Professional Documents
Culture Documents
Permeability Estimation PDF
Permeability Estimation PDF
0
0 20 40 60 80 100 20 40 60 80 100
Swi (%) Swi (%)
Fig. i-Charts for estimating permeability from porosity and water saturation.
derived techniques can provide level-by- permeability was proposed in 1927 by Koze- lowing relationship:
level (foot-by-foot) permeability values. ny and modified by Carman 3 ,4:
k'h = 1001>e 2 [(I-Swi)/Swi], ....... (2)
Well-test permeability, however, is a direct
k=1>3/[5A g 2(1-1»2], ........... (1)
measurement of the flow that provides per- where 1> e = effective porosity. Determining
meability when the contributing interval is where A g = surface area of grains exposed Swi for rocks that are not at irreducible
known but that lacks foot-by-foot resolution. to fluid per unit volume of solid material. 3 water saturation is difficult, if not impossi-
During the well test, if more than one fluid Eq. 1 describes permeability in packs of uni- ble in some cases. Height above the water
phase is produced, then the calculated per- formly sized spheres, such as powder packs. table alters Sw;' so that even in a single li-
meability may not predict well performance This formulation breaks down in other thology, Swi may vary from top to bottom.
accurately. Unlike most core- and log- sands. The greatest drawback is that A g can In many reservoirs, however, the variance
measured permeability, well-test and RFT be determined only by means of core sam- in Sw; is small, especially if we consider it
methods measure effective permeability. ples, and then only with special care and to be bulk volume. Morris and Biggs 10 ob-
Detailed, accurate reservoir characteriza- equipment. However, this model notes that served that it is generally easier to predict
tion demands the use of various measure- porosity alone cannot reliably predict per- a rock's bulk-volume irreducible water,
ments. Therefore, we need to understand the meability and that it is somehow inversely Vbwi =1>( Swi' than the actual value of Sw;'
various permeability measurement tech- related to the exposed surface area. This requires a slight modification of Eq.
niques used by the industry. Not surprisingly, the first approach to 2, made by multiplying the numerator and
finding permeability from logs did not use denominator by total porosity, 1>( :
Permeability the surface area concept directly. Starting
Measurement Techniques with Tixier,5 Wyllie and Rose 6 conjec- kY2 = 1001>e2 ( 1>( - VbWi). .. ..... (3)
tured that grain surface area was related ap- Vbwi
The three major permeability measurement proximately to irreducible water saturation,
techniques are wireline-log analysis (includ- SWi' of clean sandstone. Timur 7 extended Fig. 1 compares the free-fluid model with
ing the RFT method), laboratory testing of Wyllie and Rose's empiricism on the basis previous versions. Note the differences in
core samples, and well testing. oflaboratory studies of 155 sandstone cores. curve patterns as Swi approaches 100%.
The derivations so far suffer from two Permeability correlations with Swi and 1>
Wireline-Log Measurements. Five meth- limitations: the difficulty of deriving Swi are of limited use in carbonates because of
ods are established for obtaining permeabil- from logs and yielding zero permeabilities low effective porosity (1)e is typically 5 to
ity from wireline tool measurements 1: (1) when Swi approaches 100% and when 20 porosity units), while permeability can
empirical correlation of permeability with porosity approaches zero. These derivations span 0.2 to 2,000 md. 1 Porosity in car-
porosity, 1>, and intergranular surface area; honor only the porosity limit for zero per- bonates may not be intergranular. There-
(2) measurement of producible formation meability approximation and disregard Swi fore, using surface areas from porosity does
fluid with the nuclear magnetism log approaching 100%. The derivation of Swi not work very well. New methods 11,12 cur-
(NMLsM); (3) estimate of mineral concen- from resistivity logs needs a more method- rently being developed for carbonate rocks
trations by the geochemical logging tool ical approach. use core data to define surface area and pore
(GLTsM); (4) correlation of permeability With the introduction of the Coates- path tortuosity better.
with Stoneley wave velocity by acoustic log- Dumanoir 8 relationship of the free-fluid NML Measurements. The NML provides
ging tools; and (5) pressure/time measure- model, a new equation was derived that en- two specific products that can be related to
ment of formation fluids with the RFT tool. sured zero permeability at zero porosity and formation permeability. 13-15 One is the
Empirical Correlations. The first equa- when Sw;=IOO%. Coates and Deno0 9 ac- free-fluid index, IFf' a measure of mova-
tion relating measurable rock properties with commodated the two conditions with the fol- ble fluid (oil and water but not gas). The
~
bility is an accurate representation of a par-
a permeable formation, the Stoneley wave ticular core sample under specific labora-
attenuates by moving fluid in the pores. It tory conditions. Using this permeability
is also dispersed, meaning that different fre- value to represent reservoir formation per-
(To Uppel (To Lower quency components are slowed at different meability can be incorrect. As long as the
Sample Semple
Chamber) Chamber)' rates. measurements are consistent over a partic-
This attenuation and dispersion relates to ular interval, however, the core permeabil-
the formation's permeability, matrix or na- ity can be very useful in completion design,
Fig. 2-RFT tool setup and schematic of tural fractures. Although correlations be-
sampling system. specifically in choosing the phasing and ver-
tween Stoneley behavior and permeability tical spacing of perforation.
other is spin-lattice relaxation time, tl, the have been observed in the field, a quantita- Cores are analyzed on four length scales.
time constant involving the alignment of pro- tive prediction of permeability from Stone- The smallest scale is the sidewall core anal-
ton spin axes along magnetic fields. ley energy measurements has eluded inves- ysis, in which samples < 1 in. long are taken
IFf typically is obtained by applying a
tigators. 23 ,24 Various investigations, how- from the wellbore. The scale of the core plug
large, polarizing magnetic field to the for- ever, continue to use Stoneley waves to is also very similar, in which samples 1 to
mation and then turning it off. Signal de- directly measure permeability and as a frac- 1 V2 in. long are taken from a full-diameter
cay in solids and bound fluids is too rapid ture indicator. 25 core every 6 in. These small sample sizes
for detection with the NML tool. Only de- RFJ'Measurements. The formation-tester can bias the sample in heterogeneous for-
cay in the free fluid can be measured, and tool samples reservoir fluids and measures mations, in which permeability can vary
IFf is proportional to the number of protons formation pressure vs. time at specific depth widely from one sample to the next. Full-
in free fluid. Thus, IFf is related to Swi by stations (see Fig. 2). With the RFT, three diameter core analysis tests 6-in. sections of
sets of data can be collected to quantify per- core and samples on the medium scale. The
Swi=I-(IFf lcp), ................ (4) meability. The first two (in association with largest-scale sampling is whole-core analy-
which can be applied to Eqs. 1 through 3. pretest) are relatively quick, and the last one, sis, in which cores up to 2 ft long are test-
fl is a property of the rock and fluid wet- called superjlow, can last several minutes. ed. The number of whole-core tests is limit-
ting its pore surfaces and thus relates to pore The tool is first positioned to allow mud ed, however, primarily because of the
size. Because permeability is proportionaJ. filtrate and formation fluids to fill a first difficulty in recovering such pieces.
to the square of pore size, it is reasonable _sample chamber at a controlled, low flow Two types of permeability can be meas-
to assume I6 ,17 that k is proportional to t12. rate (first pretest). This step is followed im- ured on core samples in the laboratory: ab-
On the basis of a study of 80 sandstone cores mediately by the filling of a second sample solute and relative core permeability.
from wells worldwide, Kenyon et at. 17 re- chamber at a controlled, high flow rate (sec- Absolute Core Permeability. Two com-
lated tl to k as follows: ond pretest). This phase of the test is called monly used techniques to measure absolute
drawdown. The subsequent phase of the test, core permeabilities are the steady-state and
k= 1.6 x 1O-9tI2.3cp4.3 . .......... (5) buildup, is a measurement of increasing the pulse-decay methods. 29 ,30 The pulse-
NML responses in carbonates differ from pressure once the chambers are filled. The decay method was introduced for low-
those in sandstones. It is not surprising, final phase, superflow, involves long-term permeability rocks where attainment of
therefore, that a reasonable correlation be- drawdown while measuring the cumulative steady state can take anywhere from days
tween tl and permeability has been found in volume and transient pressures. to weeks. In both techniques, air or water
only a few carbonates. 18 Both the drawdown and buildup tests pro- can be used as the fluid medium.
GLT Measurements. GLT measures the vide a permeability value that is often reflec- The cores are cleaned and dried before
concentrations of 10 elements in a forma- tive of near-wellbore fluid movement. 26 To they are measured. In the steady-state tech-
tion by borehole nuclear spectrometry. The calculate permeability, the pressure deriva- nique, air permeability, the most common
basis for obtaining permeability from tive 27 is first plotted to identify the flow measurement, is obtained by placing the
elemental concentrations is that any changes regime and is followed by specialized plots. core in a chamber and measuring the pres-
in mineralogy are accompanied by changes For drawdown pressures, where normal (10- sure differential and stabilized flow rate of
in the size, shape, and morphology of rock cm 3 ) pretest chambers are used, the flow air pumped through the rock. Permeability
grains. 19 These changes affect the pore pattern is typically hemispherical, a mixture is calculated for this single phase with the
system geometry, which directly influences of horizontal and vertical flow with a bias Darcy equation. In the pulse-decay method,
permeability . to horizontal. Buildup permeability typically the core is subjected to a pressure pulse, as
= 162.6BJi :, qi
m I.J ....•..... (15)
k i=! hi
162.6BJi) :, qi
or k= ( I.J - •.•.••.• (16)
m i=! hi
ONEe
The calculated permeability represents the
formation thickness that responds to the
pressure and rate transients. If necessary,
use an ith-Iayer q and h to evaluate the per- Fig. 5-Productlon profile.
meability of the specific zone (provided that
no formation crossflow exists). wireline logs may then be related to the well- found in basins where enough data have
Table 1 summarizes the relationships test formation diffusivity if a means of per- been generated to establish a good value for
among the various permeability techniques. meability averaging is chosen. Whether the the constants in the geophysical equation
averaging should be arithmetic or geometric, (Eqs. 1 through 3). The standard deviation
Interrelationship. The objective of the in- or even harmonic, eludes various reservoir in the 8,850- to 8,950-ft interval, however,
terrelationship presented here is to allow engineering studies. For layers, however, is not good.
meaningful comparisons of permeabilities. the arithmetic average is preferred. 44.45 Note that the wireline-Iog values repre-
The discussions on scale factor, measure- For vertical wells, use of the layer concept sent the hydrocarbon effective permeability
ment environment, and physics show how is an accepted model. Therefore, we pre- at in-situ conditions, whereas the core values
to account for differences in permeability es- sent the following arithmetic averaging represent the absolute permeability at stan-
timates. Here, we develop a simple relation- model to relate core and log permeability
ship that allows meaningful correlations dard conditions. The question to ask is, had
to well-test permeability. we compared apples to apples, would the
among well-test and core and log permea-
N quality of correlation through the 8,850- to
bility measurements.
The strength of the wireline-Iog permea- E Fikih i · .......... (17) 8,950-ft interval be improved? Fig. 6 also
bility data lies in their capability to provide i=! represents the comparison where the core-
continuous permeability throughout a par- data measurements were corrected for in-
ticular interval. In a particular basin, the Field Example. situ pressure conditions 29 and a 0.7
ratios of permeability between various zones Example 1. Fig. 6 compare wireline-Iog relative-permeability effect to calculate
and layers are more valuable than the abso- values (from empirical correlation and Eq. hydrocarbon effective permeability. The
lute values. Such ratios can be used to corre- 12 for effective permeability) and laboratory quality of comparison has certainly im-
late with core permeability as long as the measurements at standard ambient condi- proved throughout all the intervals, especial-
scale factor, environment, and physics are tions. The quality of comparison (standard ly through the 8,850- to 8,950-ft interval.
adequately addressed. The layer-by-layer (or deviation) through the 8,550- to 8,700-ft in- The average zone well-test permeability is
foot-by-foot) permeability from cores and terval is good. Such comparisons are often indicated by the bar. With use of produc-
_......
Environment & PhysIcs should be closer to the typical well-test per-
meability. The well-test permeability of the
Pressure &
Technique Scale Temperature SaIu'- ....hod QuantIty zone of interest in the entire basin averages
""'.
AbOOIuto
Maoro<coplc
Ambient· """"...
Relative
Direct
Direct
Panneabillty
Penneabillty
near tOO md. This indicates good basin-wide
agreement. Any agreement, however, be-
tween relatively macroscopic/megascopic
Relative
Wlreline-log
K-+
NML
Megascopic
,
In-situ
AbsoM.
AbooIuto
Indirect
Indirect
Permeability
Permeability
measurement (superflow RFT) and giga-
scopic measurement (well-test) occurs if the
formation or zone is homogeneous. Open-
GLT In-sltu AbsoIuto Indirect Permeability hole logs suggest that the formation is
S_
80_
RFT
"""'"
In-situ
In-situ
,,....
In-situ
AbsoM.
""""...
Relative
Oirect?
0''''''
0'"",
PonneabIHty
ConductIvIty
ConductIvIty
homogeneous, at least vertically (see Fig. 7).
Recommend.tlons
....,,
Well-test Gigascoplc In this study, we concentrated our efforts
Short Relative Oirect Conductivity on defining the correlation axioms between
Classic Relative Direct Conductivity the various permeability sources. Identifi-
""""nood In-situ Relative 0;"", Conductivity cation of areas where the various techniques
stand out uniquely and where any correla-
tion attempt may be futile was not part of
the discuss ion. For example, zone-by-zone
tion profile data, the correlation can be com- log and the RFT pertain to in-situ conditions; permeability from geophysical logs and/or
pleted with Eq. 17, as shown in Fig. 6. however, the pretest RfT technique re- cores may not be quantitatively representa-
sponds to invaded fluid movement and the tive of the formation flow potential and in
Example 2. Fig. 7 pertains to wireline-Iog- supe rflow test pressures respond to places may be uncorrelatable, but these
derived permeability using volumetric anal- hydrocarbon fluid movement. For this rea- values have unique applications (e.g., in the
ysis. RFT data at 6,887 ft includes pretest son, the wireline-log average oil effective decision process to identify the perforation
drawdown and buildup and subsequent long- permeability (265 md), when compared to interval, perforation density, or perforation
term superfJow data. Both the geophysical the RfT results, should correlate better with phasing). We recommend that future studies
concentrate on uncorrelatable sources to
identify unique applications for them. Also,
considering the vastness of the technical area
discussed here, we suggest that studies be
.~
performed to enhance the interrelationships
among the various sources (e.g., in forma-
tions with nonuniform layers or zones).
Conclusions
I. The interrelationships among the geo-
physical wireline-log, core-analysis, and
weJl-test permeabilities depend on three im-
portaru factors: measurement scale, environ-
ment, and physics.
2. Too many correlations are made
wilhout proper regard to these factors, re-
sulting in inadequate answers. Integration
of available information pertaining to these
factors enhances correlation between the
various techniques.
3. Transient well-test data provide the best
quantitative formation permeability if pro-
duction proftle of the formation is available
and single-phase flow is maintained.
4. Geophysical-log and core-analysis per-
meabilities define layer-by-layer (or foot-by-
foot) permeability profiles.
I~
c;')
•• _
~
0 011 1'" lOa ~"'4." '. ,", ~.
WATEII oL, ~---,~- ...•., Bound Wst&r
."
D • , " I- ~ _G.!.'l"!... ~O_ Hydrocarbon IWatlf Matrix
r .. 0.-.. -8 ICAL-BS) 12
>; '- I- .... Silt
Gamma Ray : Watar Saturation Effactlve
ci 0 160 10" 0.1 laO "
Poro.ltv
0 60 " 0
Clay
-:.
I'- /
E~~::,.:.
~
I!. ..-:: la
t:3: ___
~:-.:-::
.:;t..
!) I
-, , I.... \.