Copyright 2005, Society of Petroleum Engineers

This paper was prepared for presentation at Offshore Europe 2005 held in Aberdeen,
Scotland, U.K., 69 September 2005.

This paper was selected for presentation by an SPE Program Committee following review of
information contained in a proposal submitted by the author(s). Contents of the paper, as
presented, have not been reviewed by the Society of Petroleum Engineers and are subject to
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
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
for commercial purposes without the written consent of the Society of Petroleum Engineers is
prohibited. Permission to reproduce in print is restricted to a proposal of not more than 300
words; illustrations may not be copied. The proposal must contain conspicuous
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.

Abstract

Liquid loading is a serious problem in areas where gas fields
are maturing. An analysis is provided of the production
behaviour over time of liquid-loaded wells. This shows clearly
that these wells can operate at two different rates, a stable rate
at which full production is taking place and a lower metastable
rate at which liquid loading effects play a role.

A model has been constructed which enhances the
understanding of the process of water build up and drainage in
gas wells. It assumes a single gas and water co-production
point and a single water re-injection point. As expected a
water column is built up in the well as soon as production
takes place below the critical rate. As observed in the field, for
good inflow performance a metastable flowrate can be
observed. At this state the water re-injection and water co-
production rate are equal to one another and the water column
height stabilizes.

A sensitivity analysis has been carried out to
determine how well parameters influence the metastable
flowrate, the time required to reach this metastable rate, the
corresponding water column height and the shut-in time
required to drain this water column. The results of the analysis
indicate that significant metastable flow rates occur in wells
which have good inflow performance, a low water gas ratio
and a large distance between injection and production point.

Furthermore a steady state analytical solution has
been derived for the metastable rate and stabilized water
column height confirming the numerical analysis results.



Introduction

Many of the mature offshore gas fields in the Southern North
Sea have already experienced considerable pressure depletion
resulting in significantly reduced gas flowrates and,
eventually, liquid loading. Insight into the flow characteristics
of these liquid-loaded wells will help manage tail-end
production i.e. will help define the most effective means of
accelerating and maximising their ultimate recovery.

Liquid loading

All wells producing from depleting gas reservoirs will
eventually exhibit so-called liquid loading. The liquid loading
process occurs when the gas velocity within the well drops
below a certain critical gas velocity. The gas is then unable to
lift the water co-produced with the gas (either condensed or
formation water) to surface. The water will fall back and
accumulate downhole. A hydrostatic column is formed which
imposes a back pressure on the reservoir and hence reduces
gas production. The process eventually results in intermittent
gas production and well die-out. This paper will describe the
liquid loading behaviour by modelling the build-up and
drainage of water in gas wells.

There is an extensive literature on liquid loading.
Duggan
1
introduced the concept of a critical minimum
wellhead gas velocity for the onset of liquid loading.
Oudeman
2
gives a review of the literature. The most widely
applied method for predicting liquid loading is based on an
analysis of droplet transport in vertical turbulent gas flow by
Turner et al
3
, leading to the prediction of a critical flowrate,
which we shall call the Turner rate. Coleman et al
4
confirmed
that the Turner rate is a good predictor of the onset of liquid
loading, and examined the influence of other parameters on
the process. Lea and Nickens
5
discussed ways of solving
liquid loading problems. Guo et al.
6
presented a further
improvement on the Turner criterion.

Although the Turner rate (or a modification thereof)
is accepted as a good criterion for predicting the onset of
liquid loading, Oudeman
2
pointed out that it is not a good
predictor of when the well will die, which may occur at a
lower rate. Recently, Sutton et al
7
noted that wells may operate
at subcritical rates, and gave a review of the methods for
analyzing this phenomenon.

Here we report on field data which clearly show the
existence of subcritical metastable flow rates, at which the
well still produces, even though liquid loading is occurring. A

SPE 95282

Modelling the Gas Well Liquid Loading Process
N. Dousi, Delft U. of Technology; C.A.M. Veeken, Shell E&P Europe; and P.K. Currie, SPE, Delft U. of Technology
2 SPE 95282
simple conceptual model is introduced which helps to
understand and interpret this data.

Field data demonstrating metastable flow

The production data of mature gas fields has been analyzed for
evidence of stable flow rates below the Turner flowrate. Five
examples are presented to illustrate the effects of liquid
loading. A number of techniques have been used to analyze
this production data.

Example 1. The transition period between stable and
subcritical flow can be observed in Fig. 1. In this example the
flowrate drops from a stable rate of about 220,000 m
3
/d to
50,000 m
3
/d. Liquid loading takes place within about 10 days.

Production data
0
50
100
150
200
250
300
17-feb 27-feb 9-mrt 19-mrt 29-mrt 8-apr 18-apr
Q
g
,
x
1
0
0
0
m
3
/
d
Qg, x1000m3/d

Fig. 1 Example 1. Production data demonstration start of liquid
loading

Example 2. The data consists of the daily flowing tubing head
temperature (T
fth
), in the period 1997 2002 of a NAM well.
Fig. 2 shows the temperature data for one well. For each year,
the data has been sorted, so that the lowest temperature
recorded is assigned to day 1, and the highest temperature to
day 365, and plotted as in Fig. 2. From this figure, it is seen
that in 1997 the T
fth
had the value of about 90 C for all but 50
days in the year. In subsequent years, the number of days at
lower temperatures increases steadily, reaching 160 in year
2002. Moreover, when the temperature falls below 90 C, it
takes discrete values, either about 20 C (ambient surface
temperature, indicating that the well is shut in) or about 40 C.
In 1997, the well flowed at this T
fth
for about 10 days. In 2002,
the number of days had increased to about 60. The downtime
also increases from 15 days in 1997 to 100 days in 2002.

Sorted flowing tubinghead temperature data
0
10
20
30
40
50
60
70
80
90
100
0 50 100 150 200 250
Time, days
T
f
t
h
,

C
1997 1998 1999 2000 2001 2002
Temperature at metstable flow
Temperature at stable flow
Ambient Temperature

Fig. 2 Example 2. Sorted annual tubing head temperatures of a
well that is occasionall y liquid loading.

These results are interpreted to indicate that the well
produces at a stable rate corresponding to the temperature 90
C. As liquid loading starts, there exists a metastable rate at
which the production rate is less and the tubing head
temperature correspondingly decreases to 40 C. An individual
liquid loading event for this well is displayed in Fig. 3,
illustrating the time it takes for this well to load up. In this
example liquid loading takes place within about one day.

Production data, year 2000
0
10
20
30
40
50
60
70
80
90
20-apr 21-apr 22-apr 23-apr 24-apr 25-apr
T
f
t
h
,

C

P
t
h
,

b
a
r
a
Tfth Pth Tamb
Stable regime
Metastable regime

Fig. 3 Example 2. Indi vidual recorded liquid loading event

Example 3. Here we introduce data from another NAM well.
In Fig. 4 a series of consecutive liquid loading events are
displayed. The tubing head pressure and flowrate are plotted
as a function of time.

Production data
0
10
20
30
40
50
60
70
80
90
100
20-sep 30-sep 10-okt 20-okt 30-okt 9-nov 19-nov 29-nov 9-dec
Q
g
,

x
1
0
0
0

m
3
/
d
-60
-50
-40
-30
-20
-10
0
10
20
30
40
P
t
h
,

b
a
r
Qg, x1000m3/d Pth, bar
Stable flowrate
Metastable flowrate

Fig. 4 Example 3. Indi vidual recorded liquid loading events

One can clearly distinguish two stable flowrates. The stable
and metastable flowrates are about 50,000 m
3
/d and 10,000
m
3
/d respectively. It takes about 8 to 10 days for the well to
load up. When we sort the hourly flowrate data in the same
way as the temperature data in example 2, a similar profile is
obtained, which can be observed in Fig. 5.

SPE 95282 3
Sorted flowrate data
0
20
40
60
80
0 1000 2000 3000 4000 5000 6000 7000
T, hours
Q
g
,

x
1
0
0
0
m
3
/
d
0
5
10
15
20
25
30
35
40
P
t
h
,

b
a
r
Sorted flowrate data Sorted tubing head pressure data
Stabe rate
Critical rate
Metastable rate

Fig. 5 Example 3. Sorted hourl y flowrate and tubing head pressure
data of a well that is occasionally liquid loading.

We can distinguish a regime in which the well produces at a
stable rate, between 40,000 and 60,000 m
3
/d and a regime at
which it produces at a lower metastable flowrate, between
9,000 and 11,000 m
3
/d. The metastable plateau observed is not
long because the operator systematically shuts in the well
upon observing liquid loading.

Example 4. Extended periods of metastable flow can be
observed in Fig. 6. In this example the well produces at a
metastable flowrate of about 14,000m
3
/d.

Production data
0
10
20
30
40
50
60
70
2-jun 22-jun 12-jul 1-aug 21-aug
Q
g
,
x
1
0
0
0
m
3
/
d
-10
-5
0
5
10
15
20
25
30
P
t
h
,
b
a
r
Qg,x1000m3/d Pth,bar

Fig. 6 Example 4. Extended metastable flowrate

Notice that the metastable flowrate remains fairly constant
even though the flowing tubing head pressure fluctuates
significantly.

Example 5. In order to see the bottom hole pressure in
response to liquid loading, Fig. 7 has been constructed. In this
example the flowing tubing head pressure remains fairly
constant at about 13.5 bar. Batch foam was used to kick off
this well periodically.
Production data
0
20
40
60
80
100
28-jan 7-feb 17-feb 27-feb 9-mrt
Q
g
,

x
1
0
0
0
m
3
/
d
0
5
10
15
20
25
30
P
b
h
,

b
a
r
Qg, x1000m3/d Pbh, bar

Fig. 7 Example 5. Bottom hole pressure in response to liquid
loading

As expected, the bottom hole pressure increases sharply upon
liquid loading. A significant metastable flowrate can be
observed as well.

Methods to analyse liquid loading

Study of the behaviour of individual wells confirms that when
certain wells are liquid loading the gas flowrate can drop to a
lower metastable flowrate. Furthermore the data indicates that
the time span, in which liquid loading occurs, varies.

The objective of the modelling work is to determine
what the processes are behind the field observations, and to
quantify them for various given well conditions.

Two methods are presented to analyse liquid loading.
Firstly, a simple numerical model was constructed to simulate
the transient behaviour of liquid loading and its impact on
various parameters. The second method encompasses a
procedure to determine the metastable flowrate analytically,
assuming a steady state situation. Both methods are largely
based on the modelling assumptions described in the next
chapter.

Model construction

To predict how water build-up takes place within a gas well a
simple conceptual model has been created illustrating the
effect of liquid loading, as reported in Dousi
8
. The model
incorporates well behaviour (build-up and drainage of a liquid
column) but does not take into account reservoir behaviour
(depletion and re-pressurisation).


Fig. 8 Schematic of the well model
X
3

X
2

X
1

Gas and Water
Production point
Water Injection point
Depth
Reference point
Water
column
4 SPE 95282
In Fig. 8 the model is depicted schematically. The following
assumptions are made:

Below x
1
the well is assumed to be vertical.
It is assumed that the tubing head pressure is equal to
the surface export pressure when the well is not shut
in.
The reservoir pressure at the production point x
2
equals P
res,g
. At the injection point x
3
the reservoir
pressure equals P
res,w
.
Above depth x
1
the flow behaviour within the well is
described by the model introduced by Cullender-
Smith
9
. At x
1
the pressure is P
fbh1
. The interval below
x
1
is a relatively small part of the well.
Below x
1
the pressure drop due to frictional flow
effects is neglected.
At depth x
2
a gas production point is located. A
simple Darcy model for single-phase gas describes
the inflow into the well. Water is produced together
with the gas, at a constant water gas ratio F
wg
. At x
2

the pressure is P
fbh2
.
At the bottom of the well (x
3
) a water injection point
is located. A simple Darcy inflow model for single-
phase liquids describes the flow into the reservoir. At
x
3
the pressure is P
fbh3
.
A water column can be built up between x
1
and x
3
.
Depth x
1
is always chosen higher than the top of the
liquid column.
The initial gas flowrate can be set to operate in the
liquid loading regime by changing the reservoir
pressure such that the initial gas flowrate is equal to
or lower than the critical rate described by a
simplified Turner critical rate relationship.
Upon liquid loading it is assumed all water co-
produced with the gas (with constant water gas ratio)
will fall back to form the water column.

The Cullender Smith
9
well flow model provides:

2
g
2
fth
2
fbh1
) C(Q ) B(P ) (P + = (1)

where P
fth
is the flowing tubing head pressure,

Q
g
is the gas
production flowrate and B and C are outflow constants
reflecting hydrostatic head and friction respectively. The
relationship relating the pressure P
fbh1
to the pressure P
fbh2
at
the production point is taken as purely hydrostatic, neglecting
any friction effects:

gc1 g wc1 w fbh1 fbh2
h G h G P P + + = (2)

where G
g
and G
w
are the hydrostatic gradients at the bottom of
the well. h
wc1
equals the water column length above the
production point and h
gc1
equals the height of the gas column
between the top of the water column and x
1
. If the top of the
water column is below x
2
, h
gc1
equals the distance between x
1

and x
2
.


The relationship relating the pressure P
fbh1
to the pressure P
fbh3
at the injection point is also taken as hydrostatic:

gc2 g wc2 w fbh1 fbh3
h G h G P P + + = (3)

in which h
wc2
equals the height of the water column length
above the injection point and h
gc2
equals the height of the gas
column between the top of the water column and x
1
. If the
height of the water column is zero, h
gc2
equals the distance
between x
1
and x
3
.

Assuming production and injection take place in the same gas
reservoir at pressure equilibrium:

gc g g res, w res,
h G P P + = (4)

where h
gc
equals the distance between x
2
and x
3
.

The inflow relationship at the production point x
2
is:

g g
2
fbh2
2
g res,
Q A ) (P ) (P = (5)

If P
fbh2
is larger than the reservoir pressure then the production
rate equals zero. The model neglects possible re-injection at
x
2
.

The injection relationship at the injection point x
3
is:

inj w, w w res, fbh3
Q A P P = (6)

If the reservoir pressure is larger than P
fbh3
, than the flowrate
equals zero. The model neglects possible inflow at x
3
.

The water co-produced with the gas is written as follows:

wg g prod w,
F Q Q = (7)

For modelling purposes a simplified form of the Turner
2

critical rate relationship has been used where surface area,
temperature and Z-factor are chosen as be constant such that
the relationship can be written as follows:

fth st c
P C Q = (8)

The Cullender-Smith outflow factors B and C, the gas inflow
resistance factor A
g
, the simplified Turner factor C
st
, and the
average water gas ratio F
wg
, are usually known for a
production well. In addition, the model requires knowledge of
the usually-unknown water re-injection resistance factor A
w
.
This factor is usually determined in an iterative manner by
matching the model results against the actual production data.

The numerical program takes discrete timesteps and
calculates and solves the model equations in an iterative
manner.

SPE 95282 5
Describing the process

The modelled process that takes place in the well during
liquid-loading is depicted in Fig. 9 and can be described as
follows: At time t=0 the well is producing stably at a constant
flowrate. As soon as the tubing head pressure increases above
a certain tubing head pressure, the rate drops below the critical
Turner rate. At this time the water co-produced with the gas
cannot be transported to surface and will drop to the bottom of
the well imposing a backpressure on the reservoir. Due to
water column build up, the pressure at the injection point will
rise until it exceeds the reservoir pressure. Then water will
start to be re-injected into the reservoir, essentially draining
the well partially. The column may build up so that at some
point in time it starts to cover the production interval. At this
time the column will impose a higher pressure on the
production interval as well. The increased pressure will reduce
the drawdown resulting in a drop in gas flowrate and
consequently also a drop in water production rate. A new
stabilized metastable gas rate is reached when the water
production rate becomes equal to the water re-injection rate.
At this point the water column height will stabilize as well.



Fig. 9 Schematic of model liquid loading process

Numerical model calculations

An example of the predictions of the model is given here. It
illustrates the processes that occur during liquid loading.

Example 6. The distance between the injection and production
points equals 100 m. Fig. 10 depicts the assumed tubing head
pressure over time. At time t=0 the well is producing at above
the critical Turner rate. At t=1 day the tubing head pressure is
slightly increased such that the well will produce below the
critical rate. At t=3 days the tubing head pressure is reduced to
its original level. At t=5 days the well is shut in. At t=8 days
the well is brought back onstream.

Tubing head pressure as a function of time
20
25
30
35
40
45
0 1 2 3 4 5 6 7 8 9 10
Time, days
P
t
h
,

b
a
r
a

Fig. 10 Example 6. Assumed tubing head pressure over time

Flowrates & height water column as a function of
time
0
50
100
150
200
250
300
0 1 2 3 4 5 6 7 8 9 10
Ti me, days
G
a
s

f
l
o
w
r
a
t
e
,

x
1
0
0
0
m
3
/
d










H
e
i
g
h
t


w
a
t
e
r

c
o
l
u
m
n
,

m
0
1
2
3
4
5
6
W
a
t
e
r

i
n
j
e
c
t
e
d
/
p
r
o
d
u
c
e
d
,

m
3
/
d
Height water column Gas flowrate
Water production rate Water injection rate

Fig. 11 Example 6. Production and re-injection rates and water
column height over time

Bottomhole pressures at injection and production
point as a function of time
30
40
50
60
70
0 1 2 3 4 5 6 7 8 9 10
Time, days
P
b
h
,

b
a
r
a
Reservoir pressure Pbh injection point Pbh production point

Fig. 12 Example 6. Bottomhole pressures over time.

In Fig. 11 the gas and water production rates, the water re-
injection rate and the water column height as a function of
time are displayed. Fig. 12 displays the pressures at the
injection and production point. At t=1 day the gas and water
production rates immediately reduce, because of the increased
flowing tubing head pressure. Also the water column builds up
to 100 m, the distance between the production point and
injection point. At that point in time the column will start to
cover the production point and the gas flowrate (and water
production rate) consequently decreases further. At t=1.6 days
water will start to be re-injected. This takes place as soon as
the bottom hole pressure becomes higher than the reservoir
pressure at the injection point. At t=2.5 days the height of the
water column stabilizes at 250 m. This happens as soon as the
Gas and water production
Water injection
Water column
6 SPE 95282
water injection rate equals the production rate. At t=3 days the
tubing head pressure is lowered having an immediate positive
impact on the gas flowrate. However the gas flowrate remains
lower than the critical rate. Therefore instead of draining the
water column, the height of the water column increases further
to 280 m. This is because decreasing the tubing head pressure
results in a slight decrease of pressure at both the injection
point and production point. Thus the gas flowrate and water
production rate will increase and the water injection rate will
decrease. Hence water will continue to build up. At t=4 days a
new stable point is reached. At t=5 days the well is shut in and
the water column drains at a constant rate, until the top of the
water column has reached the production point. Then the
water re-injection rate decreases until all the water is drained.

Comparison of numerical model with field data

In order to determine if the model can mimic liquid loading
realistically, the model results have been compared to actual
field data for a well with 3.5 tubing and a 7 production
liner down to a reservoir depth of about 1700 m. A downhole
pressure gauge was installed in this well to monitor the liquid
loading event. In Fig. 13 the actual and modelled gas flowrates
and the modelled height of the water column are displayed.

Gas production rates & Height water column
0
10
20
30
40
50
60
0 5 10 15 20
Time, days
F
l
o
w
r
a
t
e
,

x
1
0
0
0
m
3
/
d
0
50
100
150
200
250
H
e
i
g
h
t

w
a
t
e
r

c
o
l
u
m
n
,

m
Gas flowrate model Gas flowrate field Length water column

Fig. 13 Field and model gas flowrates and water column height
over time

Pressures field data and model
0
5
10
15
20
25
30
35
40
0 5 10 15 20
Time, days
P
r
e
s
s
u
r
e
,

b
a
r
a
Pbh production point model Pbh field Pfth Pfth field

Fig. 14 Field and model pressures over time

The actual and modelled tubing head and bottomhole
pressures are presented in Fig. 14.
Table 1 gives the relevant model parameters, where A
w
and
F
wg
have been varied to provide the best match. In this case A
w

controls the level of the metastable flowrate and the water gas
ratio mainly impacts the time the metastable flowrate is
reached. In the model the liquid loading sequence was initiated
manually after 2 days.

Table 1 - Field Example, Inflow-, Outflow- and Critical
Rate Parameters
A
g
17 [bar2/e3m3/d]
A
w
4 [bar/m3/d]
B 1.4 [-]
C 0.038 [bar2/(e3m3/d)2]
C
st
~12 [e3m3/d/(bar)0.5]

g
0.68 [kg/m3]

w
1000 [kg/m3]
F
wg
6 [m3/e6 m3]
Distance between
production and re-injection point
35 [m]

Figs. 13 and 14 show that the model realistically mimics the
liquid loading process. Note that the bottom hole pressure
gauge is located about 20 meters above the perforated interval.
Therefore when a column of water exists in the well 2 bar
needs to be added to the field data resulting in a better match
with the model prediction in Fig. 14.

Sensitivity analysis

The important properties of a liquid loaded well are:
1. Q
ms
, the metastable flowrate
2. Q
c
, the critical flowrate
3. Water column height
4. Time to reach the metastable flowrate
5. Drainage time at shut in

A sensitivity analysis was carried out to investigate
the impact of changing the input parameters on these
properties. The values in Table 2 present the base case
parameters for an existing well where the B and C outflow
factors represent a well with 5 tubing down to 4000 m. and
A
w
has been determined by matching the model against field
data.

Table 2 Base case, Inflow-, Outflow- and Critical
Rate Parameters
A
g
10 [bar2/e3m3/d]
A
w
7.5 [bar/m3/d]
B 2.2 [-]
C 0.015 [bar2/(e3m3/d)2]
C
st
16 [e3m3/d/(bar)0.5]

g
0.8 [kg/m3]

w
1000 [kg/m3]
P
fth
30 [bar]
F
wg
50 [m3/e6 m3]
Distance between
production and re-injection point
150 [m]

The following parameters were varied: A
g
, F
wg
, P
fth
,
and the distance between the production and re-injection point.
Also A
w
was varied proportionally to A
g
. The critical and
metastable rate, stable water column height, time to drain the
water column and time to reach the metastable water level
were recorded. For each scenario the reservoir pressure is
SPE 95282 7
chosen such that the well in the liquid loaded regime will
initially produce slightly below the critical rate.

In this paper only the sensitivity to A
g
and F
wg
are
highlighted. The complete results are given in Dousi
8
.

Figs. 15 and 16 depict the sensitivity to the gas inflow
resistance factor A
g
.

Flowrate and height water column as a function of
inflow resistance factor Ag
0
20
40
60
80
100
0 20 40 60 80 100
Ag, bar2/1000m3/d
Q
g
,

x
1
0
0
0
m
3
/
d
0
200
400
600
800
W
a
t
e
r

c
o
l
u
m
n
,
m
Qms, Metastable rate Qc, Critical rate Height water column

Fig. 15 Critical rate, metastable rate and water column height for
varying A
g

Shut in time and liquid loading time
0
2
4
6
8
10
12
0 20 40 60 80 100
Ag, bar2/1000m3/d
t
,

d
a
y
s
0
20
40
60
80
100
t

s
h
u
t

i
n
,

d
a
y
s
Time to reach metastable flowrate Drainage time at shut in

Fig. 16 Loading time and drainage time at shut-in for varying A
g


Four interesting conclusions can be drawn:
The metastable flowrate increases rapidly for
decreasing inflow resistance factor A
g
.
The water column grows approximately linearly with
A
g
.
For low A
g
factors the well can load within a period
smaller than 5 days, in agreement with field
observations.
Shut-in times required to drain the well increase
linearly with increasing A
g
factor.

Figs. 17 and 18 depict the sensitivity to the water gas ratio,
F
wg
.
Flowrate and height water column as a function of
Fwg
0
20
40
60
80
100
0 50 100 150 200
Fwg, m3/mln m3
Q
g
,

x
1
0
0
0
m
3
/
d
0
50
100
150
200
250
300
W
a
t
e
r

c
o
l
u
m
n
,

m
Qms, Metastable rate Qc, Critical rate Height water column

Fig. 17 Critical rate, metastable rate and water column height for
varying F
wg

Shut in time and liquid loading time
0
1
2
3
4
5
6
7
8
9
0 50 100 150 200
Fwg, m3/mln m3
t
,

d
a
y
s
0
1
2
3
4
t

s
h
u
t

i
n
,

d
a
y
s
Time to reach metastable flowrate Drainage time at shut in

Fig. 18 Loading time and drainage time at shut-in for varying F
wg


Again four interesting conclusions can be drawn:
The metastable flowrate decreases rapidly with
increasing water gas ratio.
The water column height increases marginally for
increasing water gas ratio.
The time to reach a metastable level decreases rapidly
for increasing water gas ratio.
The drainage time at shut-in is relatively insensitive
to the water gas ratio.

It can be concluded from the sensitivity analysis that wells
which produce significant metastable flowrates typically have
good inflow performance, a low water gas ratio and/or a large
distance between re-injection and production point.

Analytical predictions

The metastable flowrate and the stabilized water column
height can be derived analytically assuming steady state
conditions.

Metastable flowrate. The metastable flowrate, Q
ms,
is the gas
rate at which the rate of co-produced water Q
w,prod
(or F
wg
Q
ms
)
is equal to the amount of re-injected water Q
w,inj
:

Q
w,inj
= F
wg
Q
ms
(9)

Q
ms
and Q
w,inj
are respectively described by:

ms g
2
g fbh,
2
g res,
Q A ) (P ) (P = (10)
8 SPE 95282
and

inj w, w w res, w fbh,
Q A P P = (11)

Assuming the full column between production and injection
point is water filled a relation exists between P
fbh,w
and P
fbh,g
as
follows:

w pi g fbh, w fbh,
G Z P P + = (12)

where Z
pi
is the true vertical distance between the production
and injection points and G
w
is the hydrostatic gradient of the
co-produced water. Further assuming production and injection
take place in the same gas reservoir at pressure equilibrium:

g pi g res, w res,
G Z P P + = (13)

Where G
g
is the hydrostatic gas gradient in the reservoir. Then
Eq. (11) can be re-written as:

inj w, w g w pi g res, g fbh,
Q A G G Z P - P = + ) ( (14)

Given Eq. (9), Eq. (10) and (14) can be solved to calculate
P
fbh,g
:

+ +
= 1
)) ( ( 2
) (
1
2
1
2
2
M
G G Z P
M
P
M P
g w pi g res,
g res,
g fbh,
(15)

where

w wg
g
A F
A
M
2
= (16)

Q
ms
then follows using Eq. (10). For realistic parameters Eq.
(15) can be approximated by:

M
P
G G Z
P - P
g res,
g w pi
g fbh, g res,
1
) (
(17)

where (P
res,g
P
fbh,g
) is the drawdown pressure corresponding
to the metastable flow rate. Also, the metastable rate can be
approximated by:

M
P
A
P G G Z
Q
g res,
g
g res, g w pi
ms
1
2 ) (
(18)

Note that Eqs. (15) and (17) can be generalised to situations
where production and injection take place in different
reservoirs with different pressures by replacing Z
pi
(G
w
-G
g
) by
the pressure difference.

The sensitivity analysis performed using the numerical model
for Q
ms
and h
wc
as function of A
g
has been repeated using the
analytical predictions. The results are identical to the
numerical model results presented in Fig. 15. Note that for
very low A
g
the analytical prediction gives a metastable rate
which exceeds the stable rate while the water column height
required to re-inject the produced water becomes less than the
distance between the production and injection points.

Water column height. The above analysis indicates that the
metastable rate does not depend on the flowing tubing head
pressure. This statement is also supported by the production
data in Fig. 6. Obviously for a metastable rate to occur the
flowing tubing head pressure must be high enough or the
reservoir pressure must be low enough to result in liquid
loading. However if liquid loading occurs the flowing tubing
head pressure no longer influences the metastable rate.

However the effective true vertical height H of the water
column that forms above the production point does depend on
the flowing tubing head pressure. It is derived by matching the
pressure difference between tubing head and flowing bottom
hole pressures as follows:

g w
ms
2
fth g fbh,
G G
C(Q ) B(P P
H

+
=
2
1
2
) ) (
(19)

The total height of the stabilized water column then becomes:

pi wc
Z H h + = (20)

Fig. 19 shows how H varies as function of P
fth
for the base
case parameters in Table 2 where the reservoir pressure is
assumed fixed at 54.5 bar.

True vertical water column height above production point
vs flowing tubing head pressure
0
50
100
150
200
250
20 25 30 35 40
Pfth, bar
H
,

m
Maximum CITHP
for given reservoir
pressure
Maximum FTHP
for stable flow
Metastable
flow regime

Fig. 19 Height above production point versus flowing tubing head
pressure, base case parameters


SPE 95282 9
Four regimes can be distinguished:
P
fth
> 36.7 bar: no gas flow possible as hydrostatic
head of gas column creates overbalance at production
point,
34.6 bar < P
fth
< 36.7 bar: metastable flow is less than
indicated in Eq. (18) as there is insufficient
drawdown pressure available to produce Q
ms
,
accordingly there will be no full water column
between production and injection point,
30 bar < P
fth
> 34.6 bar: metastable flow rate as
described by Eq. (18), the height of the water column
increases as P
fth
decreases, and
P
fth
< 30 bar: stable flow regime with gas rates above
the critical rate, however metastable flow with an
increased water column is also possible. See for
example Fig. 10 and 11 for the period 3<t<5 days)

Discussion of results

Water re-injection. Metastable gas production can only take
place through simultaneous wellbore gas-water separation and
water re-injection. While the gas-water separation is the
simple result of sub-critical flow velocities the re-injection of
water relies on the existence of an effective pressure sink. The
conceptual model used in this paper creates this pressure sink
through gravity by introducing vertical separation between
production and injection points. In reality the situation can be
much more complex e.g. rely on differentially depleted
intervals. The model results indicate that in some situations the
metastable flowrates could exceed the Turner rates thus
effectively eliminating the detrimental impact of liquid
loading. Such cases have been reported (e.g. in sub-horizontal
wells) however appear to be rare e.g. because continuous
water re-injection eventually fills the pressure sink.

Reservoir behaviour. The model assumes the reservoir
behaves as a perfect tank i.e. the reservoir pressure is a
function of cumulative gas production only. For this type of
reservoir and a given flowing tubing head pressure liquid
loading is irreversible i.e. once it occurs stable production
cannot be restored as the reservoir pressure will never recover
to a value sufficient for stable production. In reality the near-
wellbore reservoir pressure will recover after the well is closed
in by filling of the near-wellbore reservoir pressure sink. The
amount and speed of pressure recovery will depend on
reservoir characteristics such as permeability,
compartmentalisation, fracturation etc. In general, lower
permeability reservoirs will show more pronounced pressure
recovery than prolific reservoirs. The presence of pressure
recovery leads to the intermittent well behaviour which
characterises most liquid loading gas wells.

Intermittent production. A large numbers of wells are
operating in the liquid loading regime through intermittent
production. In many cases intermittent production is actively
controlled i.e. the well is shut-in once liquid loading starts and
the well is opened up only after the tubing head pressure has
sufficiently recovered. It is claimed that this type of active
control accelerates and increases gas recovery. One reason
mentioned is that active control should prevent build-up of
long columns of water which take longer to drain away.
Another reason is that it could reduce the potential impairment
caused by re-injecting fresh condensed water into water
sensitive formations. Certainly active control will provide a
means of managing the available well capacity. The numerical
model presented in this paper could be expanded to include
reservoir pressure recovery in order to simulate and optimise
intermittent well production.

Managed metastable production. Metastable production
should ideally be considered as an opportunity to maximise
gas recovery and to reduce surface water processing and
disposal. The model presented in this paper could help mature
this opportunity into practical solutions.

Conclusions

The following conclusions can be drawn from this study:
Production data from mature gas fields clearly show
that gas wells can produce at metastable flowrates
below the minimum stable or Turner flowrate.
A numerical model has been developed which
reproduces the gas-well liquid-loading behaviour
observed in the field, including the metastable
flowrate. The results of the numerical model have
been confirmed by a steady-state analytical solution.
The results of sensitivity analyses show that a
significant metastable flowrate can be observed in a
well with good inflow performance, a low water gas
ratio and a large gas column.
The model can be used to analyse and optimise the
control of intermittent gas well production and to
mature the active use of metastable gas production.
10 SPE 95282
Nomenclature

A
g
= Gas inflow resistance factor, bar
2
/e3m
3
/d
A
w
= Water re-injection resistance factor, bar/m
3
/d
B = Factor related to head,-
C = Factor related to friction, bar
2
/(e3m
3
/d)
2

C
st
= Constant in simplified Turner Eq., e3m
3
/d/(bar)
0.5

F
wg
= Water gas ratio, m
3
/e6m
3

g = Gravitational constant, 9.81 m/s
2

G
g
= Hydrostatic gas gradient, bar/m
G
w
= Hydrostatic gradient of formation water, bar/m
H = Height water column above the production point, m
h
gc
= Height gas column, m
h
wc
= Height water column, m
P
fbh
= Flowing bottomhole pressure, bar
P
fth
= Flowing tubing head pressure, bar
P
res
= Reservoir pressure, bar
Q
c
= Critical gas flowrate, x1000 m
3
/d
Q
g
= Gas flowrate, x1000 m
3
/d
Q
ms
= Metastable gas flowrate, x1000 m
3
/d
Q
w,prod
= Water production flowrate, m
3
/d
Q
w,inj
= Water injection flowrate, m
3
/d
T
fth
= Flowing tubing head temperature, C
x = Position in well, m
Z
pi
= Vertical distance production-injection point, m

g
= Gas density, kg/m
3


w
= Water density, kg/m
3

Acknowledgements

We express our gratitude to the Nederlandse Aardolie
Maatschappij B.V. for their cooperation and permission to
publish this material. Special thanks to Chad Wittfeld for
supplying the bottom hole pressure data.

References

1. Duggan, J.O., Estimating flow rates required to keep
gas wells unloaded, J.Petrol.Tech.,Dec 1961, 1173 -
1176
2. Oudeman P., Improved Prediction of Wet-Gas-Well
Performance, Paper SPE 19103, SPE Production
Engineering (Aug. 1990)
3. Turner, R. G., Hubbard, M.G., and Dukler, A.E.,
Analysis and Pediction of Minimum Flow for the
Continuous Removal of Liquids from Gas Wells,
Paper SPE 2198, JPT (Nov. 1969) 1475; Trans.,
AIME 246
4. Coleman S.B., Clay H.B., Mccurdy D.G and Norrie
H.L., A New Look at Predicting Gas-Well Load-
Up, Paper SPE 20280, J.Petrol.Tech.,March 1991,
329-333
5. Lea, JF and Nickens, H.V., Solving gas-well liquid-
loading problems, J.Petrol.Tech., Dec 1961, 1173
1176
6. Guo, B., Galambor, A., Xu, C., A Systematic
Approach to Predicting Liquid Loading in Gas
Wells, Paper SPE 94081, presented at the 2005 SPE
Production and Operations Symposium held in
Oklahoma City, USA, 17-19 Apr. 2005
7. Sutton R.P, Cox S.a., Williams E.G., Stolz R.P.,
Gilbert J.V., Gas Well Performance at Subcritical
Rates, Paper SPE 80887 presented at the SPE
Production and Operations Symposium held in
Oklahoma City, Oklahoma, U.S.A., 22-25 March
2003
8. Dousi, N., Modelling Liquid Loading and Gaslift for
Gas Wells, M.Sc. Thesis, Delft University of
Technology (June 2004), TA/PW/04-05
9. Cullender, M.H. and Smith, R.V., Practical Solution
of Gas-Flow Equations or Wells and Pipelines with
Large Temperature Gradients, Trans., AIME, (1956)
207

SI Metric Conversion Factors

bar x 1.0* E + 05 = Pa
bbl x 1.589 873 E 01 = m
3

F (F 32)/1.8 = C
ft x 3.048* E 01 = m
ft
2
9.290 304* E 02 = m
2
ft
3
2.831 685 E 02 = m
3

lbm x 4.535 924 E 01 = kg
psi x 6.894 757 E + 00 = kPa
*Conversion factor is exact.