You are on page 1of 16

Permeability - It's Variations

CONTENTS
1 INTRODUCTION
2 AVERAGE PERMEABILITIES FOR SEVERAL
LAYERS
2.1 Beds in Parallel
2.2 Layers in Series - Linear Flow
3 MODELLING HETEROGENEOUS SYSTEMS

1
LEARNING OBJECTIVES
Having worked through this chapter the Student will be able to:

Appreciate and understand that permeability is an anisotropic property.

Derive equations to enable the calculation of average permeability for; layers


in parallel-linear and radial flow, layers in series linear and radial flow.

Describe briefly the impact of layered reservoirs in the context of modeling


water oil displacement in heterogeneous reservoirs.

Permeability - Its Variations

1 INTRODUCTION
Although we have defined permeability as a rock property, seldom, if ever, is a
homogeneous reservoir encountered in actual practice. In many cases the reservoir
will be found to contain several distinct units or layers of varying rock properties.
Even on a local scale the value of permeability is not necessarily the same in all
directions. Permeability is an anisotropic property, (Figure 1) i.e. its value is
dependent on direction. Porosity is an isotropic property however.
Ky

Figure 1
Permeability An Isotropic
Property.

Kz

Kx

The sedimentary nature of rocks is such that vertical permeability is less than
horizontal permeability and horizontal permeabilities in the principal directions will
also be different.
On a reservoir scale, thin streaks of very low permeability material can reduce the
effective vertical permeability to a value lower than the actual rock values would
indicate. Whereas core analysis represents microscale observations, data obtained
from well tests represent macroscale behaviour.
Figure 2 represents the variation in permeability observed using a mini-permeameter
in samples from the Leman gas field.

2
1.0

Scale 1
(cm)
0

15.5
1.0

1.0

12.0

17.5

2.5

>0.5

a
b

Figure 2
Effect of scale of
observation and
measurement
on permeability data from a
Rottliegende aeolian sand
cross bed set in the Leman
gas field 1.

36.5

"Sample plug"
Poros 17.5%
Perm. 19 mD

15.0

c
d

1.0
e
<0.5
38.5
21.0

f
>0.5

Location of minipermeability measurement

g
20.0

Permeability in mD

Department of Petroleum Engineering, Heriot-Watt University

1
Although the conditions for use of Darcy's Law state that the rock should be
homogeneous and isotropic, in reality reservoirs do not conform to this restraint. If one
examined the variation say in a core as illustrated in figure 3, very large variations in
permeability occur and vary according to the scale of measurement. Conventional
core analysis takes a sample at around 1 per foot, probe permeability is able to sample
at much closer intervals. A well test result can reflect the permeability over tens of feet.
There is considerable effort taking place now in developing up-scaling methods for
representation of permeability for different applications.

Log Permeability

300

180

25

160

200

Figure 3
Statistical Analysis of Rock
Property Evaluation

Waren and Price2 demonstrated that the most probable behaviour of a heterogeneous
systems tends towards that of the geometric mean.

k G = ( k1x k 2 x k 3 ...k n )

1/ n

(1)

Values for average permeability can be generated by considering the formation being
made up as a composite with different layers. There are two main types of layering to
be considered: linear and radial.

2 AVERAGE PERMEABILITIES FOR SEVERAL LAYERS


Simple geometry systems can be dealt with as follows:

2.1 Beds in Parallel


The horizontal system reflects the sedimentary nature of the rock: the rock material
may have been segregated as it was deposited giving different sizes, shapes etc. to
different layers in the formation. The reason for determining an average permeability
is in rationalising the permeability measured on small samples in the laboratory with
4

Permeability - Its Variations

the measurements made for example by well test analysis. Well test analysis cannot
test small sections of the reservoir (which would be uneconomical) on the same scale
as the laboratory tests. The results are therefore representative of flow through several
layers rather than only one.

Linear Flow
Consider the simple linear beds in parallel. Figure 4.
P2

P1
Q1
A1

Q2

K1

h1

A2

Q3

A3

K2

h2

K3

Figure 4
Radial Flow in Parallel

h3

The average permeability can be developed using the Darcy flow equation:

Q T = Q1 + Q 2 + Q 3
Q1 =

k1A1 ( P1 P2 ) k 2 A 2 ( P1 P2 )
,
...etc.
L
L

Q T = k A i ( P1 P2 ) / L

k1A1 ( P1 P2 ) k 2 A 2 ( P1 P2 ) k 3A 3 ( P1 P2 )
+
+
L
L
L

k=

kiAi
Ai

(2)
(3)

(4)

(5)

(6)

If all the beds have the same width the A h so k is the arithmetic average:

kA =

k h
h
i

(7)

This equation is commonly used to determine the average permeability of a reservoir


from core analysis.

Radial Circular Flow


This is the case of several superimposed layers flowing simultaneously in the well.
Each layer supplies a rate of Qi. The total rate of flow is QT = Qi. In Figure 5:

Department of Petroleum Engineering, Heriot-Watt University

1
pe

re

rw
pw

Qi =

Q1 K1

h2

Q2 K2

h3

Q3 K3

hT

Figure 5
Radial Flow in Parallel

2h i k i ( Pe Pw )
r
ln e
rw

Q = Qi =

k=

h1

h k
i

(8)

2 h T k ( Pe Pw ) 2 ( Pe Pw )
=
( k1 + k 2 + k 3 + .......)
r
r
ln e
ln e
rw
rw

(9)

(10)

hT

This value can be compared with that obtained through well flow tests or pressure
build-up tests.

2.2 Layers in Series - Linear Flow


In this case the reservoir may have been severely folded or faulted and the originally
horizontal layers are now vertical. It is assumed that the flow is now through each of
the layers towards the well. In this case, assuming a three layered system, the total flow
rate is constant through all of the layers and the total pressure drop is now the sum of
the pressure drops across each layer. Similary, the total pressure drops across each
layer. Similarly, the total length is the sum of the lengths of the individual layers, and
the area open to flow is constant. Figure 6 illustrates.
P1

P2

P4

P3

Q
A
K1

L1

K2

L2

K3

L3

Figure 6
Linear Flow in Series

Permeability - Its Variations

The average permeability of linear beds in series is obtained by adding the pressure
drop across each bed.
(P1 - P4) = (P1 - P2) + (P2 - P3) + (P3 - P4)

P1 P4 =

(11)

QL1 QL 2 QL 3 QL
+
+
=
k 1 A1
k 2A2
k 3A 3
kA

(12)

For beds of equal cross-sectional area then:

k=

L
L
ki
i

(13)

This mean is the harmonic average permeability kH.

Radial Circular Flow


In addition to natural lateral variations in permeability, wellbore damage can reduce
the permeability in the vicinity of the wellbore; also cleaning techniques, such as
acidising can increase the permeability in the vicinity of the wellbore.
For example in Figure 7:
re

r1
rw
pw
Pe

P1
k1

k2

Figure 7
Radial Flow in Series

Q1 =

Q2 =

2 k 1 h( P1 Pw
r
ln 1
rw
2 k 2 h( Pe P1 )
r
ln e
rw

(14)

(15)

Total flow:

QT =

2k avg .h( Pe Pw )
r
ln e
rw

Department of Petroleum Engineering, Heriot-Watt University

(16)
7

1
Total pressure drop:
i.e.

( Pe Pw ) = ( Pe P1 ) + ( P1 Pw )

(17)

re
r
r
Q 2 ln e Q ln 1
rw
rw
r1
=
+
2 k avg. h
2 k 2 h
2 k1h

Q T ln

(18)

At steady-state flow:
QT = Q1 = Q2

(19)

ln re / rw ln r1 / rw ln re / r1
=
+
k avg.
k1
k2

(20)

k avg. =

ln re / rw
ln r1 / rw ln re / r1
+
k1
k2

(21)

3 MODELLING HETEROGENEOUS SYSTEMS


Permeability variations can often be traced from well to well throughout the reservoir,
there by enabling a layered reservoir system to be developed. Many fields demonstrate
this layering phenomenon, leading to very large variations in permeability.
For example Figure 8 shows the considerable variation in permeability for the various
sand units making up the Brent sands in the North Viking Graben area of the North
sea. Using an average value for the permeability can lead to large errors and
misleading results in reservoir modelling.
Lower
ness
11
10

Tarbert

Upper ness

Etive

Rannoch

9
8
7

Permeability (Darcies)

6
5
4
3
2
1

9200

9300

9400

Measured Depth (Feet)

9500

Figure 8
Permeability Variation in
North Sea Field

Permeability - Its Variations

In some cases it is not possible to correlate permeabilities from well to well, and it is
more difficult to put together a reservoir model to be used to examine flow behaviour
with such a reservoir.
Modelling reservoirs with average reservoir properties can only be valid if:
reservoir sands are homogeneous
random variations in reservoir properties occur across the sand
ordered distributions of the properties observed in one well do not correlate with
other wells.
Representation of permeabilities as a function of depth used to be presented on a log
permeability -versus- depth scale. Such a representation can lead to an observation
that an average permeability might be appropriate when in reality when presented in
a true linear form the differences in permeability are more distinct. Fluid behaviour
in reservoirs obeys a linear law, as against log permeability!
Figure 9 shows the permeability variation in a reservoir when plotted on a linear scale
as against the log scale, for a section of a reservoir. It would be easy to wrongly
estimate an average permeability looking at the log permeability presentation, which
from the linear permeability presentation such as average clearly cannot be interpreted.

Permeability Distributions
10

20

30

40
Log (Permeability mD)

20
40
Normal log scale

60
80
Thickness (ft)

1000 2000 3000 4000 5000


Permeability (mD)
20
40
60

Figure 9
Permeability Distribution
on a Log and Linear Scale.

Linear scale

80
Thickness (ft)

Department of Petroleum Engineering, Heriot-Watt University

1
The difference in, for example, the behaviour of a waterflood when modelled as a
homogeneous system as against a layered system is illustrated this process is shown
schematically by figure 10.
The example demonstrates that for these heterogeneous systems with large
permeabilities contrasts, that in water flooding water flows preferentially down the
high permeability section.
The significance in these simplifications is demonstrated in the context of modelling
the behaviour of a waterflood, when modelling as a homogeneous system as against
a layered system. The illustration presented in the subsequent example is that of the
waterflooding of a reservoir with the permeability contrasts as illustrated in figure 9.
Figure 10 illustrates the process where injected water sweeps through the reservoir.
In the example given the mobility ratio of the process is less than 1. That is the injected
water is less mobile than the displaced oil. If the reservoir was homogeneous then the
permeability can be considered as an average value and the displacement simulated
on a one dimensional basis. The key issues for the reservoir engineer are; when does
the water injected arrive at the producing wells and secondly what is the oil recovery.

Gas
Limited flaring
Re-inject
qo sales

Separation

Injection
Pump
qwi

qo + qwp
qwp
Purify /
Dump /
inject

Sea Level

Seawater
for injection

Sea Bed

Reservoir

Figure 11 gives the one dimensional reservoir simulation process and figure 12 the
predicted outcome of oil production, water-cut and oil recovery. As is shown the
piston like displacement arrives at the production well around 2500 days after the start
of injection at which the oil recovery is around 48%. A very good project if this is an
accurate simulation of the process.
10

Figure 10
Water Injection Process

Permeability - Its Variations

_
k
average

Oil

Water

Figure 11
1-D Displacement Process

WATER CUT
1.0
0.8
0.6
1D
0.4
0.2
0

TIME (days)
2000

4000

6000

w = 0.5 cp

OIL RATE
1.0

2D =

kv = kh
kv = 0.1kh

1D =

1D
0.5

TIME (days)
2000

RECOVERY FACTOR
0.5

4000

6000

49.7%

0.4
1D
0.3
0.2

Figure 12
Oil Rate, Water Cut and
Recovery

0.1
0

TIME (days)
2000

4000

Department of Petroleum Engineering, Heriot-Watt University

6000

11

1
Figure 13 a,b, c and d shows the process for a layered case where the permeability
variations have been included in the simulation and the process is now two dimensional. The permeability distribution is as in figure 9 with the highest permeability in
the centre. Figure 13a shows the simulation after 600 days and the values on the
contour lines are values of water saturation. The figure clearly identifies the water
moving through the high permeability zone. In figure 13b after 1200 days water has
broken through into the producing well and this well is 'cutting' water over a short
interval. After 2400 days, figure 13c, the high permeability layer is taking the majority
of the water and a larger interval of the producing well is 'cutting' water.

0
50
20
70
600 Days

40
50

Figure 13 (a)

60
1000

2000

3000

w = 0.5 cp
kv = kh

50
20
70

1200 Days

40
50

Figure 13 (b)

60
0
50%
20
70%
2400 Days

40
Impact of gravity 50%

Figure 13 (c)

60
1000

2000

3000

0
50%

w = 0.5 cp
kv = kh

Attic oil

20
70%

4800 Days

40
50%
60

12

Figure 13 (d)
Displacement simulation in
layered reservoir

Permeability - Its Variations

It is also interesting to note an increase in the water saturation along the base of the
reservoir figure 13c and 13d. This is due to the impact of gravity, the density difference
of the denser water causing the water to move towards the base of the reservoir. After
4800 days this gravity segregation perspective is clearly seen as the low permeability
zone is being swept by this gravity impact. At the top of the reservoir no such benefit
is generated and oil is unswept. This unswept oil is sometimes termed attic oil.
Figure 14 gives the oil rate, water cut and recovery for the simulation where the
various permeability layers have been identified in the 2-D simulation. They show that
the project is not as attractive as that forecast by the previous 1-D simulation. The two
lines in the 2 D case show the impact of vertical permeability, a significant perspective
in relation to gravity flow.

WATER CUT
1.0
0.8
0.6
1D
0.4

2D

0.2
0

TIME (days)
2000

4000

6000

w = 0.5 cp

OIL RATE
1.0

2D =
1D =

1D
0.5

kv = kh
kv = 0.1kh

2D

TIME (days)
2000

RECOVERY FACTOR
0.5

4000

6000

49.7%

48.8%

0.4

39.0%

1D

2D

0.3

Figure 14
Predictions for
Homogeneous 1D and
Heterogeneous Flooding
2D

0.2
0.1
0

TIME (days)
2000

4000

Department of Petroleum Engineering, Heriot-Watt University

6000

13

1
Clearly the main impact is that the field starts producing water after around 1200 days
around half the time of the 1D prediction and the recovery at breakthrough is only
around 25%.
The example clearly demonstrates that for these heterogeneous systems with large
permeability contrasts, that in water flooding water flows preferentially down the high
permeability section.
Premature water breakthrough occurs resulting in deferment of oil production
anticipated from an average value of permeability.
With such a result the reaction is what can be done to improve the process? The
immediate reaction is not to complete the high permeability layers, forcing the fluids
through the more restricted lower permeabilities. This suggestion is illustrated in
figure 15, where the simulation clearly shows that there is little impact. Near the
injection point there is evidence of some displacement into the lower permeability
zone, but once into the formation the water finds the easiest route the fluids move
through the central high permeability zone. The only way to impact the displacement
would be to reduce the permeability of the high value layer some distance into the
formation. The use of time setting polymers could provide such a fluid diversion.
Clearly there are technical risks associated with such a process.
1000

2000

3000

0
With high permeability layers not complete

20

50%
70%

40
50%
60

The recovery of the attic oil is a challenge. Gravity segregation based methods have
been suggested where the injection of a light fluid, for example nitrogen, would have
a similar impact on the unswept oil in the upper layers as the water has on the base
layers. In the example presented it has been assumed that there is strong pressure and
flow communication across the layers if this were not the case then the flow profiles
would be significantly different.
Although the example of water flooding has been used, the phenomena will also occur
in gas injection schemes, being therefore very relevant to the development of gas
condensate reservoirs by gas cycling.
The example has illustrated the importance of permeability contrast in a formation.
The topic is further covered in some depth in the chapter on immiscible displacement.

14

Figure 15
Water Injection Profile
When High Permeability
Layer Not Completed.

Permeability - Its Variations

Figure 16 a and b illustrate the geological process which result in the permeability
decreasing with depth (16a) and the permeability increasing with depth (16b) 3.
Considerable activity in the nineties was focused on the impact of a range of
geological scenarios on permeability and the development of realistic reservoir
simulation models. The subject of upscaling, unheard of in the seventies is now an
integral part of building realistic reservoir flow models.

Coarser Sediment in
Shallow Turbulent Water

Increasing
Wave Energy

Sea Level

ize

nS

rai

G
ing

as

re

Inc

Fine Sediment in
Deeper Quiet Water

us

Advance
of Bar

eo

an

lt
mu

itio

os

p
De

Si

GR
Profile (A)

GR
Profile (B)

Injection

Permeability

Oil
Water

Depth

Figure 16a
Effect of Favourable
Permeability in
Waterflooding.

Production
(a) Favorable

Levee
Eros i o

n of

Channel
Migration

Fin

Weak
Current

Ba
nk

Strong
Current

on

siti

po

e
dD

an

Growth of
Point Bar
al
ter
La face
r
Su

e
ars

Co

tion )
e
cre
Ac e Lin
m
(Ti

GR
Permeability
Injection

Depth

Figure 16b
Effect of Unfavourable
Permeability in
Waterflooding 3.

Oil

Production
(a) Unfavorable

Water

Department of Petroleum Engineering, Heriot-Watt University

15

1
REFERENCES
1. Van Veen, F.R., "Geology of the Leman Gas Field Petroleum and the Continental
Shelf of N.W Europe." (Woodland, A.W., ed.) Supplied Science Pub. Barkins
1975. 223
2. Warren, J.E., Price, K. S., "Flow in Heterogeneous Porous Media" Society of
Petroleum Engineering Journal 1961. 153 - 169.
3. Archer, J. S., Wall, C. C., Petroleum Engineering. Graham and Trotman 1988
London

16

You might also like