GAS WELL

TESTING

1
The solution to the diffusivity equation, is applicable for slightly
compressible fluids (i.e. liquids) with relatively constant fluid
properties such as viscosity.
However for flow of real compressible gases in the reservoir,
the solution to the diffusivity equation requires adjustment
because the physical properties, such as viscosity, isothermal
compressibility and the gas deviation factor, are strong
functions of pressure.

2
The diffusivity equation for gas through porous media is:

 p 
 Z 
1   r p p     
r r  Z r  k t
This equation was derived by combining the conservation of
mass equation, Darcy’s law and the following EOS for gas
density:
pM

ZRT
3
When the left-hand side of the diffusivity equation for gas is
differentiated, the term (∂P/∂r)2 is not negligible as in the case
of slightly compressible fluids. Therefore the diffusivity equation
for gases is a non-linear partial differential equation.

Al Hussainy et al. introduced a transformation called the real

gas pseudo-pressure so that the pressure, the viscosity and
gas deviation factor could be combined in a way that allows the
linearization of diffusivity equation.

4
The real gas pseudo-pressure is defined as follows:
p

m p  2 
p
dp
po
g Z
This transformation accounts for variations in gas properties
with pressure. po is an arbitrary reference pressure.
From this equation, we have:

m p  m p  p 2 p p
 
t p t  g Z t
m p  2 p p

r  g Z r 5
The substitution of the last two equations in diffusivity equation
gives:
 m p  1 m p   g ct m p 
2
 
r 2
r r 2
k t
This equation has exactly the same form as the diffusivity
equation for slightly compressible fluids, with the pressure
replaced with the pseudo - pressure m(p).
The close analogy between diffusivity equations for gas and for
liquids suggest that the solution for the flow of real gases
should correlate as function of dimensionless variables based
on initial or average values of viscosity and compressibility.

6
Wattenbarger and Ramey showed that the pressure transient
equations could be used, with very good approximation, in
terms of m(p), i.e.
 rD2 
mD  p   p D
1
  Ei 
  

2  4t D 

Using the log-approximation to the Ei-function at the flowing

well, this equation becomes:

mD  pwf  ln t D  0,80907  2S '

1
pD 
2

7
Where S’ is the total skin factor that includes the skin resulting
from true formation damage or stimulation, S, and D is
turbulence or non-Darcy coefficient:
S '  S  Dqsc

The effect of D is to create an increasing apparent skin for

higher production rates.
The value of D varies with pressure but for simplicity it can be
considered constant as an acceptable approximation.

8
In real units the log-approximation to the Ei-function becomes:
   
m pwf   m pi   1637 qscT  kt   3,23  0,869 S '
 log
  c r 2 
kh 
  g t w  


Thus a plot of m(pwf) versus time will yield a straight line portion
corresponding to the infinite-acting radial flow regime of slope
mp, which can be used to calculate permeability:

1637 qscT
k
mph

9
The total skin factor S’ is then calculated from:

 m p   m p   k  
S '  1,151 i 1hr
 log    3,23
 m   c r 2  
 p  g t w  

The reduced pseudo-pressure at time t=1 hour, m(p1h), should

be obtained from the semilog straight line portion (extrapolated
if necessary).

10
NON - DARCY FLOW

Darcy’s law applies to gases flowing at low rates (laminar flow),

which occur in the reservoir away from the wellbore. However,
in the vicinity of the wellbore, the flow rates of gas can become
extremely high because of the converging flow.
As these rates, inertial effects can become important and
Darcy’s law no longer applies.
The inertial effects take the form of distorted flow paths and
also turbulence in different locations in the pore structure.
Although the exact nature of this microscopic flow is not known
in the reservoir, the net effect is a higher-pressure gradient
when these inertial effects become considerable.
11
For laminar flow, Darcy’s law can be rearranged in the following
form:
p 
 u
x k
Where ∂p/∂x is the pressure gradient and u is the macroscopic
(Darcy) fluid velocity.
At higher rates, when the inertial effects become important, a
quadratic relation first introduced by Forchheimer is used:

p 
  u   u 2
x k

12
The right hand side of this equation contains a term for viscous
forces and a term for inertial forces, both of which contribute to
the pressure loss.

The parameter β (ft-1) is called either the Forchheimer

turbulence factor or inertial factor and can be determined
experimentally or estimated from one of the following
correlations:

4,851*10 4
1,88 *1010 4,11*1010
    0,53 1, 47  
 5.5 k  k k 4/3
Last equation is applicable only to sandstones, dolomites and
limestone.
13
The parameter D (D/Mscf) is called the non-Darcy flow
coefficient or turbulence coefficient.
It can be estimated from:

15
 g k
D  2,22 *10
 gi rw h
The non-Darcy coefficient D and mechanical skin S can be
determined more accurately if two consecutives flow tests are
run at two different rates, yielding:

S1'  S  Dqsc1
S 2'  S  Dqsc 2

14
As discussed above, accuracy is improved for semilog analysis
tests by replacing pressure with real-gas pseudo pressure
function, m(p), which is expressed in psi2/cp.

For log-log analysis (type-curve analysis), particularly of

wellbore storage distorted data from both flow and buildup test,
accuracy also is improved by replacing time with adjusted
pseudo time, tap(p), which is expressed in hr-psi/cp:

t
t ap  p  
dt
0
 g ct

15
For convenience, although not by necessity, m(p) and tap can
be normalized to have units of psia and hours, respectively, like
the original variables, p and t.
Normalization also gives the pseudo-pressure and pseudo-time
variables magnitudes comparable with those of the
untransformed pressure and time.
The unnormalized variables m(p) and tap typically have values
of 105 to 108.
Reference values of pressure used for normalization are
arbitrary.
Some engineers prefer properties evaluated at initial reservoir
pressure, other prefer at the average reservoir pressure.

16
Lee and Wattenberger define normalized adjusted pressure,
pan, and normalized adjusted time, tna, as:

1  g z 
pna   
 m p  t na   g c t  t ap
2 p 
In terms of adjusted variables, the unsteady state equation for
gas flow is:
   
162.6q g B    kt  
pna, i - pna, wf  

log    3.23  087S '

 
kh    g c t rw2  
   

17
Where the gas formation volumetric factor at standard
conditions of pressure (14.7 psi) and temperature (60°F), Bg
(bbl/scf), is calculated at the average or initial properties of Z, T
and p, from:

ZT
Bg  0.00504
P
Note that the product qscBg is in bbl/D, similarly to the product
qBo in the oil case.

18
The similarity between equations for oil and gas indicates that
all the methods for interpreting a pressure transient test in an
oil well are also applicable in a gas well.

For semilog analysis of PBU tests, adjusted pressure and

adjusted time should be used, but the adjusted producing time,
tap used in the Horner time ratio is evaluated at current
average drainage area pressure, and therefore tap=tp
Adjusted shut-in time, ∆tna is evaluated from the integral, i.e.

d t 
t
t na   g c t 
0
 g ct

19
With µg and ct evaluated at shut-in bottom hole pressures, pws,
at values of shut-in times ∆t.

For semilog analysis of a PDD test, adjusted pressure and

actual flowing time should be used.

20
PRESSURE AND PRESSURE-SQUARED

Use of adjusted time and adjusted pressure in formulating

equations for analysis of transient tests in gas wells in not
always necessary.
Pressure data may be plotted in terms of p2 rather than m(p) if
reservoir pressures are less than 2000 psi.
Also, pressure data may be plotted directly in terms of p if the
reservoir pressures are above 3500 psi.
If in doubt whether to plot p or p2 or normalized adjusted m(p),
use m(p).
Wattenbarger and Ramey have proposed a simple method to
determine the best mode for plotting the data.
They suggest that the variation of the product µgZ with p
examined over the pressure range of interest.
The following is a graph of the variation in the product µgZ with
pressure and is typical of many gases.
If the variation in the µgZ product is small, then they suggest
that a p2 graph would suffice.
On the other hand, if the µgZ product is a linear function of
pressure, then a p-graph would be adequate regardless of the
magnitude of the pressure gradient.
If neither of these relations is applicable, then the m(p)
approach should be used.
Thus, the procedure for graphing data depends primarily on the
variation of the product µgZ.

At high reservoirs pressures, i.e. p >3500 psi, the plot µgZ

versus pressure is almost a straight line, thus:
p p pi
 cte  
g zg g z g  gi z gi
Thus for p0=0, real gas pseudo-pressure equation can be
expressed as:
 2p  p  2p 
m p      p
  z 
dp 
 z
 g  po  g 

Thus, the normalized adjusted real pseudo-pressure, becomes:

 g z    g z  2 p 
  
pna  p   m( p )  
 2p    z  p  p
    g 

And the unsteady state equation becomes:

   
162.6q g B    kt  
pi - pwf  

log    3.23  087 S '

 
kh    g c t rw2  
   
At low pressures, i.e. less than 2000 psi, the product µgZ is
almost constant, as can be observed in the µgZ vs P plot.
Thus:
 g Z  cte   g Z   gi Z i

Thus for p0=0, real gas pseudo-pressure equation can be

expressed as:

 2 p  1  2
m p      p
  z 
pdp 
 z
 g  pn  g 
The normalized adjusted real pseudo-pressure, becomes:

 g z    g z  p2  1  2
pna   m( p )   
 2p   z   p
2 p 2 p
    g  

And the unsteady state equation becomes:

   
1637q gT Z    kt  
p2 - p2  

log    3.23  087 S ' 

i wf  
kh    g c t rw2  
   
In order to test the accuracy of these three methods, Azis et al.
calculated the sandface pressure for 25 different gas well tests
conditions form reservoirs in Alberta, Canada.
They concluded that:
1. The use of average gas properties results in more accurate
solutions than the use of initial conditions. This is true
regardless of the approach used
2. When gas properties must be assumed constant at the
initial values, the use of the m(p) function yields the most
reliable results
3. In very low permeability gas reservoirs, i.e. tight gas
reservoirs, the m(p) function must be used, especially if the
reservoir is produced at very high rates.

Again, all of the rules developed for PDD tests to identify near
wellbore effects (skin flow efficiency and wellbore storage
coefficient) and the start of semilog straight line for oil Wells are
also applicable for gas Wells, regardless of which function is
used.
 CONVERTING p TO m(p)

There are several methods for converting pressure data to real

gas pseudo-pressure data:
a. The areal summation method, which involves plotting 2p/μZ
versus pressure, the area under the curve is the value of m(p)
b. The reduced properties correlation, which requires tables or
charts
c. The numerical integration method is the more common
method and will be illustrated by the following example. The
step by step procedure of the numerical integration method is:
1. Obtain viscosity and gas deviation factor. If PVT data are not
available, empirical correlations can be applied.
2. For each pressure value, calculate 2p/(μZ)
3. Estimate the mean value between the former and current
value of 2p/(μZ)
 2p   2p 
 
 Z   
 

 2p    j  Z  j 1

 Z 
 
  mean 2

4. Estimate the pressure difference between the former and

current pressure value:

p  p j  p j 1
5. Multiply results form steps 4 and 3. Thus,
*
 2p   2p 
 
 Z   
 Z 
 p
 j   mean

6. Estimate pseudo-pressure as:

*
 2p 
m p  j 
 Z 
  m( p ) j 1
 j
EXAMPLE 1
Given the viscosity and gas deviation factor data reported
in following table of a gas which has a specific gravity of
0.76. estimate the pseudo-pressure function, m(p).
viscosity,
p, psi cp Z
200 0,01117 0,9768
400 0,01176 0,9545
600 0,01236 0,9322
800 0,01297 0,9131
1000 0,0136 0,8946
1200 0,01423 0,8778
1400 0,01488 0,8631
1600 0,01555 0,8507
1800 0,01622 0,8409
2000 0,0169 0,8338
2200 0,01759 0,8295
2400 0,01828 0,828
2600 0,01898 0,8292
2800 0,01969 0,8329
3000 0,02041 0,8389
 SOLUTION

p viscosity Z 2P/μZ (2P/μZ)mean Δp (2P/μZ)mean*Δp m(p)

200 0,01117 0,9768 36660,735 18330,36748 200 3666073,496 3,67E+06

400 0,01176 0,9545 71269,9957 53965,36532 200 10793073,06 1,45E+07 7,00E+08
600 0,01236 0,9322 104148,658 87709,32666 200 17541865,33 3,20E+07 y = -0,0082x3 + 96,165x2 - 2752x + 632645
6,00E+08 R² = 1
800 0,01297 0,9131 135101,964 119625,311 200 23925062,2 5,59E+07
1000 0,0136 0,8946 164385,003 149743,4836 200 29948696,72 8,59E+07 5,00E+08
1200 0,01423 0,8778 192136,894 178260,9483 200 35652189,65 1,22E+08
1400 0,01488 0,8631 218018,819 205077,8565 200 41015571,31 1,63E+08 4,00E+08
1600 0,01555 0,8507 241904,057 229961,4382 200 45992287,65 2,09E+08
3,00E+08
1800 0,01622 0,8409 263941,268 252922,6624 200 50584532,49 2,59E+08
2000 0,0169 0,8338 283864,704 273902,9861 200 54780597,22 3,14E+08 2,00E+08
y = 56,658x2 + 49463x - 2E+07
2200 0,01759 0,8295 301557,717 292711,2107 200 58542242,14 3,72E+08 R² = 0,9996
1,00E+08
2400 0,01828 0,828 317128,088 309342,9025 200 61868580,5 4,34E+08
2600 0,01898 0,8292 330405,937 323767,0124 200 64753402,48 4,99E+08 0,00E+00
2800 0,01969 0,8329 341467,558 335936,7474 200 67187349,48 5,66E+08 0 1000 2000 3000 4000
3000 0,02041 0,8389 350427,396 345947,477 200 69189495,39 6,35E+08
TIAB DIRECT
SYNTHESIS
1. Wellbore Storage Coefficient

 0,42qscT  t 
C 
 

 m( p ) 

 i   lpu

 0,42qscT  t 
C 
 

 t * m( p )' 

 i   lpu

36
2. Permeability

711,26qscT
k 
ht * m' ( p ) r

qsc= Mscf/d T = °R

3. Skin factor
S '  S  Dqsc
 m( p ) r  ktr  
 0,5 
 ln    7,43

 t * m ' ( p ) r   c  r
t i w
2 
 

37
4. Drainage Area

 2,355qT  t pss
A
  c  h

 t * m' ( p ) 
 t i  pss

38
EXAMPLE 2
Given the viscosity, Z factor data in Example 1, the
pressure drawdown data presented in the table and the
information given below, determine:
• Reservoir permeability,
• Total skin factor,
• Forschheimer turbulence factor,
• Mechanical skin
g= 0,76 t, hr pwf, psia
0 5000
T= 255 F 0,021
0,042
4407
4121,6
0,084 3829,2
q= 5000 Mscf/d 0,146 3639,6
0,209 3556,4
0,418 3464,6
porosity= 10% 0,626 3432,1
0,835 3410,4
Cti=13,1*10-5/psi 1,044
1,46
3394
3372,1
2,088 3347,4
Pi= 5000 psia 4,176
6,246
3306
3278,4
8,352 3261,7
h= 12,4 ft 10,44 3247,8
12,528 3236,6
14,616 3225,5
rw= 1/3 ft 16,704 3217,2
18,792 3211,6

gi=0,0253 cp 20,88
41,76
3203,2
3185,7
62,64 3161
Swi=65,5% 100,44 3104,5

Bg= 0,000728 bbl/scf

SOLUTION
Since the initial pressure is greater than 3500 psia, the
pressure test can be analyzed using the p-function.
However during the flow test the pressure dropped below
3500 psia, so it maybe advisable to use the m(p)
function.
USING THE m(p) FUNCTION
1. Convert the flowing pressure pwf data to m(pwf)
(Example 1). (Use the quadratic equation)
2. Plot m(pwf) vs log t
3. Calculate K
1637 qscT
k
mph
4. Obtain Skin factor
 m p   m p   k  
S '  1,151 i 1hr
 log   3,23
 m   c r 2  
 p  g t w  
5. Obtain Forschheimer turbulence factor and non-Darcy
coefficient
4 
 g k
 
4,851*10 D  2,22 *10 15

 5.5 k  gi rw h

6. Calculate the additional skin due to turbulence, D*qsc

7. Obtain the mechanical skin factor

S  S '  Dqsc
 1.600.000.000

1.400.000.000

1.200.000.000

1.000.000.000
K=7.11 md
mp=57627531,3 S’=11,06
m(p)

800.000.000
β=5751343460 ft-1
600.000.000
D=0,00066014 D/Mscf
p1h=798616726 S= 7,76
400.000.000

200.000.000

0
0,01 0,1 1 10 100 1000
t
USING THE PRESSURE FUNCTION
1. Plot pwf vs log t
2. Calculate K
162.6qsc Bg  g
k
mh
3. Obtain Skin factor

p p  k  
S '  1,151 i 1hr
 log    3,23
 m   c r 2  
 g t w 
 5000

4500

4000
Pwf, psia

K=8.88 md
m=135,94 S’=7,71
3500
S=4,022

p1h=3388,85
3000

2500
0,01 0,1 1 10 100 1000
t, hr
USING TDS
1. Plot log (t*(∆mp)’) vs log t
2. Calculate K

711,26qscT
k 
ht * m' ( p ) r

3. Obtain Skin factor

S '  S  Dqsc
 m( p ) r  ktr  
 0,5 
 ln    7,43

 t * m ' ( p ) r   c  r
t i w
2 
 

1,0E+10

1,0E+09

K=7.67 md
S’=10.6

1,0E+08

1,0E+07
0,01 0,1 1 10 100 1000
¿QUESTIONS?