SPE 84917

Gas-lift Instability Resulted Production Loss and Its Remedy by Feedback Control:
Dynamical Simulation Results
Bin Hu, SPE, Michael Golan, SPE, Norwegian University of Science and Technology

Copyright 2003, Society of Petroleum Engineers Inc.

such as hydrodynamic slugging, which is not considered as an
This paper was prepared for presentation at the SPE International Improved Oil Recovery operation problem in most situations except that it might be
Conference in Asia Pacific held in Kuala Lumpur, Malaysia, 20–21 October 2003.
amplified by other factors. Normally, it is the marcroscopic
This paper was selected for presentation by an SPE Program Committee following review of
information contained in an abstract submitted by the author(s). Contents of the paper, as
instability that results in the severe unstable flow in the wells.
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to For many years, it has been observed that continueous gas-lift
correction by the author(s). The material, as presented, does not necessarily reflect any
position of the Society of Petroleum Engineers, its officers, or members. Papers presented at wells can be seriously unstable and sometimes even behave as
SPE meetings are subject to publication review by Editorial Committees of the Society of
Petroleum Engineers. Electronic reproduction, distribution, or storage of any part of this paper
an intermittent gas-lift due to instabilities with systematic
for commercial purposes without the written consent of the Society of Petroleum Engineers is background. Large fluctuation in the well flow rate can result
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 in poor separation, limit the production capacity and cause
acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O.
Box 833836, Richardson, TX 75083-3836 U.S.A., fax 01-972-952-9435.
flaring and shutdown.
Besides the operation problems, production loss is another
important issue for unstable gas-lift wells. The field tail
Abstract production will be prolonged due to the reduced production
Gas-lift wells can be unstable due to various reasons. The rate since it will take more time to get the same recovery. This
fluctuating, and sometimes chaotic unstable production will tremendously increase the cost of operation. The
behavior affects many offshore fields, particularly at their concentraion of this paper is to investigate the gas-lift
decline stages. The hazardousness of the flucuation to instability resulted production loss since this has not been
operation safety and smoothness has been awared. But the widely addressed and caused enough attention. Particularly,
production reduction due to the instabilities has not been when the downstream receiving facitlity and process can
widely addressed. In this paper, a commercial available handle the flow fluctuation, one seldom has the motivation to
dynamical multiphase flow simulator is used to conduct deal with it. Therefore, the production reduction is ignored.
investigation on the production loss. Two types of This paper emphasizes the notion that production loss is the
instabilities, casing heading and density wave instability, norm for oscillating wells and the instabilities have to be
which both can result in production loss are simulated. Two seriously treated even when they do not cause any operation
hypothetic wells, abstracted from typical North Sea gas-lift problem. It also demonstrates that the pragmatic method such
wells, are built within the simulator. Parametric simulation as static choking performed by skilled operators can only
study is then performed by changing the well settings. The stabilize the well at a cost of large production reduction.
average production rate of the unstable well can be obtained However, dynamical choking based on feedback control can
from the simulation and compared with the steady state not only stabilize the well but also save part of the production
prediction. The difference of the two results gives the loss. Thus, the advantage of applying feedback control to
production loss due to instability. Furthermore, the paper suitable unstable wells is promoted.
presents simulation results of applying feedback control to the
unstable wells. The production loss is reduced as an effect of Gas-lift instabilities. Two systematic instability mechanisms
stabilization by feedback control. that can result in serious unstable phonomena are discussed in
this paper. One is casing heading, in which the gas injection
Introduction through gas-lift valve is non-critical and annulus flow
Gas-lift is one of the most widely used artificial lift method in dynamics is involved in the instability. Another mechanism is
oil production. Economically optimised gas-lift design called density wave instability, in which the gas injection is
requires that gas-lift wells be operated at the up-slope of their critical and thus flow in annulus is isolated from any
lift performance relationship curve[7]. This leads the gravity instability taking place in the tubing. Since this instability has
effect becomes the dominant factor of gas and liquid two- never been discussed in prtroleum gas-lift application, we
phase flow in the tubing. Two-phase vertical flow under borrow this name from the similar phenomenon observed in
gravity domination is often unstable. This is particularly true airlift pumps that are often used in the mining industry[8][12].
in the gas-lift wells. The instabilities of gas-lift can either be Casing heading phenomenon has been observed and
marcroscopic or microscopic. By marcroscopic, we mean that studied for some years and its mechnism is now quite clear. It
the instability has a systematic background. And by needs two necessary conditions to occure. One is that the two-
microscopic on the other hand, we refer to the local instability
2 SPE 84917

phase flow in the tubing is gravity dominant. The other relies increase in the pressure drop due to increased mixture density
on the large volume of compressible gas in the annulus. If the will result in the reduction of the liquid inflow and thus the
port size of the gas-lift valve is not small enough, any pressure mixture density, and vise versa. But this self-controlling
purterbation in the tubing will be promoted by the changed mechanism is more or less delayed due to the out-of-phase
gas injection rate accordingly. Assuming that a negative effect between the well influx and the total pressur drop along
pressure purterbation from its steady state is arbitrarily applied the tubing. To a certain level, it breaks down and the well
in the tubing side, then gas injection through gas-lift valve will becomes unstable. The occurrence and characterization of
be increased due to the increased pressure difference across density wave instability will be discussed in authours’
the valve. The tubing pressure then will be further decreased another publication.
due to the increased gas injection since the well is operated at
the up-slope of its lift performance curve. This trend will Feedback Control. Recently, feedback control has emerged
continue until that the annulus pressure is low enough and gas as a cost-effective option for stabilizing gas-lift wells[9][13]. For
injection starts to decrease. The pressure in the annulus is the wells that cannot be stabilized by installing NOVA valve,
reduced due to the out-of-balance between the gas flowing or are technically and economically forbidden to install
into annulus through the casing choke and the gas flowing out NOVA valve, feedback control is an effective method for
of annulus through gas-lift valve. Unfortunately, this slowing stabilizing wells. The paper will even demonstrate that the
down effect of annulus sometimes can only result in limit production loss is reduced as an effect of stabilization by
cycle rather than stabilized flow. The reason is that the feedback control.
annulus response is much delayed due to its huge volume and
the compressibility of the gas. It behaves as a buffer tank in Dynamical Simulation Study
this situation. Using dynamical simulator to investigate the general principle
The hazardousness of casing heading to operation safety of gas-lift instabilities and production loss is preferable since
and smoothness has been awared. The previous investigations it is the most cost-effective way of doing the investigation
on casing heading mainly concentrated on its characterization comparing with well experiments and laboratory experiments.
and developing simple stability criteria[1][3][4][6][10][11][18]. The With the development of new dynamical multiphase flow
purpose of obtaining the criteria was to avoid casing heading simulator, this becomes more and more realistic.
at the design stage of gas-lift. Due to the complexity of OLGA2000© is selected as our simulation tool. The
gas/oil/water three-phase flow and their PVT properties, even objective of developing OLGA was to simulate slow transient
the latest developed criterion is not convincing when multiphase pipe flow such as terrain slugging, pipeline startup
comparing with field data and the performance of dynamical and shutin, flow rate variation and pigging. The fisrt version
numerical simulator[16]. In fact, predicting gas-lift instability at of OLGA was working in 1983, but the main development
its design stage has no too much meaning since the well was carried out later by a joint research program, involving
conditions are always changing. This consideration led into two research institutions and several oil companies. OLGA is
the development of new type of gas-lift valve based on nozzle based on an extended two-fluid model, accounting for flow
venturi principle[17]. The new valve can easily attain crtitcal regimes and droplet entrainment and deposition. Correlations
flow under small pressure difference for the gas injection and for holdup and flow friction are selected based on flow regime
thus suppress the occurrence of casing heading while at the considerations. The OLGA model had been tested against
same time maintaining low gas injection cost. experimental data over a substantial range in geometry scale,
The new valve, which is named as NOVA, has been pressure and variety of fluid[5].
selected by many new gas-lift designs. It is also used to Besides OLGA, there are several other multiphase flow
replace the orifice valve for the unstable wells in operation[14]. simulators in the market. The dynamical performances of the
But not all the installation of NOVA valve gave a successful simulators have been tested and evaluated in different
story. Eliminating casing heading does not mean assured well applications and experiments. It was observed that none of the
stability. Density wave instability is a newly identified simulators could give satisfied results in all tests[2][15]. To
phenomenon in deep gas-lift wells in North Sea, which make our simulation investigation results be meaningful and
produce from depleted reservoirs. The unstable gas-lift wells reliable, we take the following methodology in our simulation
are all equipped with NOVA valve and no pressure oscillation study. First we use casing heading problem to verify the
in the annulus is observed. dynamical performance of OLGA since the basic
Since the gas injection rate is constant, then any variation characteristics of casing heading are already known from
in the liquid inflow to the wellbore will easily result in the previous studies. If OLGA simulation can reflect those
density change of the two-phase mixtures in the tubing due to characteristics, then we have enough reason to trust the
the change of phase fraction. Clearly, the mixture density simulation results from other type of gas-lift instabilities
change will result in the change of hydrostatic pressure drop, performed by OLGA.
thus the total pressure drop, particularly in a gravity A hypothetic well is built within OLGA2000 for
dominating system. The initiated mixture density change due simulating casing heading. The well parameters were
to phase fraction variation at the bottom of the well will travel abstracted from some North Sea gas-lift wells. The main
along the tubing as density wave, which sometimes is also parameters are as follows.
called continuity wave or void wave. This density wave does 2048 m in depth
not necessary introduce instability to the system since the well 5 inch tubing
has a self-controlling effect. This attributes to that any 10 inch casing
SPE 84917 3

2.75 inch production choke under three different gas injection rate, in which the first is the
0.5 inch injection orifice base case. The third case has the highest gas injection rate,
The reservoir parameters are which is approximately one third higher than the base case
PR = 150 bara rate. It shows that the oscillation amplitude is decreased and
TR = 108 °C frequency is increased when increasing gas injection, and
PI = 26Sm3/D/bar finally the well is stabilized when gas injection rate is
Unloading valve is not considered in the simulation. Since high enough.
the purpose of this study is to investigate the dynamical Effect of gas-lift valve size. The port size of gas-lift valve
characteristics of the well, what kind of fluid the well is has strong impact on well stability. Valve with smaller port
producing is not a key issue here. We assume that there is no size can increase stability since it can suppress big flow
water in the produced fluids. GOR is approximately 80 variations. Figure 2 shows the time series of bottomhole
Sm3/Sm3 for the oil sample used in the simulation. flowing pressures for three cases with different port sizes, in
The hypothetic well has three boundary conditions. In the which the smallest size is 3/8 inch and the biggest is ¾ inch. It
reservoir side, reservoir pressure, temperature and productivity is clear that the oscillation amplitude is decreased when using
index are given. On the surface, separator pressure is assumed smaller size orifices. At 3/8-inch case, the well is
constant. While at the casing head, there is a fixed gas fully stabilized.
injection rate. Besides, a steady formation temperature field Effect of productivity index. Productivity index also can
along the well path is assumed and a constant overall heat change the stability of the well. Since increased gas injection
transfer coefficient is used. The base case is that production will decrease the hydrostatic pressure drop by decreasing
choke is fully open, separator pressure equals 15bara and mixture density, thus promotes the instability, a higher
injection gas rate is 72000Sm3/D. productivity index should have a stabilizing effect by
For density wave instability study, a similar hypothetic supplying more reservoir fluid to compensate the mixture
well is built with further simplification. Since gas injection density reduction caused by increased gas fraction. This is also
through gas-lift valve is constant, then the flow in annulus and clearly demonstrated by the simulation. Figure 3 shows the
through the valve is dropped from the simulation, instead, a effects of productivity index. Increasing productivity index
constant gas source is given near the bottom of the well. reduces the oscillation amplitude and the well becomes stable
Because density wave instability is a newly identified when it is big enough. In this simulation, the productivity
phenomenon, we decide to remove all the weak factors from index of case three is about three times of that in the base case.
our simulation so that the results are more representative and Effect of choke opening. Choking is the most used
with common sense. This leads that air and water are used as pragmatic way of stabilizing unstable wells. The role of
test fluids in the simulation instead of gas and oil, and the choking relies on that smaller choke opening can increase
system is assumed isothermal. Detailed information of the friction loss for the whole system. The lift performance curve
well and reservoir are at smaller choke opening will become more flat than that at
2500m in depth bigger opening. OLGA simulation results on choking effect
5 inch tubing are given in figure 4. It shows that choking decreases
2.75 inch choke instability. When the choke opening is about 40%, the
PR =50 to 260 bara oscillation amplitude is already significantly reduced.
Psep=10bara Continuing to close the choke to 20% opening, the well is
PI=34.5Sm3/D/bar then stabilized.
OLGA2000 has an internal steady-state simulator. In our Summary. OLGA2000 simulation results succefully
study, all the simulation is initiated from steady-state demonstrate the main characteristics of casing heading
calculation results. problem. The limit cycle of casing heading process has been
clearly reflected by the simulations. There could be some
Results and discussions offsets in numbers if OLGA simulation can be compared with
Characterisation of casing heading. As mentioned in lab experiments, but from the above simulations, we believe
previous section, casing-heading problem is simulated to that OLGA at least has the capability to qualitively capture the
verify OLGA dynamical performance when simulating main dynamics of gas-lift instability. This gives us the
unstable gas-lift wells. Effects of four factors on casing confendence of using OLGA in the following investigations
heading are simulated. These include gas injection rate, size of since they are rather qualitive than quantative.
gas-lift valve, productivity index and choke opening. For each Production loss due to casing heading. The production
factor, two additional cases are tested upon the base case to loss due to casing heading is then investigated. At the basic
see the tendency of stability. settings of the base case, we first study the production loss at
Effect of gas injection rate. Increasing gas injection rate different gas injection rate. The gas rate is from 0 to
can increase the stability if the well is operated near its 100000Sm3/D. The steady-state prediction of oil production
optimum gas-lift point. This attributes to two factors caused rate for a given gas rate is determined by OLGA internal
by increased gas injection. One is the increased friction loss steady-state simulator. In fact all the dynamical simulation is
due to increased flowrate in the tubing. Another is that high initiated by the steady-state simulation results. The real
gas rate can make the flow in the annlus more stiff, thus production rate under casing heading is a time average value
reduce the delay effect. Figure 1 gives the simulation results based on the dynamical simulation results for 24hrs. The
showing the change of well bottomhole flowing pressure averaged rate is normalized to the steady-state prediction. If
4 SPE 84917

the normalized value is not unity, then it is judged that the The well will be stable if gas rate is high enough. The
production rate is changed due to instability. Depending on stable region is mainly located in the upper right direction on
the normalized value, production rate could be increased or the map. All the stable cases are market with small circles.
decreased. Figure 5 shows the results for different gas The stability map is quite representative. It is clear that density
injection rate. It is clear that production rate is reduced in the wave instability can easily occure in a well with low reservoir
well settings that are of interest. Up to 35% production loss pressure and low gas injection rate.
can happen due to casing heading. When increase the gas rate Production loss due to density wave instability. The same
to a certain level, the well is stabilized and the dynamical as casing heading problem, density wave instability can also
simulation result is the same as that predicted from steady- cause flow rate change compared with steady-state
state simulation. predictions. Figure 11 shows the normalized production rate
Figure 6 gives the results at different choke opening. Here, of unstable production at different gas injection rate. Reservoir
all production rates are normalized to the steady-state pressure is 90bara for all the simulations. The interesting
prediction when choke is fully opened. The gas injection rate observation here is that unstable production due to density
is the same as the base case, choke is adjusted to different wave instability not only can reduce the production but also
openings from 0 to 100%. The well is unstable at large choke can increase the production if the gas rate is low enough. For
opening and loses its production by 20% compared with an example, according to the simulation result, the well can
steady-state prediction when choke is fully opened. When produce as twice as its steady-state prediction when the well
reduce the choke opening to 10%, the well is absolute stable. just steps over the threshold from “no production” to
In this situation, the production rate is about 60% of the “unstable production”.
steady-state prediction when choke is fully opened. This In most situations, the range of gas rate that can increase
demonstrates that static choking can result in big production production is not of interest in practical gas-lift wells. The
loss even it is effective to stailize the wells. If comparing the normal gas rate is much larger than that. For the high gas rate,
real production rate (the dashed line in the figure) between production loss rather than gain is often observed as shown in
large choke opening and small choke opening, we find that the figure 11, in which up to 24% production reduction can occure
difference is quite small. For an example, the well in fact at a certain gas injection rate. We have even observed that up
produces almost the same amount of oil at the situations of to 50% production loss is possible from other well settings.
20% and 100% opening. This might mislead the operators by
the impression that choking does not result in production loss, Effects of feedback control. As indicated in casing heading
thus it is not necessary to tackle the instabilities by other problem, static choking can only stabilize the well without
measures rather than choking. saving the production loss. Dynamical choking based on
feedback control can not only stabilize unstable wells, but also
Basic characteristics of gas-lift density wave instability. save the production loss. The results of applying feedback
From some North Sea gas-lift wells, it was found that they control to both casing heading and density wave instability are
could be unstable even when NOVA valves were applied. This presented as follows.
observation initiated a new study in gas-lift instability for the Control structure and controller design. Since this paper
situation that gas injection rate at the bottom of the well is mainly discusses the production loss and the possibility of
constant and annulus flow is not involved in the instability. using feedback control to remedy it, only very simple
Primary investigation shows that density wave instability can feedback control structure is tested, in which bottomhole
occure in deep depleted gas-lift wells. flowing pressure is controlled by production choke. PI
Figure 7 is a stability map for the second hypothetic well. controller is selected to carry out the task.
For a given reservoir pressure, particularly when it is not high The controller settings are determined by following steps.
enough to overcome the hydrostatic head and make the well First, the optimum setting of the bottomhole flowing pressure
natrually flow, three phonomena, indicated as “no is determined from OLGA steady-state simulation. In fact, the
production”, “unstable” and “stable production” in the map, setting should be lit bit higher than the steady-state prediction.
can be observed when increasing gas injection rate from low Then the integration time is selected proportion to the period
to high. When the gas rate is too low to lift the liquid, gas time of the oscillation. A small gain value is determined so
phase will flow through the liquid and release near the top of that stability can be achieved. The gain value is adjusted by
the well. The liquid level may be oscillating within the tubing, trial-and-error so that the setting of bottomhole flowing
but no production is observed at the wellhead. This kind of pressure is as close as possible to the steady-state prediction
phonmenon is marked by the cross sign on the stability map. and a quick stabilization can be realized.
If increasing the gas rate to a certain level, a burst-like Active control of casing heading. Figure 12-14 give the
liquid production is observed. The well starts to produce results of stabilizing the base case of casing heading. The
unstably. Figure 8 gives the outlooking of the production at simulation is initiated by steady-state calculation, and then the
this situation. Long periodic time is its typical characteristic. If limit cycle is developed. At the beginning, the choke opening
continue increasing the gas rate, both the frequency and the is 96%. The controller is started after 8 hours from the
amplitude of the oscillation will increase. This is shown in beginning of the simulation. Figure 12 shows the variation of
figure 9. The oscillation will become more harmonic-like if choke opening before and after controller is started. Figure 13
the gas rate is increased further as shown in figure 10. All the gives the change of pressure and production rate accordingly.
unstable cases are marked by black dots on the map. It is seen that after about 3 hours the well is stabilized. The
choke opening is then about 88%. The production rate is
SPE 84917 5

stabilized at the level that is very close to its steady-state 2. Density wave instability can occure in gas-lift wells
prediction indicated in the figure at time 0. In fact, normalized equipped with NOVA valve, particularly when the
production rate is about 95% marked by a black dot in figure reservoir is depleted and gas injection rate is low.
5, which is much higher than that without using feedback 3. The performance of dynamic simulator, OLGA2000,
control. The original rate is only 78%, so 17% production has been verified by reflecting the main
is saved. characteristics of casing heading problem. It shows
If we keep the choke opening determined by the controller promising capability in simulating
and switch back to manual control, the well will be unstable gas-lift instabilities.
again. This is shown in figure 14. It demonstrates the function
of feedback controller. Acknowledgement
Active control of density wave instability. Figure 15-17 This research is supported by PETRONICS project sponsored
show the results of applying feedback control to the case given by ABB, Norsk Hydro AS, Norwegian Research Concil and
by figure 9. The controller is started after 5.5 hours from the Norwegian University of Science and Technology.
beginning of the simulation. Figure 15 shows the choke Scandpower Petroleum AS is acknowledged for offering
opening variation. It finally approaches 40% when the well is OLGA2000 acdemic licence.
stabilized 10 hours later after the controller is started. Figure
16 shows that the stabilized production rate is also quite close Refereces
to steady-state prediction. In fact 96% normalized rate is 1. Alhanati, F.J.S., Schmidt, Z., Doty, D.R. and Lageref, D.D.:
obtained after stabilization by feedback control. This is "Continuous Gas-Lift Instability: Diagnosis, Criteria, and
marked in figure 11 by a black dot. Comparing with the case Solutions", SPE 26554, presented at the 68th Annual Technical
without using feedback control, in which the rate is only 76%, Conference and Exhibition Held in Houston, TX,
3-6 October 1993.
20% production is saved. 2. Asheim, H.: "Verification of Transient, Multi-Phase Flow
If keep the choke opening determined by feedback Simulation for Gas Lift Application", SPE 56659, presented at
controller and switch back to manual control, the well is the SPE Annual Technical Conference and Exhibition held in
unstable again even the choke opening is only about 40%. Houston, Texas, 3-6 October 1999.
This is shown in figure 17. Comparing the simulation results 3. Asheim, H.: "Criteria for Gas Lift Stability", JPT (November
in figure 9 and figue 17, we see the instability is reduced as an 1988), pp1452-1456.
effect of static choking. This demonstrates that static choking 4. Avest, D.ter and Oudeman, P.: "A Dynamic Simulator to Analyze
can also reduce the density wave instability. and Remedy Gas Lift Problems", SPE 30639, presented at the
Summary. Both casing heading and density wave SPE Annual Technical Conference and Exhibition held in Dallas,
Texas, 22-25 October 1995.
instability can be stabilized by feedback control. The 5. Bendiksen, K.H., Malnes, D., Moe, R., Nuland, S.: ”The
production loss can be partly remedied as an effect of Dynamic Two-Fluid Model OLGA: Theory and Application”,
stabilization. In our simulation, only a simple feedback control SPE-PE, 171, May 1991,
structure is tested and the controller tuning is also rough. Even 6. Blick, E.E., Enga, P.N. and Lind, P.C.: "Theoretical Stability
so, the simulation results clearly demonstrate the function of Analysis of Flowing Oil Wells and Gas-lift Wells", SPE
feedback control when applying to unstable gas-lift wells. Production Engineering (November 1988), pp504-514.
Further study on the details of control structure and controller 7. Clegg, J.D.: "Discussion of Economic Approach to Oil
design is suggested. Production and Gas Allocation in Continuous Gas Lift", J. Pet.
. Tech., pp. 301-302 (February 1982).
8. De Cachard, F. and Delhaye, J.M., “Stability of Small Diameter
Conclusions Airlift Pumps”, Int. J. Multiphase Flow 24, 17-34, 1998.
The information provided in this paper can be used in gas-lift 9. Eikrem, G.O., Foss, B., Imsland, L., Hu, B. and Golan, M.:
design, operation and optimisation. The results remind the “Stabilization of Gas Lifted Wells”, IFAC 2002 WC, Barcelona,
operators of taking necessary measures to smooth the unstable Spain, July 2002.
production even though sometimes it does not harm the safety. 10. Grupping, A.W., Luca, C.W.F. and Vermulen, F.D.: "Continuous
The results also show the potential of reduce production loss Flow Gas Lift. Heading Action Analysed for Stabilization", Oil
by feedback control, thus, gives a new option to optimise the & Gas Journal, pp47-51, July 23, 1984.
production. Following conclusions can be made from 11. Grupping, A.W., Luca, C.W.F. and Vermulen, F.D.: "Continuous
this study. Flow Gas Lift. These Methods Can Eliminate or Control Annulus
Heading", Oil & Gas Journal, pp186-192, July 30, 1984.
1. Unstable gas-lift can definitely result in production 12. Hjalmars, S., “The Origin of Instability in Airlift Pumps”, J.
reduction. Around 20-40% of production loss is Appl. Mech. 399-404, June 1973.
observed due to gas-lift instability for typical well 13. Jansen, B., Dalsmo, M., Nøkleberg, L., Harvre, K., Kristiansen,
settings in our simulations. Static choking can V. and Lemetayer, P.: "Automatic Control of Unloading Lifted
normally stabilize the gas-lift at a cost of losing Wells", SPE 56832, presented at the 1999 SPE Annual Technical
production. But dynamic choking not only can Conference and Exhibition held in Houston, Texas,
stabilize the gas-lift wells, but also reduce the 3-6 October 1999.
production loss. In the simulated cases, the wells 14. Juan, F., Bermudes, G. and Cuauro, A., “A Solution to Instability
produce closely to its steady-state predictions when Problems in Continuous Gas-lift Wells Offshore Lake
Maracaibo”, SPE 53959, presented at the 1999 SPE Latin
feedback control is applied. American Petroleum Engineering Conference, Caracas,
Venezuela 21-23 Aprill 1999.
6 SPE 84917

15. Lopez, D., Dhulesia, H., Lerocher, E., Duchet-Suchaux, P.: 1.4E+07

“Performance of Transient Two-Phase Flow Models”, SPE

38813, Annual Tech. Conf. And Exhibition, San Antonio, TX, 5- 1.3E+07

8 October 1997.
1.2E+07
16. Tinoco, M.M.: "Validation and Improvement of Stability Criteria
for Gas-Lift Wells", M.Sc. Thesis, The Univ. of Tulsa. (1998). 1.1E+07
17. Tokar, T., Schmidt, Z. and Tuckness, C.: "New Gas Lift Valve

Pwf (Pa)
Design Stabilizes Injection Rates: Case Studies", SPE 36597, 1.0E+07

presented at the 1996 SPE Annual Technical Conference and

Exhibition held in Denver, Colorado, 6-9 October 1996. 9.0E+06

18. Xu, Z.G. and Golan, M.: "Criteria for Operation Stability of Gas
8.0E+06
Lift", SPE 19362, unsolicited, June 9, 1989.
Gas injection 1
Gas injection 2
7.0E+06
Gas injection 3

6.0E+06
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04
Time (s)

Figure 1. Effect of gas injection rate on stability, gas injection rate

1<gas injection rate 2<gas injection rate 3.

1.4E+07

1.3E+07

1.2E+07

1.1E+07
Pwf (Pa)

1.0E+07

9.0E+06

8.0E+06

Orifice port size 1

7.0E+06
Orifice port size 2
Orifice port size 3
6.0E+06
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04
Time (s)

Figure 2. Effect of orifice size on stability, orifice port size

1>orifice port size 2>orifice port size 3.

1.4E+07

1.3E+07

1.2E+07

1.1E+07
Pwf (Pa)

1.0E+07

9.0E+06

8.0E+06

PI1
PI2
7.0E+06 PI3

6.0E+06
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04
Time (s)

Figure 3. Effect of productivity index on stability, PI1<PI2<PI3

SPE 84917 7

1.4E+07 8.E+04

1.3E+07 7.E+04

1.2E+07 6.E+04

No production

Gas injection rate (Sm3/D)

1.1E+07 Unstable
5.E+04
Stable
Pwf (Pa)

1.0E+07 4.E+04

9.0E+06 3.E+04

8.0E+06 2.E+04

Choke opening 1
7.0E+06 Choke opening 2 1.E+04
Choke opening 3

6.0E+06 0.E+00
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260
Time (s) PR-Psep (bar)

Figure 4. Effect of choke opening on stability, choke opening Figure 7. Stability map for density wave instability.
1<choke opening 2<choke opening 3.

1.0
1.E+07 350

0.9

9.E+06 300
0.8

Production rate at wellhead (m /D)

0.7 8.E+06 250
Normalized Production Rate

3
0.6
7.E+06 200
Pwf (Pa)

0.5

Open loop dynamical simulation results 6.E+06 150

0.4
With feedback control

0.3 Bottomehole flowing pressure

5.E+06 100
Production rate at wellhead
0.2
4.E+06 50
0.1

0.0 3.E+06 0
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04 7.E+04 8.E+04 9.E+04 1.E+05 0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04
Gas injection rate (Sm3/D) Time (s)

Figure 5. Production loss due to instability. Figure 8. Outlooking of density wave instability, PR is 90bara and
3
air injection rate is about 18000Sm /D.

1.0
1.E+07 2100

0.9

9.E+06 1800
0.8

Production rate at wellhead (m /D)

0.7 8.E+06 1500
Normalized production rate

0.6
7.E+06 1200
Pwf (Pa)

0.5

Steady-state simulation 6.E+06 900

0.4
Unstable (from dynamic simulation) Bottomhole flowing pressure
Production rate at wellhead
0.3
5.E+06 600

0.2

4.E+06 300
0.1

0.0 3.E+06 0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04
Choke opening (-) Time (s)

Figure 6. Effect of static choking on production rate. Figure 9. Outlooking of density wave instability, PR is 90bara and
3
air injection rate is about 40000Sm /D.
8 SPE 84917

1.E+07 1400 1.E+07 3500

9.E+06 1200 1.E+07 3000

Production rate at wellhead (m /D)

Production rate at wellhead (m /D)

8.E+06 1000 1.E+07 2500

3
Pressure (Pa)
7.E+06 800 8.E+06 2000
Pwf (Pa)

Bottomhole flowing pressure

6.E+06 600 6.E+06 1500
Production rate at wellhead

5.E+06 400 4.E+06 1000

Bottomhole flowing pressure

4.E+06 200 2.E+06 Annulus pressure near gas-lift valve 500
Production rate at wellhead

3.E+06 0 0.E+00 0
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04 0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04 7.E+04 8.E+04 9.E+04
Time (s) Time (s)

Figure 10. Outlooking of density wave instability, PR is 90bara and Figure 13. Variation of pressure and production rate before and
3
air injection rate is about 54000Sm /D. after controller is started.

2.0
1.E+07 3500

1.8
1.E+07 3000
1.6

Production rate at wellhead (m /D)

1.4 Open loop dynamical simulation results 1.E+07 2500
Normalized production rate

3
With feedback control
1.2
Pressure (Pa)

8.E+06 2000
1.0

6.E+06 1500
0.8

0.6
4.E+06 1000

0.4 Bottomhole flowing pressure

Annlus pressure near gas-lift valve
2.E+06 500
Production rate at wellhead
0.2

0.0 0.E+00 0
1.0E+04 1.5E+04 2.0E+04 2.5E+04 3.0E+04 3.5E+04 4.0E+04 4.5E+04 5.0E+04 5.5E+04 6.0E+04 0.0E+00 5.0E+03 1.0E+04 1.5E+04 2.0E+04 2.5E+04 3.0E+04
3
Gas injection rate (Sm /D) Time (s)

Figure 11. Production loss due to density wave instability. Figure 14. Unstable production when controller is switched back
to manual control.

1.2
1.2

1.0
1.0

0.8
0.8
Choke opening (-)
Choke opening (-)

0.6
0.6

0.4
0.4

0.2
0.2

0.0
0.0
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04 7.E+04 8.E+04 9.E+04 1.E+05
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04 7.E+04 8.E+04 9.E+04
Time (s)
Time (s)

Figure 12. Variation of choke opening before and after controller Figure 15. Variation of choke opening before and after controller
is started. is started.
SPE 84917 9

1.E+07 2000

9.E+06 1800

8.E+06 1600

Production rate at wellhead (m /D)

7.E+06 1400

3
6.E+06 1200
Pwf (Pa)

5.E+06 1000
Bottomhole flowing pressure
4.E+06 Production rate at wellhead 800

3.E+06 600

2.E+06 400

1.E+06 200

0.E+00 0
0.E+00 1.E+04 2.E+04 3.E+04 4.E+04 5.E+04 6.E+04 7.E+04 8.E+04 9.E+04 1.E+05
Time (s)

Figure 16. Variation of pressure and production rate before and

after controller is started.

1.E+07 1500

9.E+06 1350

8.E+06 1200
Production rate at wellhead (m /D)

7.E+06 1050
3

6.E+06 900
Pwf (Pa)

Bottomhole flowing pressure

5.E+06 750
Production rate at wellhead

4.E+06 600

3.E+06 450

2.E+06 300

1.E+06 150

0.E+00 0
0.0E+00 5.0E+03 1.0E+04 1.5E+04 2.0E+04 2.5E+04 3.0E+04 3.5E+04 4.0E+04
Time (s)

Figure 17. Unstable production when controller is switched back

to manual control.