You are on page 1of 5

Criteria for Gas-Lift Stability

Ha~ald Ashelm, SPE, U. of Trondheim

Summary. Severe flow instability (heading or annulus heading) is known from operations of gas-lift systems. Here, two simple
stability criteria are developed and compared with reported field data. The stability problems experienced for the cases examined
would have been identified with these criteria and corrected at the design stage.

Introduction
The currently used principles for gas-lift design were established In the current work, two simple criteria are developed providing
during the early 1950's. 1-3 They provide relations between (1) gas causal relationships between gas-lift design parameters and flow
injection pressure and the most efficient point of injection and (2) stability. The criteria developed do not substitute the more advanced
gas injection rate and the production rate to be expected. From these approaches of unstable flow behavior, but they may provide a prac-
relations, standardized procedures for gas-lift design have been tical method for the design of stable gas-lift systems.
worked out. 4 Works on application and optimization of the
procedures have provided further insight into the interrelations be- Mechanisms of Gas-Lift Instability. Fig. 1 shows an abstraction
tween gas-lift design and economic performance. 5-8 of a gas-lift system. It is assumed that the high-pressure lift gas
Often-unstated assumptions of gas-lift design are that it will be enters the surface inlet of the gas conduit (surface piping and
possible to inject gas at a constant downhole rate and that the casing/tubing annulus, or dedicated lift string) at a constant rate.
resultant production rate will be stable. This is not necessarily true; The lift gas will flow through the gas conduit and enter the tubing
severe flow instability is well known in the actual operation of gas- through a subsurface injection port. The gas inflow rate into the
lift systems. tubing is governed by the pressure difference across this port, be-
Variations in pressure and flow rate are observed in all multiphase tween the gas conduit and the tubing. By conventional gas-lift
flow systems, even in pumping wells, because of redistribution of design, constant inflow of lift gas is assumed. As mentioned, the
gas and liquids. They cause relatively small short-duration pressure tubing pressure may show temporary variations, causing temporary
and flow changes. Alone, this has little effect on the continuity of variations in the gas inflow rate. The question addressed here is
production. In a gas-lift system, however, it may trigger system how the gas-lift system will respond to this.
instabilities. If an increase of gas inflow causes increased pressure difference
API9 recommends that, for the sizing of pipes receiving gas- between the gas conduit and the tubing, then the gas inflow to the
lifted production, a "surge factor" of 40 to 50% should be added tubing will increase further. This positive feedback leads to unstable
to the estimated steady-state flow rate, compared with 20% for flow behavior, as described by Bertuzzi et at. 2 If an increased flow
naturally flowing wells. 9 Intended as guidelines for cases where of gas causes decreased pressure difference between the gas conduit
more definite information is lacking, these numbers may indicate and the tubing, gas flow will decrease. Under this condition, the
something about the uncertainties concerning the flow instabilities gas-lift system will be stabilized by negative feedback.
during gas lift.
Bertuzzi et al. 2 observed that when the lift-gas input rate was Stability Criteria
reduced below a certain minimum, violent heading would occur In Appendices A and B, first-order stability analyses for gas-lift
and the liquid production would eventually cease. They postulated systems are performed. This gives two explicit stability criteria.
that "a sudden drop in pressure in the tubing brought about a sudden The first quantifies stabilization as a result of the inflow responses
surge of gas into the tubing. The volume of gas surging into the of reservoir fluid and lift gas; the second quantifies stabilization
tubing is dependent on the pressure and volume of gas in the an- caused by depletion of the gas conduit pressure.
nular space. If the pressure in the annular space dropped too much,
gas ceased to flow into the tubing. " More recently, gas-lift insta- Inflow Response. If the inflow rate of the heavier reservoir fluids
bilities have led to shutdowns of wells in the Claymore field. 10 is more sensitive to pressure than the lift-gas flow rate, then the
This was amended by replacement of the downhole injection valve average density of the flowing fluid mixture will increase in response
by a fixed orifice. to a decrease in tubing pressure. This causes the tubing pressure
Flow instabilities have also been observed and analyzed for simple to increase again, which stabilizes the flow. Appendix A shows that
air-lift pumps. 11,12 This is related to gas-lift instability. However, stabilization by the inflow response requires (Criterion 1)
the inflow mechanisms of a gas-lift system are considerably more
complicated than for an air-lift pump. Besides, the friction damp- PgseBgqgs} J
ening will be much larger in a gas-lift system because of order-of- FI = - - > 1. ........................ (1)
qLse (EAj)2
magnitude-larger flow length. Thus, the dominating mechanisms
of instability will be quite different. By this criterion, stability is promoted by a high flow rate of lift
During the last few years, attempts have been made to under- gas, a high productivity index, and a small injection port.
stand and to quantify gas-lift instabilities with numerical techniques.
One approach is to make a dynamic numerical model of the gas- Pressure-Depletion Response. If the first criterion is not fulfilled,
lift system, assuming that instabilities that occur when the model a decrease in the tubing pressure will cause the gas flow rate to
is run on a computer represent physical flow instabilities, as increase more than the liquid flow rate. This will cause a decreasing
Grupping et al. did. 13,14 This succeeds in demonstrating unstable tubing pressure, but will also deplete the gas conduit pressure. If
flow behavior by numerical means. The other approach is to apply the gas conduit pressure depletes faster than the tubing pressure,
linear stability analyses directly on a mathematical model of the then the pressure difference between the gas conduit and tubing
flow system. Fitremann and Vedrines l5 performed linear stability will decrease, and so will the lift-gas rate. This stabilizes the flow.
analyses for a gas-lift system. The results after low-pressure sim- Appendix B shows that stability corresponds to Criterion 2:
plifications were shown to correspond to small-scale laboratory ex-
periments. No field data comparisons were attempted.
Copyright 1988 SOCiety of Petroleum Engineers

1452 Journal of Petroleum Technology, November 1988


By this criterion, stability is promoted by a small gas conduit
volume, a high gas flow rate, and a high inflow-response ratio. A .---_ _ _ ----+ Production

high tubing pressure, provided by higher wellhead backpressure,


will be stabilizing if the downhole gas injection volume is main-
tained constant. rr======= 4 - - - Lift gas

Relationships to Existing Recommendations


and Models Tubing

Some of the above considerations can be recognized in existing


design rules. API 4 recommends that the size of the injection port
Gas conduit (annulus or
be chosen so that a pressure differential of 690 kPa [100 psi] is dedicated string)
established across the injection port. For many gas-lift installations,
this will secure a stable inflow-response ratio: F1 > l.
Bertuzzi et at. 2 observed that the use of an auxiliary, small-
diameter lift-gas string, would stabilize wells that were unstable
when injected through the casing/tubing annulus. Criterion 2 shows Downhole choke
that stability can always be achieved by choice of a sufficiently small
gas-lift conduit.
On the basis of experience and numerical simulation, Grupping
et at. 14 stated that gas-lift stabilization should "be based on the
principle that the choking effect exercised at the surface injection
orifice, relative to that of the downhole orifice, should be de- Reservoir
creased." By the current model, this conclusion can be derived from
Criterion 1.
Fitremann and Yedrines 15 observed that the pressure drop at the
gas injection point has a strong stabilizing effect. This again corre-
sponds to Criterion 1, expressed most clearly by Eq. A-9.
In the pressure-depletion-response analyses, Appendix B, the Fig. 1-Gas-lift system.
Sffects of inertia and friction danIpening in the tubing are neglected.
These "second-order" effects are included by Fitremann and Ye-
drines in their model and presumably also by Grupping et al. 13 pressure surges and prevented injection from the lower injection
Fitremann and Yedrines' analysis showed that waves of three prin- port. Stability was later achieved by replacing the bottomhole in-
cipal modes may establish in the tubing. The higher-frequency waves jection valves by two fixed 9.5-mm [2r64-in.] orifices. Well C-6
are danIpened by flow friction; the lowest frequency is undampened. showed a considerably lower productivity index than Well C-2;
The lowest frequency is the continuity wave, created by a change therefore, stability problems were expected. Well C-6 was therefore
in the inflow-mixture density. equipped initially with fixed 9.5-mm F%4-in.] downhole orifices.
By neglecting inertia and friction, the current analysis is based With this arrangement, the well gave no stability problems.
on the assumption that the higher-frequency waves are sufficiently The data reported for the Claymore wells are listed in the first
dampened in a field-scale installation, so they can be neglected. two columns of Table 1. The production and injection rates and
The field cases analyzed below appear to support this assumption. the pressure at the injection point are the design parameters reported
for the system:The effective injection port size for the valve type
Examination of Reported Field Cases originally installed in Well C-2 was estimated from the performance
Data on gas-lift instability are scarce in the literature. The cases chart given. A tubing-pressure-controlling feature of the valve ap-
reported by DeMoss and Tiemann 10 and by Bertuzzi et al. 2 are parently did not work and was neglected. Table 2 lists the estimated
primarily studies of stable gas-lift performance. However, they fluid properties and downhole flow rates (volumetric flow rates at
contain enough information to examine the stability criteria, de- gas injection conditions).
veloped above vs. actual gas-lift performance. Bertuzzi et al. 's data were collected from an experimental well.
DeMoss and Tiemann report that Well C-2 in the Claymore field In Cases 1 through 4 the gas was injected through a small-diameter
turned out to be unstable when gas-lifted. The instability caused auxiliary string. In these cases, no stability problems were experi-

TABLE 1-DATA REPORTED

Claymore Claymore Bertuzzi et al. Bertuzzi et al. Bertuzzi et al.


Well C-2 Well C-6 Case 2 Case 7 Case 12
Vertical depth to injection port, It [m) 7,600 [2317) 7,865 [2397) 4,500 [1372) 3,810 [1161) 3,810 [1161)
Tubing 10,' in. [mm) 4.78 [121.4) 4.78 [121.4) 1.995 [50.7) 1.995 [50.7) 1.995 [50.7)
Tubing 00,' in. [mm) 5.51 [140.0) 5.51 [140.0) 2.375 [60.3) 2.375 [60.3)
CaSing 10,' in. [mm) 8.7 [221) 8.7 [221) 5 [127) 5 [127)
Gas-string 10, in. [mm) 0.824 [20.9)
Liquid production rate, BID [m 3/s) 14,000 [0.0258) 12,000 [0.0221) 374 [0.000688) 541 [0.000995) 541 [0.00114)
Gas injection rate, MscflD [std m /d)3 11,200 [3.67) 12,000 [2.87) 68.3 [0.0224) 192.3 [0.0630) 507.9 [0.1664)
WOR, 1t3/lt 3 0.025 0.04 306 105 18.8
Nominal injection port size," in. [mm) 0.91 (23) 24/64 [9.53) 14/64 [5.6) 14/64 [5.6)
Orifice efficiency factor 0.9 0.9 0.9 0.9 0.9
Injection-gas specific gravity 0.81 0.81 0.668 0.668 0.668
Oil specific gravity 0.884 0.884 0.846 0.846 0.846
Water specific gravity 1.07 1.07 1.07
Formation gaslliquid ratio, 1t3/lt 3 ",0 13 11.2 29.1 67.5
Temperature at injection port, OF [K) 172 (351) 172 (351) 166 (348) 162 (346) 162 (346)
Pressure at injection port, psi [kPa) 1,610 [11 100) 1,600 [11 030) 1,035 (7140) 590 (4070) 600 [4140)
Productivity index, BID-psi [m 3/s· Pal 26 [6.94x10- 9 ) 14.4 [3.84x10- 9 ) 1.88 [5.02x10- 1O ) 1.88 [5.02x10- 1O ) 1.88 [5.02x10- 1O )

·Values from tubing tables based on nominal diameters given .


.... Two valves/orifices are used for the Claymore wells. The equivalent port size for the valves in Well C-2 is estimated from valve performance curve given. 10

Journal of Petroleum Technology, November 1988 1453


TABLE 2-ESTIMATED FLUID PARAMETER AND FLOW RATES. IN SI UNITS AS APPLIED IN THE CALCULATIONS

Claymore Claymore Bertuzzi et a/. Bertuzzi at al. Bertuzzi at al.


Well C-2 Well C-6 Case 2 Case 7 Case 12
z factorof injected gas 0.79 0.79 0.9 0.92 0.92
FVF of injected gas 0.00877 0.00882 0.0154 0.0274 0.0270
Downhole density of injected gas, kglm 3 113 112 52.9 29.7 30.2
Downhole density of reservoir fluid mix,
kg/m 3 884 884 911 593 377
Downhole fluid (oil, gas, water) rate, m 3 /s 0.0258 0.0221 0.000807 0.00179 0.00323
Downhole gas injection rate, m 3 /s 0.0322 0.00254 0.000345 0.00173 0.00449

enced. Case 2 examined here is the lowest-flow case. According Pt = tubing pressure, Pa [psi]
to our stability criteria, this would be the least stable. Ptf = tubing-head flowing pressure, Pa [psi]
For Bertuzzi et ai.'s Cases 5 through 18, the gas was injected Pti = tubing flowing pressure at gas injection point, Pa
into the annulus. The well was equipped with a 5.6-mm [1%4-in.] [psi]
downhole orifice. They reported that "liquid production could be
Pwf = bottornhole flowing pressure, Pa [psi]
varied only over a range of about 10 percent by varying the input
!::"Pj = friction loss, Pa [psi]
gas rates. If the gas input rate was reduced below the minimum,
heading would occur and the flow would eventually cease. " Cases qfi = flow rate of reservoir fluids at injection point, m 3/s
7 and 12 are examined here. Case 12 had the highest reported liquid [ft 3/sec]
production and should be stable. In Case 7, the liquid production qgi = flow rate of lift gas at injection point, m 3/s [ft 3/sec]
is about 13 % lower than for Case 12; it should therefore be at least qgse = flow rate of lift gas at standard conditions, std m 3/s
on the border of instability. . [scf/sec]
The data reported by Bertuzzi et ai. are listed in the three last qLse = flow rate of liquids at standard conditions, std m 3/s
columns of Table 1. Table 2 lists the estimated fluid properties and [scf/sec]
downhole flow rates (volumetric flow rates at gas injection con- R = universal gas constant, Nm/kmol' K [ft-Ibf/gmol' OF]
ditions). For both Bertuzzi et ai.'s data and the Claymore cases, t = time, seconds
0.9 was used for the orifice efficiency factor.
'rei = conduit gas flowing temperature at the injection
Table 3 summarizes the stability criteria calculated for the cases
examined. As seen, the estimates correspond nicely to observed point, K reF]
behavior. A discrepancy occurs for Bertuzzi et ai.'s Case 12, which Tti = tubing fluid flowing temperature at the injection
was reported as stable but is predicted to be unstable. This is a case point, K [OF]
of high flow rate in a small tubing with significant, but not suffi- v = flow velocity, m/s [ft/sec]
cient, stabilization by conduit pressure depletion (F2 =0.83). It is Ve = gas conduit volume, m 3 [ft3]
possible that tubing-flow friction dampening, which is neglected VI = tubing volume downstream of gas injection point,
in the criteria development, may smooth out the flow variations m 3 [ft3]
in this case. However, there may also be other explanations. Wei = mass injection rate of gas into conduit volume, kg/s
[Ibm/sec]
Conclusions
wti = mass injection rate of gas into tubing, kg/s [Ibm/sec]
On the basis of limited comparison with reported field data, the z = gas z factor
theoretically founded criteria appear to identify potentially unstable
wells and to provide quantitative guidelines for stabilization. The
o = small perturbation of steady state
stability problems experienced for the cases examined would have Pa = tubing-averaged fluid density, kg/m 3 [lbm/ft3]
been identified with these criteria and corrected at the design stage. Pfi = reservoir fluid density at injection point, kg/m3
[lbm/ft3]
Nomenclature Pgi = lift-gas density at the injection point, kg/m 3
Ai = injection port size, m 2 [ft2] [lbm/ft3]
AI = tubing flow area, m 2 [ft2] Pgsc = lift-gas density at standard surface conditions,
Bfi = FVF of reservoir fluids at injection point kg/std m 3 [lbm/sct]
Bg = FVF of gas at injection point Pi = mixture density of reservoir fluids and lift gas at
D = vertical depth to injection point, m [ft] injection point, kg/std m 3 [Ibm/sct]
E = orifice efficiency factor, here assumed to equal 0.9
F 1 ,F2 = stability criteria References
g = acceleration of gravity, m/s2 [ft/sec 2] I. Poettmann, F.H. and Carpenter, P.G.: "Multiphase Flow of Gas, Oil,
J = productivity index, std m 3 /s'Pa [scf/sec'psi] and Water Through Vertical Flow Strings with Application to the Design
of Gas-Lift Installations," Drill & Prod. Prac., API (1952) 257-317.
M = gas molecular weight 2. Bertuzzi, A.F., Welchon, J.K., and Poettmann, F.H.: "Description
Pei = gas conduit pressure at the injection point, Pa [psi] and Analysis of an Efficient Continuous-Flow Gas-Lift Installation,"
PR = reservoir average pressure, Pa [psi] Trans., AIME (1953) 198, 271-78.

TABLE 3-RESUL TS

Predicted Observed
WelllCase ~ ~ Behavior Behavior
Well C-2 0.06 0.76 Unstable Unstable
Well C-2 after valves replaced
by 20/64-in. orifices 1.9 Stable Stable
Well C-6 0.76 2.7 Stable Stable
Bertuzzi et a/. Case 2 5.2 Stable Stable
Bertuzzi et a/. Case 7 0.09 0.28 Unstable Unstable (?)
Bertuzzi et a/. Case 12 0.55 0.83 Unstable Stable (?)

1454 Journal of Petroleum Technology, November 1988


3. Gilbert, W.E.: "Flowing and Gas-Lift Well Performance," Drill. & The flow of lift gas can be described by the orifice equation for
Prod. Prac., API (1954) 126. constant-temperature flow:
4. Gas Lift, Vocational Training Series, Prod. Dept. API, 6.
5. Blann, J.R., Brown, J.S., and DuFresne, L.P.: "Improving Gas-Lift
Performance in a Large North African Oil Field," JPT(Sept. 1980)
1486-92. ..................... (A-6)
6. Kanu, E.P., Mach, J., and Brown, K.E.: "Economic Approach to Oil
Production and Gas Allocation in Continuous Gas Lift," JPT (Oct. 1981)
1887-92.
Assuming constant-temperature gas flow across the port implies
7. Clegg, J.D.: "Discussion of Economic Approach to Oil Production and
Gas Allocation in Continuous Gas Lift," JPT(Feb. 1982) 301-02. a slight approximation because the gas may undergo some expansion
8. Blann, J.R. and Williams, J.D.: "Determining the Most Profitable Gas cooling. As long as the pressure difference across the valve is small
Injection Pressure for a Gas Lift Installation," JPT (Aug. 1984) 1305-11. compared with the total pressure, this error will be negligible.
9. API RP 14E, Design and Installation of Offshore Production Platform The change in injection gas flow caused by a change in tubing
Piping Systems, API (1984). pressure can be expressed mathematically by differentiation of Eq.
10. DeMoss, E.E. and Tiemann, W.D.: "Gas Lift Increases High-Volume A-6:
Production From Claymore Field," JPT (April 1982) 696-702.
11. Hjalmars, S.: "The Origin of Instability in Airlift Pumps," Trans.,
Appl. Mech., ASME (1973) 41, 399-404. oqgi = _ (EAi)2
. ................................ (A-7)
12. Apazidis, N.: "Influence of Bubble Expansion and Relative Velocity
of the Performance and Stability of an Airlift Pump," Inti. J. Multiphase
Flow (1985) 11, No.4, 459-79.
13. Grupping, A.W., Luca, C.W.F., and Vermeulen, F.D.: "Heading Here, the conduit volume has been assumed large enough to
Action Analyzed for Stabilization," Oil & Gas J. (July 30, 1984) 47-51. prevent any significant pressure decrease. Decreasing conduit
14. Grupping, A.W., Luca, C.W.F., and Vermeulen, F.D.: "These pressure would have an extra stabilizing effect, which is considered
Methods Can Eliminate or Control Annulus Heading, " Oil & Gas J. separately in Appendix B.
(July 30, 1984) 186-92. With these derived expressions for the flow rates and their
15. Fitremann, J.M. and Vedrines, P.: "Non Steady Gas-Liquid Flow in derivatives, the criterion for a stable inflow performance may be
Pipes and Gas-Lifted Wells," Proc., Second Inti. Conference on Multi-
expressed as
Phase Flow, London (June 19-21, 1985) 245-62.

Pgiq~i Bftl
Appendix A-Inflow Response F , = - - - - - > l ........................... (A-8)
A decrease in the downhole tubing pressure will cause increased qft (EA i)2
flow of both reservoir fluid and lift gas. If the flow of gas increases
relatively more than the flow of liquid, the density of the fluid or, equivalently, in terms of pressures,
mixture decreases. This reduces the static head and the flow friction
and thus may accentuate instabilities. On the other hand, if the 2Pt' In(p ·Ipt')
density increases in response to decreasing pressure, both the static F, = I CI I > 1. .......................... (A-9)
head and the flow friction will increase and the system will be stabi- PR-Pwj
lized by negative feedback. Thus, a criterion for stability becomes
It is convenient to express the criterion in terms of surface flow
rates:
OPi
F, = - <0 .................................... (A-I)
OPli
Fl = PgscBgq~sc _1_ > 1. ..................... (A-tO)
The density of the inflowing mixture is expressed by the amount qLsc (EAJ2
of reservoir fluids and lift gas and by their respective densities:

Appendix B-Pressure-Depletlon Response


Suppose that the system is unstable by the criterion derived in Ap-
pendix A. Then a decrease in tubing pressure will cause increased
inflow of lift gas. However, the increased inflow of lift gas will
The change in density of the inflowing fluid mixture resulting also deplete the gas conduit pressure. If the gas conduit pressure
from change (perturbation) of the inflow rates can be expressed depletes faster than the tubing pressure, the gas flow rate will soon
mathematically by the differentiation of density equation: reverse to stabilize the flow:

aqg/at<o . ..................................... (B-1)

From Eq. A-3, it can be seen that the postulated stability criterion By the orifice equation (Eq. A-6), the requirement for decreasing
is fulftlled for flow is that the gas conduit/tubing pressure ratio decrease. Thus,
the flow of lift gas decreases when

p. -ap ·/at
F2=--.!!... Cl > 1. ........................... (B-2)
Pci -apti 1at
Neglecting inflow transients, the flow of reservoir liquids is
assumed proportional to the pressure difference between the The change of gas conduit pressure is expressed by the general
reservoir and the well bottom: gas equation:

apci zciRTci
qft =Bftl(ji R -Pwj)' .............................. (A-5) --=O(Wci-Wli)---' ........................ (B-3)
at VcM

Journal of Petroleum Technology, November 1988 1455


The flow of gas into the gas conduit-gas volume-is assumed By combining Eqs. B-6 and B-8 with Eqs. A-2 and A-4, we can
constant. The mass flow rate is converted to vQlume flow rates at express the pressure response of the tubing caused by a change in
tubing pressure to correspond with the parameter system used here: gas inflow as

apci Pti TciZci apti _ D Pji -Pgi qji (F I)"


--=----oqgi . .......................... (B-4) - - - g - - - - - - 1- uqgi' ............. (B-9)
at Vc TtiZ ti at Vt qji -qgi

With acceleration neglected, the tubing pressure can be described The conduit pressure depletion criterion can now be expressed
by a general momentum balance: by combining Eqs. B-2, B-4, and B-9:

Pti=Pwf+PagD+IlPf. ............................ (B-5)

Consider a small change in input gas/liquid ratio. This will prop-


agate as a continuity wave with no friction dampening. The pressure
response to such a change is found by differentiation of the flow PtiTciZci
c= ==1. ............................... (B-Il)
equation above. Assuming constant wellhead pressure and no PciTtiZti
friction dampening of the continuity wave in the tubing yields
PPt apa The parameter C defined above is the gas volume factor between
- =gD- . .................................. (B-6)
pt at the gas conduit and the tubing. For all foreseeable cases, this will
be close to unity.
To quantify the variation in tubing-averaged density, we may start
from the continuity equation. Neglecting acceleration, the continuity SI Metric Conversion Factors
equation becomes
bbl x 1.589 873 E-OI
ap ap ft x 3.048* E-Ol
-+v-=O . .................................. (B-7) ft3 x 2.831 685 E-02
at ax gal x 3.785 412 E-03
in. x 2.54* E+OO
Consider a step change in input fluid density. Until the corre- Ibm x 4.535 924 E-Ol
sponding continuity wave has propagated to the wellhead, the
outflow fluid density will be as before. Then the average density
variation can be derived from the continuity equation as above: 'Conversion factor is exact. JPT
apa qji+qgi Original SPE manuscript (SPE 16468) received for review Feb. 2.1987. Paper accepted
-=---OPi' .............................. (B-8) for publication Oct. 20, 1987. Revised manuscript received Feb. 1, 1988.
at At

1456 Journal of Petroleum Technology, November 1988

You might also like