1. Introduction
The energy required for the transportation of raw materials and the production of most manufactured goods depends on crude oil [
1]. The explosive increase in the number of automobiles and the demand for production facilities have accelerated dependence on oil [
2,
3]. For that reason, the demand for oil has steadily increased over the past few decades. Until 2030, the share of the oil in the world’s primary energy supply by the source is expected to remain the same [
4,
5].
FPSOs (Floating, Production, Storage, and Offloading) have become the primary production units of crude oil for both shallow and deep seas in recent decades with explosive demands. FPSOs play the role of refining crude oil supplied through risers, storing it in the cargo tanks, and unloading it to the shuttle tankers while floating in the sea. For this reason, FPSO units have inevitably been regarded as one of the leading offshore technology solutions for the oil and gas industries [
6]. A strong demand for oil has consequently led to an increase in the number of FPSOs. As of 2018, compared to 151 FPSOs in 2009, there are about 215 FSPOs worldwide, including those under operation, construction, and orders [
7,
8,
9,
10].
In addition to the increase in the number of FPSOs, the size of production and storage facilities of the FPSOs tends to become larger to meet the growing demand for oil. This leads to an increase in oil production in the deep sea, and it becomes essential to expand the production and storage facilities of FPSOs as shuttle tankers have longer round-trip days [
9].
Figure 1 represents the average daily oil production of the FPSOs that have been delivered or are scheduled to be delivered by MODEC, one of the representative companies in FPSO engineering [
11].
An increase in the oil production at the topside facilities of FPSOs will contain more gases, which leads to a rise in blow-by gas. The blow-by gas is defined as the discharge of the gas from a process component through a liquid outlet [
12,
13]. The blow-by gas occurs when free gas escapes with the liquid, which is caused by the failure of the gas pressure control valve installed in the separator to close [
14]. When oil loading into the cargo tanks from the second stage separating system in a FPSO, there is a chance of delivery of the blow-by gas from the second stage separator to cargo tanks by the failure of the gas pressure control valve installed in the second stage separator. That is, the blow-by gas may be generated due to some causes such as clogging in a control valve installed in the separator after the general dry docking or at the initial start-up of the commissioning operation. This could lead to a dangerous situation for the safety of the FPSO such as the explosion of the cargo tanks [
14]. The role of the blanket gas system, which is consist of an inert gas system and venting system [
12], is becoming increasingly important to safely discharge large amounts of gas, such as blow-by gas, when it goes down to the cargo tanks. In addition, to properly discharge the increased blow-by gas, the current blanket gas system has the disadvantage that the pipe must be large. A novel blanket gas system is needed because of the difficulty of arrangement to the increases in the piping size.
The primary contribution of this paper is the proposal of a blanket gas system configuration and control methods that can safely discharge a large amount of gas, such as the blow-by gas, without reaching the cargo tank design pressure when it enters the cargo tanks. First, we suggest a configuration that can control the pressure of the cargo tanks when the blow-by gas is generated. There have been many studies related to the safety of FPSOs. Some researchers have focused on the study of safety for FPSOs. Jaffee et al. [
15] researched fire and explosion assessments for FPSOs caused by gas and oil leaks. Lei et al. [
16] suggested a novel method for optimizing the FPSO design parameters which affect the strength of the hull girder. Lars [
17] studied the application of the FPSO design for risk analysis and assessment tools. Some studies have been conducted in relation to the efficient operation of the FPSOs. Ali et al. [
18] provided an optimization procedure for topside processes that can reduce fuel consumption and CO
2 emissions. Max Mauro et al. [
19] described waste heat recovery and optimization of power production by an organic Rankine cycle for reducing fuel consumption and CO
2 emissions. Some researchers focused on that how to operate the blanket gas system safely and efficiently. Elisabet et al. [
20] proposed an optimal blanket system of an onshore multiple-tank facility using various connections between each tank, considering economic, environmental, and safety issues. Duddy et al. [
21] provided a solution for a simple and effective way of disposing of inert gas and volatile organic compounds from the FPSO cargo tanks using a cargo vent educator. Béla et al. [
22] introduced a variety of blanket gas control methods. However, research is necessary on a novel blanket gas system that can adequately handle large volumes of gases under abnormal conditions such as blow-by gases. Using piping, which was previously used as a purging line of the inert gas supplied from inert gas generators, is applied for a blanket gas system as dual lines with a vent line when a certain pressure exceeds.
Then, appropriate control methods could be applied to the blanket gas system. The application of the feed-forward control is considered for taking the manipulated variable change before the disturbance, which is the weakness of the feedback control, affects the process. In other words, the feed-forward control is used to ensure adequate control performance of the process when the disturbance is expected to affect the process [
23]. Some researchers have focused on developing advanced controllers for FPSOs. Tannuri et al. [
24] investigated sliding mode control on the dynamic positioning system of the turret moored FPSO. Yuanhui et al. [
25] suggested an improved active disturbance rejection controller for the application of the dynamic positioning system of FPSOs. Hwang et al. [
26] suggested a new type of controller that combines a cascade controller and a feed-forward controller that is often used in industries. However, due to the nature of ships that are difficult to obtain external help for maintenance, the structure of the controller should be simple. In addition, even if the rule of thumb used in the industry is applied for the tuning of the controllers, good control performance should be guaranteed.
Considering these points, we suggest novel configuration and control methods for the blanket gas system in this paper. By dynamic simulations from the HYSYS (Aspen Technology, Bedford, MA, USA) showing good accuracy between actual test results and simulations [
27,
28], we review the performance of the proposed new configuration with a control strategy and compare them with a typical configuration applied in FPSOs. The feasibility of the actual piping arrangement for the blanket gas systems is additionally examined.
3. Design Basis
First, it is necessary to confirm how much the blow-by gas is occurring as crude oil production increases in the FPSOs. Only processes related to the blow-by gas are considered in the HYSYS steady-state simulation. After calculating the flow rate of the blow-by gas, the appropriate pipe size is selected. Then, the review is carried out for the relevant equipment of the blanket gas system used in the dynamic simulation of the HYSYS to evaluate the performance of the control system. Finally, the theoretical background of the equipment applied in the simulations is introduced.
3.1. Blow-By Gas Calculation
The blow-by gas defined as the discharge of the gas from a separation system through a liquid outlet is calculated using the information in
Table 2. In order words, the three-phase fluid flows into the separation system at the same time as shown in
Figure 5 and this fluid flows into the cargo tanks along with the crude oil because of the fail-close of the water and gas control valves. During the calculation of the flow rate of the blow-by gas, it is assumed that there are no heat loss and pressure loss in the cargo tanks and the relevant piping.
Figure 6 shows a schematic diagram used for calculating the flow rate of the blow-by gas. The calculation starts from the state where the primary separation was completed since the processes and equipment from the flow rate of the production head to the primary separation system is superfluous information for making the heat and mass balance of the blow-by gas. During the heat exchange in the crude oil heat exchanger, we try to keep a 2 °C temperature difference between No.2 stream (②) and No.5 stream (⑤) considering a logarithmic mean temperature difference (LMTD) and a minimum temperature approach [
32,
33]. Equation (1) represents the LMTD considered in this calculation [
34].
where
and
are the temperature difference between the two streams at ends 1 and 2.
To prevent damage to the coating applied to the cargo tanks, the temperature of the crude oil should be lowered below a certain level. In addition, the pressure is selected in consideration of the design pressure of the cargo tanks and the pressure loss in the piping. Based on this philosophy explained, the heat and mass balance in
Table 3 is calculated by the HYSYS steady-state simulation.
According to the classification rules, the selection of the maximum flow rate for the blanket gas system should be considered 125% of the largest production rate, the capacity of the inert gas system, blow-by gas from the process plant, or the tank-tank transfer by the cargo pump [
35]. The maximum volumetric flow rate for the blanket gas system is defined as Equation (2).
Referring to
Table 2 and
Table 3, the blow-by gas is the largest value so this is selected as the main case for the design of the blanket gas system. The
obtained by applying Equation (2) is 20,917 m
3/h.
Based on the volumetric flow of the blow-by gas, the pressure of the discharged gas at the end of the vent line should be higher than the atmospheric pressure. If the pressure of the discharged gas is lower than the atmospheric pressure, there may be a problem in the gas discharge. In other words, it is necessary to select pipes of the appropriate size in consideration of the pressure loss in the pipe. The pipe selected by this philosophy is 32-inch for the single line and 20-inch for the dual lines.
3.2. Equipment
The scope of a steady-state simulation for the blow-by calculation and dynamic simulations for confirmation of the control performance are limited from the No.2 separation system to the blanket system. Only the flow rate, compositions, pressure, and temperature of the fluid from the No.1 separation system are used. In these simulations for the steady-state and dynamic, the blanket gas system is simulated using the Peng-Robinson equation of state in the HYSYS.
3.2.1. Plate Heat Exchanger
The plate heat exchanger is applied for calculating the energy balance for the hot and cold fluids in the steady-state simulation. Equation (3) applies to the plate heat exchanger [
36].
where
is mass flow of the hot side,
is mass flow of the cold side,
is enthalpy of inlet of hot side,
is enthalpy of outlet of hot side,
is enthalpy of inlet of cold side,
is enthalpy in outlet of cold side.
The duty of the heat exchanger,
, is used in the simulation. Especially, the overall heat transfer coefficient and area of the heat exchanger are important when the dynamic simulation is applied. The duty of the heat exchanger is defined in Equation (4).
where
is heat transfer coefficient,
is the area of the heat transfer,
is log mean temperature difference.
3.2.2. Control Valve
For valve sizing, the industry standard ANSI/ISA S75.01 is applied. The ANSI/ISA is considered state-of-the-art in valve sizing with more accuracy than other methods [
37]. The general flow equation for a two-phase flow is given in Equation (5).
where
is overall mass flow,
is adjusted vapor mass fraction,
is piping geometry,
is flow coefficient,
is liquid mass density,
is absolute pressure at upstream,
is absolute pressure at downstream,
is the gas expansion factor,
is pressure drop ratio, and
is vapor mass density.
For the piping geometry factor () is considered as 1 if the diameter of the inlet and outlet pipes are the same size.
In the dynamic simulation, a definition of the
is applied for determining the flow coefficient of the control valve [
38]. The
is explained by Equation (6) [
39].
where
is the mass flow of fluid,
is the density of fluid,
is the opening ratio of the control valve,
is inlet pressure of the control valve,
is the outlet pressure of the control valve.
3.2.3. Pump
The pump is used for increasing the pressure of the liquid stream [
40]. After passing through the crude oil heater, the pressure lowers due to friction. At this time, the pressure needs to be pressurized through the pump to pass through the crude oil heat exchanger and crude oil cooler. The centrifugal type is applied for pressurizing the crude oil. For the dynamic simulation, the power and characteristic curve are used. For calculating the actual power of the pump, efficiency is necessary. The ideal power (
) required in the pump is expressed as Equation (7) [
37].
where
is mass flow,
is density,
is the inlet pressure of pump,
is the outlet pressure of the pump.
In addition, the efficiency (
) of the pump is one of the important factors for dynamic simulation. The actual power required for the pump varies depending on the efficiency. The actual power (
) required in the pump and efficiency are defined as Equations (8) and (9).
3.2.4. Controllers
During the dynamic simulation, two controllers are used. One is the proportional-integral (PI) controller. A manipulated variable (
MV) is calculated by the difference between a process variable (
PV) and a set value (
SV) [
41]. Based on this philosophy, the PI controller makes a signal to adjust the control valve in the process.
A PI controller is the most common structure used in the process control industries. However, derivative action is excluded in this simulation because it may make some derivative kicks sometimes [
42]. The controller output signal with the parallel PI structure is represented in Equation (10). For the error, the value is shown in Equation (11).
The other controller is the feed-forward controller. The feed-forward control is different from the feedback control in that the manipulated variable is adjusted beforehand when the control system is felt for load changes or disturbances [
43]. The transfer function of the feed-forward controller is shown in Equation (12) [
37].
where
is feed-forward controller input,
corresponds to feed-forward controller output,
is delay time,
is lead time constant, and
is lag time constant.
The input,
, of the feed-forward controller is expressed in Equation (13).
where
is the measured value of disturbance variable,
is the minimum value of the disturbance variable,
is the maximum value of disturbance variable.
The controller output is combined with the response to the feed-forward control and PI control outputs. The control schematic diagram with the feed-forward controller from the vent header and PI controller from the purge header is indicated in
Figure 7.
4. Simulation Results and Discussion
The main purpose of the performance analysis is to confirm how efficiently the blow-by gas can be processible and the proposed control scheme can properly adjust the blanket system compared to the existing system (Case A).
Then, the feasibility study on the piping arrangement is carried out. Because the FPSO has the crude oil lines supplied from the topside, the fire-fighting lines, and various lines so that installing 32 inches of the large pipes is a big challenge to the engineers.
4.1. Performance Analysis
The performance analysis is carried out under the condition that the inert gas is provided to prevent an explosion in the cargo tanks whilst the crude oil produced from the topside modules is supplied to the cargo tanks of the FPSO. In these conditions, we consider the situation in which the blow-by gas is generated as a problem occurs in the separation system.
The dynamic simulations are proceeded to refer to the described operation conditions. From the dynamic simulation, the actual vented volumetric flow, the inert gas volumetric flow supplied to the cargo tanks, and the pressure inside the cargo tanks are obtained, then it is compared and analyzed for making the performance analysis.
4.1.1. Simulation Results for Case A
Case A shows the existing system, and the inert gas pressure is controlled by the CV1 and CV2 depicted in
Figure 3. When the pressure of the cargo tanks is increased by the blow-by gas, so it is common to control the inert gas pressure using the CV1 and CV2 rather than immediately turning off the inert gas generator due to the large amounts of the inert gas supplied. Despite regulating the inert gas pressure by the CV1 and CV2, the PV valve opens to control the cargo tank pressure when the set value of the PV valve is reached. Based on these conditions,
Figure 8 represents the result of how much the gas is vented for the pressure control inside the cargo tanks.
As the PV valve momentarily opens, the vented gas is so large at the initial time. Then, the volumetric flow rate is lowered to about 18,700 m3/h and settles over time at approximate 24,300 m3/h.
Because the outgoing flow rate is greater than the incoming flow rate at the initial time, the pressure drops to 111.5 kPa at the discharge side of the cargo tanks, and the pressure starts to rise again when the flow rate to the atmosphere via the vent decreases. After the pressure inside cargo tanks rises to 118.5 kPa, the pressure gradually decreases over the time. It is explained in
Figure 9.
As the pressure in the cargo tanks drops due to the immediate opening of the PV valve, the inert gas is supplied for compensating the pressure inside the cargo tanks. A flow of up to 1230 m
3/h is provided to the cargo tanks. As the vent flow rate decreases compared to the blow-by gas flow rate, the pressure rises. As the pressure rises again inside the cargo tanks, the flow rate of the inert gas injected into the cargo tanks is reduced.
Figure 10 shows the flow rate of the inert gas to the cargo tanks.
Except for the initial opening of the PV valve, the stable flow rate of the 32 inches of the pipe is 24,300 m3/h, and it is confirmed that it operates within the design pressure of 125 kPa of the cargo tank. However, it is checked that a rather large amount of inert gas is supplied to the cargo tanks due to the problem of the blanket system piping configurations and control.
4.1.2. Simulation Results for Case B
Unlike Case A, this configuration includes a separate purge header for the venting purpose. When the pressure transmitter in
Figure 4 feels the over-pressure, the CV1 and CV2 are operated according to the control signal of the PC1 and PC2.
Figure 11 shows the flow rate handled by each header when the blow-by gas is occurring. Over time, the purge header and vent header are under a stable condition with a flow rate of 12,800 m
3/h and 14,500 m
3/h. The total flow rate that can deal with the two headers is 27,300 m
3/h.
When the PV1 and PV2 installed in two headers are suddenly opened, the drop in the pressure is relatively small compared to Case A with 32-inch piping because the drop in the vented flow rate is not large.
Figure 12 depicts how the tank discharge pressure is changed.
As the purge header is used for the gas venting, a relatively small flow rate of the inert gas compared to Case A is allowed into the cargo tanks. In addition, the flow rate of the inert gas injected into the cargo tanks is small owing to the small pressure drop of the cargo tanks. The information on the flow rate of the supplied inert gas is depicted in
Figure 13.
Compared with the existing system (Case A), the pressure can be stabilized relatively quickly over the time because the discharge gas flow rate is large. However, to guarantee a better performance, faster response to the disturbance of the control system of the purge header is essential.
4.1.3. Simulation Results for Case C
The biggest difference between Case B and Case C is whether the feed-forward control using the pressure transmitter installed in the vent header is applied. In normal conditions, the PI controllers are used for adjusting the pressure inside cargo tanks by the pressure transmitter installed in the purge header, but the feed-forward controllers are applied to control the CV1 and CV2 by the transmitter at the vent header when a disturbance such as the blow-by gas occurs.
When the blow-by gas occurs, more gas flow is released at the initial time compared to the Case B whilst swiftly adjusting the CV1 and CV2 by the feed-forward controllers. This occurs because the purge header is used for venting whilst quickly shutting off the inert gas supply to the cargo tanks. In addition, the fluctuation of the gas flow rate largely occurs at the initial time by the same reason.
Figure 14 shows how much each header can allow for releasing the gas.
At the beginning time of the opening of the CV1 and CV2 by the feed-forward controllers, a large amount of the gas is released, causing the tank pressure to significantly drop compared to Case B. When the tank pressure drops, the inert gas is quickly supplied through the purge header into the cargo tanks by the feed-forward controllers. That is, there is a restriction in using the purge header for venting. As a result, the pressure is risen again. When the pressure rises again, since the supply of inert gas is stopped and purge header is used again for venting so that the discharged gas flow rate becomes a stable condition over the time. The changes in the pressure inside the cargo tanks are able to confirm in
Figure 15.
The supply of the inert gas into the cargo tanks is swiftly adjusted by the feed-forward control. As a result, about 1100 m
3/h of the inert gas is provided to the cargo tanks for a short time when the pressure is dropped to about 113 kPa at about 20 s. After that, as the pressure rises, the inert gas is not supplied to the cargo tanks. The trend of the supplied flow rate of the inert gas is available in
Figure 16.
In Case B, the pressure inside the cargo tanks is 118.4 kPa at 3500 s. With a faster response by the feed-forward controller to the disturbance for the purge header in Case C, it shows 117.5 kPa at 3500 s. Since one of the dual lines is also used as an inert gas supply line, it is confirmed that the pressure of the cargo tanks is swiftly and stably controlled by the combination of the PI controller and feed-forward controllers.
Table 4 shows a summary for comparison of simulation results. For settling time, it is measured based on the pressure at which the PV valve closes, 117 kPa.
4.2. Piping Arrangement
In relation to the piping arrangement, it is assumed that the installation spaces of the riser platform and various equipment need a width of 5 m and a passage of 1 m between each line. The tank gauging system is requested to install for each cargo tank. It is assumed that the tank gauging system needs a total width of 1 m for the entire cargo tanks. In addition, although not depicted in
Figure 17, it is assumed that water and gas injection lines vertically installed to be connected to the riser platform are requested for a total of 10 m width. These explanations are summarized as follows.
- -
Riser platform and equipment: 10 m
- -
Total passage width: 7 m
- -
Tank gauging system for entire cargo tanks: 1 m
- -
Water and gas injection lines: 10 m
When all the values assumed above are added up, a total width of 28 m is required. A width of 30 m is remained for the piping arrangement when subtracted from the width of the ship described in
Table 2.
In most cases, the fire-fighting lines, the topside lines to cargo tanks, and the offloading lines are large, followed by the blanketing system lines and the small lines related to the crude oil. However, the quantity of the piping is a lot when the piping line size is small, so that a similar width is required compared to the large piping. In other words, each piping line is possible for using a width of 5 m.
Considering the piping stress, it is impossible to arrange the piping in a straight line. Therefore, it has no choice but to arrange it in a bending shape using an elbow. If a 32-inch elbow is used, the H shown in
Figure 18 is about 1.63 m [
44]. That is, a total height of 3.26 m of the elbow is required when two elbows are applied. Considering the pressure loss and stress for the piping, a straight pipe may need to connect it to the elbows. Taking these considerations into account, a width of 5 m is required to install a pipe excluding the pipe supports. In the case of a 20-inch elbow, the H is about 1.02 m [
44]. A width of 3.17 m including two elbows is required to install a 20-inch pipe without the pipe supports. From these points, it is advantageous to use dual 20-inch pipes rather than one 32-inch pipe considering the installation of the pipe supports and the handleable area of the deluge valves for the fire-fighting system.
In the case of the pipe weight, it is about 57.8 tonnes, which shows similar results, when it is applied to the total length of 367.9 m [
45]. There is no difference in terms of capital expenditure (CAPEX) because the price of the pipe is proportional to its weight. However, this is limited to the costs of piping used for vents. In the existing system, the purge header is used only for the inert gas supply, and if it is included, the newly proposed system has an advantage in terms of CAPEX.