OTC 21776
Application of CFD to a Flare Drum Carryover Issue
M.J. Lewis, CD-adapco; D. Cooper, CD-adapco
Copyright 2011, Offshore Technology Conference
This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, USA, 2–5 May 2011.
This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents of the paper have not been
reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect any position of the Offshore Technology Conference, its
officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the written consent of the Offshore Technology Conference is prohibited. Permission to
reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.
Abstract
Process diagnostics suggested there was an oil carry-over problem with a Flare Knock-Out Drum; the liquid level inside the
vessel was observed to drop during flaring events. Further indicators of this came from a temperature increase in the vessel.
Oil carry-over can lead to poor combustion and resultant damage to the flare tip.
CD-adapco worked together with a Client to find a solution to this oil carry-over problem. STAR-CCM+, the general
purpose CFD software, was used to identify the cause of the carry-over; suggest potential design solutions; quantify
improvements of the best designs; and to provide design information for use to size and manufacture the modifications.
The simulations showed the inlet device to be the cause of the problem, directing the incoming flow onto the liquid surface
inside the drum resulting in liquid being re-entrained into the gas stream and carried over to the flare tip. The inlet device was
redesigned with the aid of CFD, without reducing the residence time inside the vessel, such that the inlet gas stream was
distributed throughout the drum. The result was a dramatic reduction in liquid re-entrainment.
Introduction
The flare system on production facilities is used to safely dispose of unwanted hydrocarbon fluids which arise primarily during
emergencies such as to relieve excess pressure or to remove hydrocarbon inventory from the production systems. It can also
be used to safely dispose of excess gas during start-up or shut down of the facility. The fluids routed to the flare system are
largely composed of vapour but liquids can be present in the stream due to multiphase relief, liquid entrainment or retrograde
condensation due to the temperature and pressure reduction the fluids undergo. Flare knock out drums are designed to
disengage these entrained liquids from hydrocarbon fluid streams routed to the flare disposal system. This is to prevent liquid
being carried over to the flare tip where it won't be fully combusted and can hence pose a hazard as "burning rain" or cause
pollution as the liquids fall onto the ground or water.
The internal design of a flare knock out drum is kept as simple as possible to avoid the potential for blockages due to
internals failure as this would incapacitate the flare system and prevent safe disposal of the fluids. A liquid level is normally
kept in the flare knock out drum to prevent vapour break out in the liquid disposal pump suction or gas breakthrough to a
lower pressure system. The flare knock out drum is usually equipped with level and temperature instrumentation to monitor
the system and to take executive action to shut down total production to prevent safe operating limits being exceeded.
Under normal flaring operations, the liquid level would be expected to rise in the flare knock out drum. For one platform
the liquid level was observed to fall markedly in the flare knock out drum each time there was a significant flaring event. This
liquid wasn't fully combusted in the flare tip, and consequently a sheen could be observed on the sea surface following a
flaring event confirming that the phenomenon wasn't an artifact of the level measurement becoming unreliable due to a
disturbed liquid. A typical event is illustrated in Figure 1 where the liquid level, flaring rate and liquid temperature for the
flare knock out drum are plotted. This information is quite limited in value in that it only shows what is happening but does
not provide an explanation as to why the liquid level should decrease. Several mechanisms were postulated for this including
liquid re-entrainment due to poor fluid flow distribution and evaporation of the liquid. Diagnostics such as neutron backscatter
or gamma scanning techniques of the flare knock out drum would not provide any further insight into what was causing the
drop in liquid level, rather this would probably only confirm the observations. In order to find a solution to the problem it was
decided to utilize Computational Fluid Dynamics (CFD) to gain insight into the flow distribution in the vessel.
Initially, the as is design was simulated using a typical flaring event. The insight gained into the causes in the drop of
liquid level was used to propose and rank several alternative designs. Finally, the most appropriate design was quantitatively
compared with the original design
2 OTC 21776
Figure 1: Process diagnostics of the flare drum. The flaring rate is shown at the bottom in blue. The liquid level is shown at the top
in red. The vessel temperature is shown at the top in yellow.
CFD methodology
STAR-CCM+, the general purpose CFD software, was used for all simulations. STAR-CCM+ employs finite volume
techniques for solving the conservation equations of mass, momentum, energy and turbulence [1].
The volume mesh is generated using arbitrary polyhedra, as validated for flow and thermal solutions [2, 3]. This type of
meshing technique affords the best comprise between numerical accuracy and robustness of meshing on complex geometrical
shapes.
Turbulence was modelled using the standard k-ε model [4, 5], where the near-wall behaviour adapts to a y+ independent
(or all y+) treatment [6].
The oil and gas phase behavior is captured with use of the Volume of Fluid (VOF) algorithm. This is implemented as a
volume tracking method, where the liquid phase is represented by a passive scalar α, solved using a high order differencing
scheme to resolve the gas/liquid interface. The gas density was assumed to follow ideal gas law behavior. Then the properties
of the mixture (density, thermal conductivity etc) are weighted averages on α of the single components properties.
Problem Investigation
The vessel is shown in Figure 2; it consists of;
• the vessel.
• an inlet duct with an inlet device.
• an outlet duct to the flare.
• a liquid outlet including vortex breaker. N.B. for all simulations described herein this outlet is closed.
Figure 2: The flare drum geometrical features. The inlet ducting and simple inlet device can be seen on the top right. The outlet
ducting to the flare is shown on the top left. The liquid outlet including a vortex breaker is shown at the bottom.
OTC 21776 3
The inlet device shown in Figure 2 is the original design of the inlet device. This is a flat plate angled down towards the
lower surface of the flare drum.
Operating and Boundary Conditions
Simulations for two operating conditions were undertaken; these are presented in Table 1. Both of these operating conditions
showed the same behavior, and consequently, only Case 1 with a maximum flaring gas mass flow rate of 137,000 kg/h is
reported here.
Maximum Flaring Gas Flow Rate Gas Temperature
o
Operating Condition 1 137,000 kg/h 50 C
o
Operating Condition 2 342,000 kg/h 35 C
Table 1: Operating conditions undertaken
An inlet mass flow rate was applied at the entrance to the inlet duct shown in Figure 2. The gas mass flow rate was ramped up
from zero to the maximum flow rate over a period of 15s to 25s (depending on the maximum flow rate), and then the
maximum flow rate was maintained for 2 minutes.
An outlet pressure boundary was applied at the exit of the outlet duct shown in Figure 2. In reality the flow restriction of
the flare tip creates a back pressure in the tank which increases with flow rate. Therefore, the pressure was initially set to
atmospheric and increased according to the pressure/flow rate curve of the flare tip (see Figure 3). Note that this causes a
gradual pressurization of the entire tank, which increases the gas density and thus (for fixed inlet mass flow conditions)
reduces inlet velocity.
Figure 3: Flare Tip Flow vs. Pressure Curve
Original Design Simulation Results
There were obvious pointers to the design of the inlet device from the initial simulations. The flaring gas flow enters the
vessel through the inlet duct. As the inlet flow interacts with the inlet device it is directed onto the liquid surface. The
combination of the high mass flow rate and the high velocity of the gas impinging on the liquid surface causes the surface to
be disturbed. A standing wave is setup by this directly below the inlet device. Subsequently, droplets and ligatures are
stripped from the liquid and transported through up to the gas outlet to the flare. This process is shown in Figures 4 to 6.
A B C
Figure 4: Velocity contour plots on a vertical slice through the centre line of the vessel with the original design of the inlet device. A
shows velocity contours at 5s, B at 75s and C at 120s.
4 OTC 21776
Figure 4 shows three contour plots of velocity contours on a vertical slice through the centerline of the vessel at 3 instances
during the simulation. Three instances are chosen to show the transient behavior during the start-up period of the flaring event
and the behaviour during the maximum flow rate. The higher velocities can be seen in the inlet ducting, and it is clear that the
inlet device directs the gas down onto the liquid surface.
Figure 5 shows three iso-surface plots of the interface between the oil and the gas in the vessel at 3 instances during the
simulation. During the initial ramping up of the gas flow rate, it can be seen, in Figure 5A, that there is little disturbance of the
liquid surface. However, once the flow rate reaches the maximum flow rate the surface is disturbed and a standing wave is
clearly evident below the inlet device.
A B C
Figure 5: Iso-surface plots showing the interface between the oil and the gas in the vessel with the original design of the inlet device.
A shows the interface at 5s, B at 75s and C at 120s
Figure 6 shows three contour plots of density on a vertical plane through the centreline of the vessel at 3 instances during
the simulation. The oil is shown in red as the denser fluid; the gas is shown in blue as the less dense fluid. This reinforces the
observation of the disturbance to the liquid surface, but it also shows the stripping of liquid from the surface as demonstrated
in Figure 6C.
A B C
Figure 6: Density contour plots on a vertical plane through the centre line of the vessel with the original design of the inlet device.
The oil is shown in red as the denser fluid; the gas is shown in blue as the less dense fluid. A shows density contours at 5s, B at 75s
and C at 120s
Solution Investigation
It became clear that a solution to the liquid re-entrainment problem would require a redesign of the inlet device of the flare
drum. Re-designing the inlet device focussed on distributing the incoming gas flow around the vessel without significant short
circuiting, realized six possible designs for the inlet device. These are shown in Figure 7 and can be summarized as;
• Flat plate perpendicular to inlet ducting
• Flat plate directed upwards
• Square saucer perpendicular to inlet ducting
• Vertical concave half pipe.
• Horizontal concave half pipe.
• Convex plate
OTC 21776 5
A B C
D E F
Figure 7: Geometrical variations of possible inlet devices. A shows a flat plate perpendicular to the inlet ducting. B shows a flat plate
angled towards the upper surface of the vessel. C shows a saucer perpendicular to the inlet ducting D shows a vertical half pipe. E
shows a horizontal half pipe. F shows a convex plate.
In order to analyze the suitability of each of these alternatives quickly and efficiently, a series of single-phase, steady state
simulations were undertaken. This provided qualitative comparisons of the devices in a realistic turnaround time. These
results can be seen in Figure 8 which shows streamline plots for each alternative inlet device. For the;
• Flat plate perpendicular to inlet ducting: the flow remains directed downwards onto where the liquid surface
would be (Figure 8A).
• Flat plate directed upwards: although more flow is directed upwards, there is significant flow down towards where
the liquid surface would be. The flow directed upwards moves directly to the gas outlet increasing short circuiting
(Figure 8B).
• Square saucer perpendicular to inlet ducting: The flow down towards where the liquid surface would be is greatly
reduced, but there is a large increase in the flow along the top of the vessel directly towards the gas outlet greatly
increasing short circuiting (Figure 8C).
• Vertical concave half pipe: flow is split that it is directly upwards and downwards. This not only does not reduce
the impingement on the liquid surface, but would increase short circuiting (Figure 8D).
• Horizontal concave half pipe: the flow is split horizontally towards the outer surface of the vessel. It then follows
the contours of the vessel down to where the liquid surface would be (Figure 8E).
• Convex plate: flow is directed onto the rear surface of the vessel, which channels the flow down towards where
the liquid surface would be (Figure 8F).
The best compromise between ease of construction and reduction in the jetting down on the surface was the horizontal
concave half pipe.
6 OTC 21776
A B C
D E F
Figure 8: Velocity streamlines within the vessel. A shows the streamlines for the flat plate perpendicular to the inlet duct. B shows
the streamlines for the flat plate angled towards the upper surface of the vessel. C shows the streamlines for the saucer
perpendicular to the inlet duct. D shows streamlines for the vertical half pipe. E shows streamlines for the horizontal half pipe. F
shows the streamlines for the convex plate.
Design Verification
In order to quantify the expected improvement of the horizontal concave half pipe design over the original design, a direct
comparison simulation was undertaken using the approach outlined for the initial simulation. This resulted in the following.
Figure 9 shows three iso-surface plots of the interface between the oil and the gas in the vessel at 3 instances during the
simulation. Figure 10 shows three contour plots of density on a vertical plane through the centreline of the vessel at 3
instances during the simulation.
The flaring gas is now split horizontally towards the outer surface of the vessel. It then follows the contours of the vessel
down to the liquid surface. The flow is distributed over a wider region than was seen for the original design, and
consequently, the disturbance of the liquid surface it is significantly reduced. This process is shown in Figures 9 to 5 with
comparison to Figures 10 to 6.
A B C
Figure 9: Iso-surface plots showing the interface between the oil and the gas in the vessel with the original design of the inlet device.
A shows the interface at 5s, B at 75s and C at 120s.
OTC 21776 7
A B C
Figure 10: Density contour plots on a vertical plane through the centre line of the vessel with the new design of the inlet device. The
oil is shown in red as the denser fluid; the gas is shown in blue as the less dense fluid. A shows density contours at 5s, B at 75s and
C at 120s
Figure 11 shows the volumetric flow rate of oil out of the gas outlet of the vessel. Figure 12 shows the total volume of oil
in the vessel. It is clear from Figures 11 and 12 that there is a dramatic reduction in liquid re-entrainment and subsequent carry
over for the new design of the inlet device compared to the original design.
Further to this, it was thought that further improvements could be made by perforating the plates of the inlet devices.
These results are also shown in Figures 11 and 12. When comparing a given inlet device with and without perforations there is
a small improvement. This improvement is a result of a further distribution of the incoming gas flow over the vessel.
Manufacturing of the plates with perforations would add complications and cost. Consequently, the improvement gained by
adding the perforations was not seen to balance the increase in cost, and these options were not taken further.
Figure 11: Volume flow rate of the oil through the flare outlet. The original design of the inlet device is shown in black. The new
design of the inlet device is shown in red. The original design of the inlet device with perforations is shown in blue. The new design
of the inlet device with perforations is shown in pink.
8 OTC 21776
Figure 12: Total volume of oil in the vessel. The original design of the inlet device is shown in black. The new design of the inlet
device is shown in red. The original design of the inlet device with perforations is shown in blue. The new design of the inlet device
with perforations is shown in pink.
Conclusions
Flare knock out drums are designed to disengage entrained liquids from hydrocarbon fluid streams routed to the flare disposal
system. This is to prevent liquid being carried over to the flare tip where it won't be fully combusted and can hence pose a
hazard as "burning rain" or cause pollution as the liquids fall onto the ground or water.
Under normal flaring operations, the liquid level would be expected to rise in the flare knock out drum. Process diagnostics
suggested there was an oil carry-over problem through a drum; the liquid level inside the vessel was observed to drop during
flaring events. Further indicators of this came from a temperature increase in the vessel. This liquid carried over wasn't fully
combusted in the flare tip, and consequently a sheen could be observed on the sea surface following a flaring event confirming
that the phenomenon wasn't an artifact of the level measurement becoming unreliable due to a disturbed liquid. In order to
find a solution to the problem it was decided to utilize Computational Fluid Dynamics (CFD) to gain insight into the flow
distribution in the vessel.
CD-adapco worked together with a Client to find a solution to this oil carry-over problem. STAR-CCM+, the general
purpose CFD software, was used to identify the cause of the carry-over; suggest potential design solutions; quantify
improvements of the best designs; and to provide design information for use to size and manufacture the modifications.
The simulations showed the inlet device to be the cause of the problem, directing the incoming flow onto the liquid surface
inside the drum resulting in liquid being re-entrained into the gas stream and carried over to the flare tip. The inlet device was
redesigned, without reducing the residence time inside the vessel, such that the inlet gas stream was distributed throughout the
drum. The result was a dramatic reduction in liquid re-entrainment.
References
[1] STAR-CCM+, release version 4.06, CD-adapco, November 2009, www.cd-adapco.com,
[2] M. Peric, “Flow simulation using control volumes of arbitrary Polyhedral shape”, ERCOFTAC Bulletin 62, September 2004
[3] I. Demirdžić, S. Muzaferija, “Numerical method for coupled fluid flow, heat transfer and stress analysis using unstructured moving
meshes with cells of arbitrary topology”, Comput. Methods Appl. Mech. Engrg., 125: 235-255, 1995.
[4] Jones, W.P. and Lander, B.E. 1972. “The Prediction of Laminarization with a Two-Equation Model of Turbulence”, Int. J. Heat and
Mass Transfer, 15, pp. 301-314.
[5] Launder, B.E. and Sharma, B.I. 1974. “Application of the Energy Dissipation Model of Turbulence to the Calculation of Flow Near a
Spinning Disc”, Letter in Heat and Mass Transfer, vol. 1, no. 2, pp 131-138.
[6] Reichardt, H. 1951. “Vollstaendige Darstellung der turbulenten Geschwindigkeitsverteilung in glatten Leitungen”, Z. Angew. Math.
Mech, 31(7), pp. 208-219.
