

ABSTRACT: In this study, a dynamic simulator for three-phase

gravity separators in oil production facilities is proposed. The mass
Downloaded via 91.106.57.67 on December 15, 2023 at 06:48:37 (UTC).

conservation equation is established to calculate the pressure, water

level, and oil level in the separator and the mass balance equation of
the dispersed phase to calculate the oil−water separation efficiency.
The proportional integral controllers are applied to control the water
level, oil level, and pressure in the separator by setting the opening of
the three outlet valves of oil, gas, and water. The model is verified
using field data by means of the given valve opening and given
proportional integral controller parameters, respectively. Subse-
quently, the verified simulator is applied to study the dynamic
behavior of the separator filling process and the effect of pressure, oil
level, and water level setpoint changes on the separator operating
status. A detailed analysis of the changes in the liquid level, pressure, and opening of three outlet valves is presented. Then, the
effects of operating conditions such as the inlet flow, water setpoint, and weir height on the separation efficiency are discussed. This
simulator can be applied for the design of oil, gas, and water three-phase separation processes. In addition, through this simulator,
the parameters that are difficult to be measured by instruments during the operation of the separator can be calculated, providing
technical support for the construction of the digital twin of the separator.

1. INTRODUCTION control the running state of the separator. The establishment of

The digital twin is one of the buzzwords of the last several years dynamic simulators plays an indispensable role in the digital twin
in the petroleum industry.1 Digital twins can provide innovative practice of three-phase gravity separators.12
ways of using simulations to reshape the digital transformation Many studies have been carried out to investigate the dynamic
of Oil & Gas 4.0,2 which enable operators to respond to behavior and separation efficiency of separators, most of which
potential equipment failures pre-emptively and plan remedial use computational fluid dynamics (CFD).13−15 However, due to
measures before failures occur.3 Meanwhile, real-time data can the huge demand for computing resources, CFD-based models
be gathered from sensors and simulator results to know the cannot be applied to the practice of digital twins.16,17 Few
accurate state of the equipment.4 Digital twins are believed to studies have been carried out from the perspective of
help identify opportunities for process improvement, produc- conservation of mass and first-principles.8
tion data improvement, operating cost reduction, and health, Sayda and Taylor18 completed the dynamic modeling of the
safety, and environment (HSE) risk reduction.5 separator in 2007 by developing a simple phase equilibrium
In the process of oil production, the produced fluid gushing model that extended the American Petroleum Institute (API)
from the Christmas tree is generally a mixture of water, light static design criteria. On this basis, a simple proportional integral
hydrocarbons, and oil,6 which need to be separated before they (PI) controller was added to realize the control and level loop
can be metered and exported.7 Thus, the separators are an process dynamics identification.19 Backi et al.8 presented an
important unit of the gas−oil separation plant, and controlling
the separator operation8 is crucial to achieving the required Received: December 30, 2022
efficiency and specification of the fluids exiting from the Accepted: January 26, 2023
separator for further processing.9 The water level, oil level,10 Published: February 3, 2023
pressure, and separation efficiency are the key parameters for the
operation of the three-phase gravity separator.11 The digital twin
of the three-phase gravity separator can effectively monitor and
© 2023 The Authors. Published by
American Chemical Society https://doi.org/10.1021/acsomega.2c08267
6078 ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Figure 1. Schematic of a three-phase gravity separator.

approach for the control and simulation of the liquid level, level can be calculated according to eq 3. The first part of eq 3 is
pressure, and oil−water separation efficiency including the volume of the separator cylinder, and the second part is the
simplified droplet balance calculations. Three PI controllers volume of the elliptical head of the separator edge.
were applied to control the flowrate of gas, oil, and water phase
outlets, but the valve is not taken into account. Das et al.20
proposed a coalescence-based dynamic gravity separator model
(1)
to obtain the oil content in the water outlet and the water
content in the oil outlet. In addition, a number of mature
commercial simulators have been proposed, such as HYSYS,21 (2)
OLGA,22 and K-spice,23 which can simulate the processes of the
separator. However, the models and algorithms of commercial
software are not available to the public.
In this work, a dynamic simulator for a three-phase gravity
separator in oil production facilities is proposed. The remainder
of the paper is organized as follows. In Section 2, the model (3)
development is shown in detail. The model verification is
presented in Section 3. The discussions about dynamic behavior where, V is the volume of the liquid on the left side of the weir,
due to separator filling and setpoint change and how the m3. Q is the flowrate of liquid, m3/s. Δt is the time step, s. L is the
separation efficiency is affected by the operating conditions are length of the left part of the weir plate in the separator, m. D is
performed in Section 4. Finally, the conclusions are presented. the diameter of the separator cylinder, m. H is the liquid level, m.
hi is the surface width of the ellipse head, m. Subscripts l, w, and o
2. MODEL DESCRIPTION stand for total liquid, water, and oil, respectively. Superscripts t
and t+1 stand for the current time step and the next time step,
The separation process is shown in Figure 1. In the separation respectively. Subscripts in and out stand for input parameters
zone on the left side of the weir, the mixed liquid is usually and output parameters, respectively.
divided into three continuous phases, with a continuous water The pressure in the separator can be calculated according to
layer at the bottom and a continuous oil layer above the water eqs 4 and 5.
layer. The gas phase is at the top of the separator.24 The water
level refers to the water level in the separation area on the left
side of the weir plate, and the oil level refers to the oil level on the
right side of the weir plate. (4)
In the separation zone on the left side of the weir, the oil
droplets dispersed in the water and the water droplets dispersed
in the oil are separated from their respective bulk phase. In this (5)
work, it is assumed that the droplets have reached the dynamic
equilibrium of aggregation and fragmentation before entering where n is the molar amount of gas in the separator, mol. ρg is the
the separation section. Therefore, the phenomenon of droplet density of gas, kg/m3. M is the molar mass, kg/mol. P is the
aggregation or fragmentation is ignored in this work. It is pressure in the separator, Pa. z is the compression factor, which
assumed that neither the oil phase outlet nor the water phase is calculated by the Peng−Robinson equation of state.25R is the
outlet contains gas; ignore the part of gas dissolved in oil or gas constant, 8.314 J/(mol·K). T is the temperature, K.
water. Moreover, only mass conservation is considered inside Subscript g stands for gas.
the separator, and energy conservation is ignored. The outlet flow of the oil, gas, and water from the separator
2.1. Governing Equations of the Liquid Level and depends on the pressure difference between the upstream and
Pressure. For the calculation of the liquid level on the left side downstream of the valve, the physical properties of the fluid, and
of the weir, it is considered that the liquid is incompressible. The the properties of the valve. Referring to the industry standard
volume of the fluid and water on the left side of the weir can be ISA-75.01.01-2007,26,27 the mass flowrate at the outlet of the
obtained according to eqs 1 and 2 respectively, and the liquid separator is determined by eqs 6−8
6079 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Figure 2. Flow of each small control body in the gravity separator model; convection flow is shown in yellow and the non-convective flow of water and
oil droplets is shown in green and red, respectively.

(6)

(7)
(11)

where σ is the droplet number density, m−3. N is the number of

(8)
droplets. V is the volume of the control body, m3. v is the velocity
where G is the total mass flowrate through the valve, kg/h. α is of the droplet, m/s. A is the area of the interface of the control
the adjusted vapor mass fraction, dimensionless. Hg is the inlet volume, m2. Superscripts i and j represent the index of the
gas mass fraction of the valve. Fp is the pipe geometry parameter. control body. Subscript k stands for the index of the droplet
Fp is a dimensionless quantity; if the diameters of the inlet pipe group. Subscripts x and y represent the parameter in the x and y
and the outlet pipe of the valve are equal, Fp is equal to 1. If the directions, respectively. Subscripts left and bottom represent the
inlet and outlet diameters of the valve are different, the value of left and bottom interfaces of the control volume, respectively.
Fp can be found from the valve manufacturer’s catalog. Cv is the Assuming that the flow above and below the oil−water
effective flow coefficient, m2. ρl is the liquid density, kg/m3. Pinlet
interface is an independent flow, the lateral velocity of the
and Poutlet represent the pressure at the inlet and outlet of the
valve, respectively, kPa. Y is the gas expansion coefficient. Cv, Xt, droplets in the gravity separator is expressed as eq 12.
and Fk are provided by the valve manufacturer.
2.2. Mass Balance Equation of the Dispersed Phase.
(No + Nw)Nx grids are generated in this section. The separator is
divided into Nx column areas horizontally, Nx − 1 columns on
the left side of the weir plate, and one column on the right side of
the weir plate. The oil layer and the water layer are divided into
No and Nw small control bodies in the vertical direction, which (12)
can ensure that the oil−water interface is not in the control body.
The flow of dispersed phase particles due to volume flow is where ΔV is the volume change of the oil or water phase on the
called convective flow, and the flow of dispersed phase particles left of the weir plate, m3. A is the cross-sectional area of the oil or
driven by buoyancy is called non-convective flow, as shown in water layer, m2.
Figure 2. The velocity29 of the dispersed phase in the vertical direction
For ease of calculation, the droplets in the control body are is expressed as eqs 13−15.
classified according to their particle size. The number of droplets
in a unit volume is defined as the droplet number density,28 as
shown in eq 9. The number density of the kth dispersed phase
droplets in the control volume (i, j) at time step t+1 is expressed
as eqs 10 and 11

(9)

(10) (13)

6080 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Figure 3. Control logic flow of the PI controller.

proportional gain coefficient, and the dimensions of KPP, KPol,

and KPwl are pa−1, m−1, and m−1, respectively. e(t) is the error at
(14) time t obtained by subtracting the input value from the set point
value, and the unit is consistent with the input variable. KI is the
integration time constant, and the dimensions of KIP, KIol, and
KIwl are all s.
(15) The control logic of system pressure is shown in Figure 3a.
where vy is the velocity of the dispersed phase in the vertical The difference between the pressure setting point and the
direction, m/s. ρ is the density of oil or water, kg/m3. Δρ is the calculated pressure value at the previous time step is the input of
density difference between the droplet and the continuous the PI controller. The PI controller calculates the flow rate of the
phase, kg/m3. d is the diameter of the droplet, m. μ is the gas outlet by eqs 6−8. Then, the system pressure at the current
continuous phase viscosity, Pa·s. 0.153 and 1.74 are dimension- moment can be obtained by eq 9. Similarly, the control logic for
less constants. the oil level and water level is shown in Figure 3b,c.
2.3. Controllers of the Liquid Level and Pressure. In this
work, the dynamic change process of the gravity separator is 3. MODEL VERIFICATION
controlled by three PI controllers,30,31 and the incremental The dynamic simulator is validated with field data obtained from
algorithm is applied as the control algorithm. The algorithm a separator running in an offshore platform. The horizontal
model is as shown in eq 16 below separator has a diameter of 2.2 m and a length of 5 m. The height
of the weir plate is 1.6 m, and the distance from the separator
inlet is 3.5 m. The components of the separator inlet fluid are
shown in Table 1. Where, components 10−14 are estimated as
unknown components. Details of estimated unknown compo-
nents are shown in Table 2.
The downstream pressure of the oil valve, gas valve, and water
valve is set to 200 kPa. The initial water level in the separator is
set to 1.547 m, the initial oil level on the right side of the weir
(16) plate is 1 m, and the initial pressure in the separator is 1150 kPa.
where the subscripts P, ol, and wl represent pressure, oil level, The gas mole flowrate, oil mole flowrate, and water mole
and water level, respectively. f(t) is the controller output value at flowrate at the separator inlet are 3.696, 14.88, and 1724 kmol/h,
time t, dimensionless. For water level control, oil level control, respectively. The gas volumetric flow rate, oil volumetric flow
and pressure control, the output values are the opening of the rate, and water volumetric flow rate at the separator inlet are
water outlet valve, the opening of the oil outlet valve, and the 0.1936, 8.715, and 31.12 m3/h, respectively. The valve
opening of the air outlet valve, respectively. KP is the coefficients (Cv)32,33 of the three valves of oil, water, and gas
6081 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Table 1. Components of the Separator Inlet Fluida

mole mole fraction % mole fraction % mole mole fraction % mole fraction %
no component fraction % (in gas) (in oil) no component fraction % (in gas) (in oil)
1 N2 0.00658 2.6485 0.08321 8 n-pentane 0 0 0
2 CO2 0.00136 0.2413 0.02928 9 water 98.942 2.5879 0.3125
3 methane 0.240 93.911 4.8015 10 NBP275* 0.0838 0.001 9.8174
4 ethane 0.00136 0.4059 0.05858 11 NBP443* 0.255 0 29.8992
5 propane 0.000454 0.09113 0.0305 12 NBP624* 0.259 0 30.3454
6 n-butane 0.000680 0.02548 0.02024 13 NBP779* 0.168 0 19.6352
7 i-butane 0.000227 0.08763 0.05794 14 NBP958* 0.042 0 4.9088
a
NBP: normal boiling point.

Table 2. Details of Estimated Unknown Components

name NBP275* NBP443* NBP624* NBP779* NBP958*
NBP (°C) 274.9 443.1 624.2 779.2 958.2
molecular weight 203.2 390.5 639.9 884.1 1121.8
density [(kg/m3)] 810.9 888.2 957.1 1008.6 1060
Tc (°C) 444.1 590.8 738.3 859.5 995.1
Pc (kPa) 1610 968 558 340 181
Vc(m3/kmol) 0.874 1.539 2.835 4.877 9.644
eccentricity 0.698 1.1031 1.488 1.7725 2.061

are 52.8215, 57.3018, and 0.3924, respectively. Fp, Xt, and Fk are the pressure and liquid level at the setpoint. The openings of the
1, 0.7, and 0.9, respectively. The particle size distribution of the three outlet valves, the separator pressure, and the liquid level
water droplets in the inlet of the separator is shown in Figure 4. are output by the simulator. The simulation results of pressure,
oil level, water level, and valve openings are compared with the
operation data to verify the accuracy of this simulator.
According to Figures 6−8, the AADs for the separator
pressure, water level, oil level, and water volume fraction in oil in
the separator are 0.15, 0.0022, 0.043, and 26.55%, respectively.
The AADs for valve openings for the oil phase and water phase
are 5.55 and 1.142%, respectively. According to the simulation
results shown in Figure 6, the oil level and water level can be
controlled at the setpoint, but the separator pressure is higher
than the setpoint of 1150 kPa, both from the field data and the
simulation results. This is because the flow capacity of the gas
phase outlet valve is insufficient. Even with the gas valve fully
open, the separator pressure is still higher than the setpoint, both
from the field data and the simulation results.

4. RESULTS AND DISCUSSION

4.1. Dynamic Behavior of the Separator Filling
Process. In this part, the filling dynamics of a separator with
Figure 4. Particle size distribution of water droplets in the inlet of the an initial pressure of 500 kPa is simulated with the verified
separator. simulator. The setpoints of pressure, oil level, and water level are
1150 kPa, 1 m, and 1.59 m, respectively. The boundary
3.1. Simulation with the Given Valve Opening. In this
conditions and PI control parameters are same with the
part, the openings of the three valves of oil, water, and gas are
0.0879, 0.19, and 1. The simulation results of pressure, oil level parameters shown in Section 3.
on the right side of the weir, water level on the left side of the As shown in Figure 9, the separator pressure increases rapidly
weir, and water volume fraction in oil in the outlet are compared from 500 kPa when the separator begins to be filled with the
with the operation data of the separator to verify the accuracy of produced fluid. Initially, since the liquid level and pressure are
this simulator, which are shown in Figure 5. The average lower than the setpoint, the three outlet valves are completely
absolute deviations (AADs) for separator pressure, oil level, closed, resulting in a rapid increase in the pressure and liquid
water level, and water volume fraction in oil in the separator are level. Subsequently, the pressure continues to rise, and when the
0.0943, 0.405, 0.139, and 17.45%, respectively. pressure is approaching the setpoint, the gas valve opening
3.2. Simulation with Given Setpoints of the Oil Level, rapidly increases to 100% under the action of the PI controller.
Water Level, and Pressure. Different from Section 3.1, in this Similarly, as the liquid level approaches the set point, the valve
part, the setpoints of pressure, oil level, water level, and PI opening also increases. Water level and oil level reach the
controller parameters are given in Table 3. The openings of the setpoint successively and remain stable. After the oil level and
three outlet valves are controlled by the PI controllers to adjust water level reached a stable level, the gas phase space in the
6082 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Figure 5. Comparison of the simulation results of pressure, oil level, water level, and water volume fraction in oil with the operation data of the
separator.

Table 3. Given Data in the Simulation

pressure oil level water level
setpoint 1150 kPa 1m 1.547 m
KP 8 pa−1 2 m−1 10 m−1
KI (s) 4 50 25

separator remained constant. Then, the pressure decreased in

fluctuation, and at 1 h, it gradually became stable.
4.2. Dynamic Behavior due to the Setpoint Change. In
this section, based on the stable operation status in Section 4.1,
the setpoints of pressure, oil level, and water level are modified,
respectively, to study the dynamic behavior of the separator.
4.2.1. Change of the Pressure Setpoint. Figure 10 shows the
evolution of the separator operating parameters over time after
the pressure setpoint is increased from 1150 to 1350 kPa. After
the pressure setpoint is raised due to the separator operating
pressure being lower than the setpoint under the action of the PI Figure 6. Comparison of the simulation results of pressure, oil level, and
controller, the gas outlet valve opening is quickly adjusted from water level with the operation data of the separator.
70% to completely closed, resulting in an increase in the
6083 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

separator pressure. Subsequently, the separator pressure

gradually stabilized at 1350 kPa. As the pressure in the separator
increases, the pressure before the liquid phase outlet valve
increases. In order to ensure the stability of the oil level and
water level, the opening of the oil valve and water valve is slightly
reduced.
4.2.2. Change of the Oil Level Setpoint. Figure 11 shows the
evolution of the separator operating parameters over time after
the oil level is increased from 1 to 1.2 m. After the oil level
setpoint is increased, due to the oil level being lower than the
setpoint, the oil valve is closed by the controller, resulting in the
enhanced oil level. When the liquid level approaches the
setpoint, the opening of the oil valve gradually increases, and
finally the oil level is stabilized at the setpoint by the controller.
In addition, an increase in the oil level directly leads to a
decrease in the gas space, which in turn leads to an increase in
the separator pressure. Since the separator pressure is higher
Figure 7. Comparison of the simulation results of water volume fraction than the setpoint, the gas valve is opened by the controller, and
in oil with the operation data of the separator. finally the gas pressure is stabilized at the setpoint. During the
dynamic adjustment process, the water level does not fluctuate
significantly. However, in the initial stage, due to the increase of
the separator pressure, the water flow rate out of the separator
through the water valve increases, which in turn leads to the
decrease of the water level in the separator. Under the action of
the regulator, the water level is finally stabilized at the setpoint.
4.2.3. Change of the Water Level Setpoint. Figure 12 shows
the evolution of the separator operating parameters over time
after the water level is changed from 1.547 to 0.9 m. After the
setpoint of the water level is lowered, the water level on the left
side of the weir plate is much higher than the setpoint; thus, the
water valve is fully opened by the controller. As a result, the
water level drops rapidly, causing the liquid level on the left side
of the weir to drop rapidly to 1.036 m, which is much lower than
the height of the weir (1.6 m).
The oil flow to the right side of the weir plate decreases to
zero, and the oil level at the right side of the weir decreases. In
addition, the drop in liquid level leaves vast free space for the gas,
Figure 8. Comparison of the simulation results of valve opening with which in turn leads to a rapid drop in pressure in the separator.
the operation data of the separator.
To compensate for this change, the gas valve and oil valve are

Figure 9. Parameter changes during the separator filling process.

6084 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Figure 10. Parameter changes due to pressure setpoint change from 1150 to 1350 kPa.

Figure 11. Parameter changes due to oil level setpoint change from 1 to 1.2 m.

closed by the controller, and then the liquid level on the left side Unless otherwise specified, the setting of case is the same as that
of the weir plate and the liquid level on the right side of the weir in Section 3.2.
plate gradually rise. Finally, every parameter is stabilized at the The effect of inlet flow on separation efficiency is shown in
setpoint by the controllers. Figure 13. The flow rates are set from 400 to 2000 kmol/h,
4.3. Influencing Factors of Oil−Water Separation respectively, and other setting conditions are the same as in
Efficiency. Oil−water separation is the main function of the Section 3.2. It can be seen from Figure 13 that as the flow rate
separator, and efficient oil−water separation performance is increases, the oil−water separation efficiency in the separator
becomes lower. With the increase in the flow rate, the residence
crucial to the operation of the separator.34 The core principle of
time of the fluid in the separator gradually decreases and the
the gravity separator is that the dispersed phase, which has a
lateral velocity of the fluid increases. The dispersed phase does
density difference with the continuous phase, escapes by floating not have enough time to escape from the continuous phase. It
or sinking within a certain period.35 The longer the residence can be found that with the increase of the flow rate, the slope of
time of the fluid in the separator, the higher the separation the water volume fraction in oil will decrease. Assuming that the
efficiency, and the residence time is affected by parameters such flow rate is infinite, it means that the lateral velocity of the fluid is
as the flow rate and liquid level setpoint.36,37 The effects of infinite, and all the dispersed phase water droplets have no
operating conditions such as inlet flow, water setpoint, and weir chance to settle, resulting in the water volume fraction in oil
height on the separation efficiency are discussed in this section. being infinitely close to the inlet volume fraction.
6085 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Figure 12. Parameter changes due to water level setpoint change from 1.547 to 0.9 m.

The effect of water level on separation efficiency is shown in respectively. The ratio of the water level to weir height is
Figure 14. The water level setpoint is set from 0.7 to 1.58 m, between 93.2 and 27.5%. Other setting conditions are the same
respectively, and other setting conditions are the same as in as in Section 3.2. It can be seen from Figure 15 that with the
Section 3.2. It can be seen from Figure 14 that as the water level increase of the weir height and the unchanged water level, the oil
increases, the oil volume fraction in the water outlet decreases phase space on the left side of the weir plate increases and the
while the water volume fraction in the oil outlet increases. When residence time of the oil phase increases, which directly leads to a
the water level rises, the cross-sectional area of the water phase significant decrease in the water volume fraction in oil outlets. In
on the left side of the weir increases; thus, the flow velocity of the addition, the increase of separation efficiency in the oil phase
water phase decreases, and the residence time increases. In also leads to more free water entering the bulk water phase,
addition, due to the constant height of the weir, the increase in resulting in a slight decrease in the oil volume fraction in the
the volume of the water phase compels the volume of the oil water phase.
phase to decrease, resulting in an increase in the flow velocity of
the oil phase in the separator and a decrease in the residence 5. CONCLUSIONS
time. The increased water level results in a reduced oil volume In this work, a dynamic simulator for three-phase gravity
fraction in water outlets and an increased water volume fraction separators in oil production facilities is proposed. In this
in oil outlets. simulator, the mass conservation equation is established to
The effect of weir height on separation efficiency is shown in calculate the pressure, water level, and oil level in the separator.
Figure 15. The water level setpoint is set as 0.55 m, the oil level The mass balance equation of the dispersed phase is established
setpoint is set as 0.5 m, the weir height is set from 0.59 to 2 m, to calculate the oil−water separation efficiency. The PI
6086 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

outlet valves of oil, gas, and water. The simulator is validated

using field data with the given valve opening and given PI
controller parameters, respectively. The verification results show
that the accuracy of this simulator is within an acceptable range.
Subsequently, the verified simulator is applied to study the
dynamic behavior of the separator filling process, and the effect
of pressure, oil level, and water level setpoint changes on the
separator operating status. A detailed analysis of the changes in
the liquid level, pressure, and the opening of three outlet valves is
presented. The liquid level, pressure, and valve opening in the
separator affect each other, and the change of any parameter will
cause fluctuations in other parameters. Under the action of the
PI controller, the three parameters of the water level, oil level,
and pressure are stabilized in turn. Then, the effects of operating
conditions such as inlet flow, water setpoint, and weir height on
the separation efficiency are discussed. The increase of the inlet
flow reduces the separation efficiency, while the increase of the
Figure 13. Variation of the water volume fraction in the oil outlet and weir height enhances the separation efficiency. The increase of
oil volume fraction in water outlet with the inlet flowrate. the water level increases the water volume fraction in oil while
reducing the oil volume fraction in water.
This simulator can be applied for the design of oil, gas, and
water three-phase separation processes. In addition, through this
simulator, the parameters that are difficult to be measured by
instruments during the operation of the separator can be
calculated, providing technical support for the construction of
the digital twin of the separator.

■ AUTHOR INFORMATION
Corresponding Authors
Shangfei Song − Beijing Key Laboratory of Urban Oil and Gas
Distribution Technology / National Engineering Research
Center of Oil and Gas Pipeline Transportation Safety/ MOE
Key Laboratory of Petroleum Engineering, China University of
Petroleum-Beijing, Changping, Beijing 102249, China;
orcid.org/0000-0001-8275-951X; Email: song.sf@
cup.edu.cn
Figure 14. Variation of the water volume fraction in the oil outlet and Jing Gong − Beijing Key Laboratory of Urban Oil and Gas
the oil volume fraction in the water outlet with the water level.
Distribution Technology / National Engineering Research
Center of Oil and Gas Pipeline Transportation Safety/ MOE
Key Laboratory of Petroleum Engineering, China University of
Petroleum-Beijing, Changping, Beijing 102249, China;
Email: [email protected]
Authors
Xuanzhang Liu − CCCC Second Harbor Consultants Co.,Ltd.
Wuhan, Hubei 430060, China
Chenxuan Li − Beijing Key Laboratory of Urban Oil and Gas
Distribution Technology / National Engineering Research
Center of Oil and Gas Pipeline Transportation Safety/ MOE
Key Laboratory of Petroleum Engineering, China University of
Petroleum-Beijing, Changping, Beijing 102249, China
Zhe Li − Xi’an Jiaotong University the College of Energy and
Power Engineering, Xi’an, Shaanxi 710049, China
Shijia Zhang − CNOOC EnerTech Equipment Technology
Research & Design Center, Tianjin 300452, China
Wei Wu − CNOOC EnerTech Equipment Technology Research
Figure 15. Variation of the water volume fraction in the oil outlet and & Design Center, Tianjin 300452, China
the oil volume fraction in the water outlet with the weir height. Bohui Shi − Beijing Key Laboratory of Urban Oil and Gas
Distribution Technology / National Engineering Research
Center of Oil and Gas Pipeline Transportation Safety/ MOE
controllers are applied to control the water level, oil level, and Key Laboratory of Petroleum Engineering, China University of
pressure in the separator by setting the opening of the three Petroleum-Beijing, Changping, Beijing 102249, China
6087 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

Qi Kang − Research Institute of Tsinghua University in Δt, time step (s)

Shenzhen, Tsinghua University, Shenzhen 518057, China Vlt+1, volume of liquid on the left side of the weir at time t+1
Haihao Wu − Beijing Key Laboratory of Urban Oil and Gas (m3)
Distribution Technology / National Engineering Research Vot+1, volume of oil on the left side of the weir at time t+1 (m3)
Center of Oil and Gas Pipeline Transportation Safety/ MOE Vwt+1, volume of water on the left side of the weir at time t+1
Key Laboratory of Petroleum Engineering, China University of (m3)
Petroleum-Beijing, Changping, Beijing 102249, China V(i,j), volume of the control volume (i,j) (m3)
Complete contact information is available at: ΔVot, difference of oil volume at time t (m3)
https://pubs.acs.org/10.1021/acsomega.2c08267 ΔVwt, difference of water volume at time t (m3)
vk,xt, velocity of the kth particles in the x direction at time t
Notes (m/s)
The authors declare no competing financial interest. vy, velocity of the dispersed phases in the vertical direction
(m/s)

■ ACKNOWLEDGMENTS
This work was supported by the National Natural Science
vk,yt, velocity of the kth particles in the x direction at time t
(m/s)
vo,yt, velocity of the oil particles in the x direction at time t (m/
Foundation of China (grant nos 52104069, U20B6005, and s)
51874323); the National Key Research and Development vw,yt, velocity of the water particles in the x direction at time t
Program of China (grant no: 2022YFC2806203); the China (m/s)
Postdoctoral Science Foundation (grant no: 2022M713460); Xt, provided by the valve manufacturer
and the Science Foundation of China University of Petroleum, Y, gas expansion coefficient
Beijing (grant no: 2462020YXZZ045), all of which are gratefully z, compression factor
acknowledged.

■ NOMENCLATURE
■
α
GREEK SYMBOLS
adjusted vapor mass fraction
Aleft(i,j), area of the left interface of the control volume (i,j) σk(i,j)(t) droplet number density of k particles in the control
(m2) volumes i and j at time t (1/m3)
(i,j)(t)
Abottom(i,j), area of the bottom interface of the control volume Δσk,x difference of the droplet number density of kth
(i,j) (m2) particles in the control volume (i, j) at time t in the x
Aleft,o, cross-sectional area of the oil layer (m2) direction (1/m3)
Aleft,w, cross-sectional area of the water layer (m2) Δσk,y(i,j)(t) difference of the droplet number density of kth
Cv, effective flow coefficient particles in the control volume (i, j) at time t in the y
D, diameter of the separator cylinder (m) direction (1/m3)
d, diameter of droplets (m) μ continuous phase viscosity (Pa·s)
eol(t), error in oil level at time t (m) ρ density (kg/m3)
eP(t), error in pressure at time t (kPa) ρg density of the gas (kg/m3)
ewl(t), error in water level at time t (m) ρl density of the liquid (kg/m3)
Fk, provided by the valve manufacturer Δρ density difference between droplets and the
Fp, pipe geometry parameter continuous phase (kg/m3)
fol(t), oil level controller output value at time t
f P(t), pressure controller output value at time t
f wl(t), water level controller output value at time t
■
HSE
ABBREVIATIONS
health, safety, and environment
G, total mass flowrate through the valve (kg/h) PI proportional integral
g, gravitational acceleration, 9.8 m/s2 API American Petroleum Institute
H, liquid level (m) CFD computational fluid dynamics
Hg, mass fraction of the gas phase at the valve inlet AAD average absolute deviation
hi, surface width of the ellipse head (m) NBP normal boiling point
L, length of the left part of the weir plate in the separator (m)
Mgt, gas molar mass at time t (kg/mol)
Nk(i,j), number of droplets of the kth particle in the control
■ REFERENCES
(1) He, L.; Wen, K.; Gong, J.; Wu, C. A multi-model ensemble digital
volume (i,j) twin solution for real-time unsteady flow state estimation of a pumping
nt, molar amount of gas in the separator at time t (mol) station. ISA Trans. 2022, 126, 242.
Pt+1, pressure in the separator (Pa) (2) Kuang, L.; Liu, H.; Ren, Y.; Luo, K.; Shi, M.; Su, J.; Li, X.
Pinlet, pressure at the inlet of the valve (kPa) Application and development trend of artificial intelligence in
Poutlet, pressure at the outlet of the valve (kPa) petroleum exploration and development. Petrol. Explor. Dev. 2021,
Qin,gt, inlet flowrate of gas at time t (m3/s) 48, 1−14.
Qin,ot, inlet flowrate of oil at time t (m3/s) (3) Sircar, A.; Yadav, K.; Rayavarapu, K.; Bist, N.; Oza, H. Application
of machine learning and artificial intelligence in oil and gas industry. Pet.
Qin,wt, inlet flowrate of water at time t (m3/s) Res. 2021, 6, 379−391.
Qout,gt, outlet flowrate of gas at time t (m3/s) (4) Bimastianto, P.; Khambete, S.; AlSaadi, H.; Couzigou, E.; Al-
Qout,ot, outlet flowrate of oil at time t (m3/s) Marzouqi, A.; Chevallier, B.; Qadir, A.; Pausin, W.; Vallet, L.Digital
Qout,wt, outlet flowrate of water at time t (m3/s) Twin Implementation on Current Development Drilling, Benefits and
R, gas constant, 8.314 J/(mol·K) Way Forward. Abu Dhabi International Petroleum Exhibition &
Tt+1, temperature at time t+1 (K) Conference, 2020; D041S111R001, Vol. Day 4 Thu, November 12, 2020.

6088 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089
ACS Omega http://pubs.acs.org/journal/acsodf Article

(5) Libing, G.; Lyu, Z.; Suo, H.; Liu, X. Market analysis and (24) Hernandez-Martinez, J. L.; Martinez-Ortiz, V.A Method To Size
development trend of petrochemical process simulation software. Gas-Liquid Horizontal Separators Handling Nonstable Multiphase
Chem. Ind. Eng. Prog. 2021, 40, 1−14. Streams. SPE Latin America and Caribbean Petroleum Engineering
(6) Luo, X. M.; He, L. M.; Liu, X. Y.; Lü, Y. L. Influence of separator Conference, 2014.
control on the characteristics of severe slugging flow. Petrol. Sci. 2014, (25) Peng, D.-Y.; Robinson, D. B. A New Two-Constant Equation of
11, 300−307. State. Ind. Eng. Chem. Fundam. 1976, 15, 59−64.
(7) Vileiniskis, M.; Remenyte-Prescott, R.; Rama, D.; Andrews, J. (26) America, I. S. o flow equations for sizing control valves; ISA-
Fault detection and diagnostics of a three-phase separator. J. Loss Prev. 75.01.01-2007(IEC 60534-2Mod); United States of America, 2007.
Process. Ind. 2016, 41, 215−230. (27) Grace, A.; Frawley, P. Experimental parametric equation for the
(8) Backi, C. J.; Grimes, B. A.; Skogestad, S. A Control- and prediction of valve coefficient (Cv) for choke valve trims. Int. J. Pres. Ves.
Estimation-Oriented Gravity Separator Model for Oil and Gas Pip. 2011, 88, 109−118.
Applications Based upon First-Principles. Ind. Eng. Chem. Res. 2018, (28) Shi, G.; Song, S.; Shi, B.; Gong, J.; Chen, D. A new transient
57, 7201−7217. model for hydrate slurry flow in oil-dominated flowlines. J. Petrol. Sci.
(9) Stewart, M.Three-Phase Oil and Water Separators. In Gas-Liquid Eng. 2021, 196, 108003.
And Liquid-Liquid Separators; Stewart, M., Arnold, K., Eds.; Gulf (29) Pacek, A. W. Chem. Eng. Res. Des. 2008, 86, 535−536.
Professional Publishing, 2008; Chapter 4, pp 131−174. (30) Guo, H.-p.; Zou, Z.-j.; Liu, Y.; Chen, J.RANS Simulation of Self-
(10) Arvoh, B. K.; Skeie, N.-O.; Halstensen, M. Estimation of gas/ Propulsion Ship Model Test Coupled With PI Controller. The 27th
liquid and oil/water interface levels in an oil/water/gas separator based International Ocean and Polar Engineering Conference, 2017.
on pressure measurements and regression modelling. Sep. Purif. (31) Zhang, M.; Wu, J. P.; Lin, C.; Xu, K.Rethink the tradeoff between
proportional controller and PI controller. 7th IEEE International
Technol. 2013, 107, 204−210.
Symposium on Computers and Communications, Taormina, Italy, 2002,
(11) Hjertaker, B. T.; Johansen, G.; Jackson, P. Level measurement
2002; pp 57−62.
and control strategies for subsea separators. J. Electron. Imag. 2001, 10,
(32) Malavasi, S.; Somaschi, M. C. Discussion on ″Experimental
679−689.
parametric equation for the prediction of valve coefficient (Cv) for
(12) Yu, W.; Patros, P.; Young, B.; Klinac, E.; Walmsley, T. G. Energy
choke valve trims″ by Andrew Grace and Patrick Frawley, International
digital twin technology for industrial energy management: Classi- Journal of Pressure Vessels and Piping 88 (2011) 109-118. Int. J. Pres.
fication, challenges and future. Renew. Sustain. Energy Rev. 2022, 161, Ves. Pip. 2015, 125, 1−2.
112407. (33) Grace, A.; Frawley, P. Experimental parametric equation for the
(13) Vijayan, V.; Vivekanandan, M.; Venkatesh, R.; Rajaguru, K.; prediction of valve coefficient (Cv) for choke valve trims. Int. J. Pres. Ves.
Godwin Antony, A. G. CFD modeling and analysis of a two-phase vapor Pip. 2011, 88, 109−118.
separator. J. Therm. Anal. Calorim. 2021, 145, 2719−2726. (34) Ahmed, T.; Russell, P. A.; Makwashi, N.; Hamad, F.;
(14) Xue, X. H.; Sun, G. G.; Wan, G. J.; Shi, M. X. Numerical Gooneratne, S. Design and capital cost optimisation of three-phase
simulation of particle concentration in a gas cyclone separator. Petrol. gravity separators. Heliyon 2020, 6, No. e04065.
Sci. 2007, 4, 76−83. (35) Skjefstad, H. S.; Stanko, M. Experimental performance evaluation
(15) Samanta, A.; Modak, N. Influence of operating parameters in and design optimization of a horizontal multi-pipe separator for subsea
particle spreading, separation, and capturing in a hybrid free flow oil-water bulk separation. J. Petrol. Sci. Eng. 2019, 176, 203−219.
magnetophoretic bio-separator. Phys. Fluids 2020, 32, 112012. (36) Liu, S.; Zhang, J.; Wang, L.-S.; Xu, J.-Y. Separation mechanism
(16) Mostafaiyan, M.; Saeb, M. R.; Alorizi, A. E.; Farahani, M. and influential factor study on vane-type-associated petroleum gas
Application of evolutionary computational approach in design of separator. Sep. Purif. Technol. 2020, 250, 117274.
horizontal three-phase gravity separators. J. Petrol. Sci. Eng. 2014, 119, (37) Behin, J.; Aghajari, M. Influence of water level on oil-water
28−35. separation by residence time distribution curves investigations. Sep.
(17) He, X.; Liu, Q.; Zhang, H.; Kui, M.-Q.; Tan, X.-H.; Li, X.-P. Purif. Technol. 2008, 64, 48−55.
Numerical investigation of the performance of moisture separators
based on two-way coupling model by Lagrangian-Eulerian method-
ology. Ann. Nucl. Energy 2019, 124, 407−417.
(18) Sayda, A.; Taylor, J.Modeling and Control of Three-Phase
Gravilty Separators in Oil Production Facilities. American Control
Conference, 2007, New York City, USA, July 11−13, 2007, 2007.
(19) Haekal, M.Control-oriented modeling and simulation of a three-
phase gravity separator and its level loop process dynamics
identification. 2014.
(20) Das, T.; Heggheim, S. J.; Dudek, M.; Verheyleweghen, A.;
Jäschke, J. Optimal Operation of a Subsea Separation System Including
a Coalescence Based Gravity Separator Model and a Produced Water
Treatment Section. Ind. Eng. Chem. Res. 2019, 58, 4168−4185.
(21) Mamudu, O. A.; Igwe, G. J.; Okonkwo, E.Process Design
Evaluation of an Optimum Modular Topping Refinery for Nigeria
Crude Oil Using HYSYS Software. SPE Nigeria Annual International
Conference and Exhibition, 2016; SPE-184342-MS, Vol. All Days.
(22) Ali, Z.; Anuar, A.; Grippo, N.; Ramli, N. E.; Rahim, N.Unifying of
Steady State and Transient Simulations Methodologies for Increasing
Oil Production of Integrated Network of Wells, Pipeline and Topside
Processing Equipment. Abu Dhabi International Petroleum Exhibition &
Conference, 2021; D022S184R002, Vol. Day 2 Tue, Nov 16, 2021.
(23) Becquin, G.; Castane Selga, R.; Abrol, S.; Busboom, A.; Doder,
D.; Jain, A.; Glomsaker, T.; Hyllseth, M.; Ruigrok, C.Subsea Multiphase
Boosting Station System and Controls Optimization. 17th International
Conference on Multiphase Production Technology, 2015; BHR-2015-J3,
Vol. All Days.

6089 https://doi.org/10.1021/acsomega.2c08267
ACS Omega 2023, 8, 6078−6089