Professional Documents
Culture Documents
This study concluded that the scale of the heterogeneity determining the uncertainty of a petrophysical measurement,
was small (averaging slightly less than 1 in.), producing heterogeneity is explicitly ignored.
significant heterogeneity at the scale of the core plug, but To address the sample-to-sample dispersion caused by
having little impact on the response from the logging tool. heterogeneity, we introduce the “heterogeneity index,”
This has significant implications for other aspects of defined as a “Parameter, associated with the result of a
carbonate petrophysics: measurement that characterizes the dispersion of the values
• Comparison of core and log properties such as porosity that could reasonably be attributed to the measurand in
and permeability must be done with care, because the response to different volumes within some larger reference
impact of heterogeneity is variable from zone to zone. volume.”
This work predicts the relative impact of heterogeneity The heterogeneity index is intrinsically scale-dependent. It
on core plugs, minipermeameter points, and logs. is related to the measurement volume, the spatial variability of
• The fact that standard wireline logs are insensitive to the property, and the size of the reference volume. Thus, the
most of the heterogeneity in these formations explains overall uncertainty in the mean value of the reference volume
the difficulties in determining RRTs from standard logs, is a combination of the measurement error and the
because the properties distinguishing the RRTs from heterogeneity index.
one another (macrofossils and molds) are below the For example, if a density logging tool with a measurement
resolution of standard logs. volume of 12 in. vertically and 3 × 4 in. horizontally, with a
This analysis was performed in one dimension directly measurement error of one porosity unit (p.u.), records a
around the wellbore. It was assumed that the heterogeneity is measurement of 15 p.u. when straddling the boundary between
at least locally isotropic in three dimensions; in particular, that a 10-p.u. bed and a 20-p.u. bed, the measurement error is still
there is no vertical variation in formation properties over the 1 p.u., but the heterogeneity index will be 5 p.u. The
scale of an inch or two. This assumption fails in the presence uncertainty in the formation porosity measured at the bed
of bed boundaries that do not intersect the wellbore at an angle
boundary will then be ± 5 p.u. ( 1 + 5 ), and will drop to
2 2
of precisely 90°. Those cases are characterized by locally large
heterogeneity indices and a large correlation length. 1 p.u. when the center of the sensor is more than 6 in. from the
Fortunately, this is not a common occurrence in the data of bed boundary.
this study. Thus, given a porosity measurement with volume of
The effect of heterogeneity on standard 1-in. core-plug investigation v , the heterogeneity index is defined as:
measurements may be understood by considering the porosity σ hi (v) ; .......................................................................... (1)
readings that would be obtained from a large number of plugs
taken from the formation at a single depth, from all sides of and a porosity measurement tool with volume of
the hole using a wireline core driller. In a heterogeneous investigation v and inherent measurement error of σ me (where
formation, the porosity measured in each plug will vary
considerably, whereas a relatively homogeneous formation σ denotes ±2 standard deviations), will then have an overall
will yield a narrow range of measured porosity. uncertainty of measurement in the reference volume given by:
Log measurements often yield relatively smoothly varying
values even over intervals in which the core indicates σ Φ (v) = σ me
2
(v) + σ hi2 (v) ....................................... (2)
significant heterogeneity. If the measurement volume of these
different observations is correctly accounted for, then the Porosity logs for a particular tool (e.g., density porosity),
heterogeneity index analysis introduced here allows us to or porosity measurements (e.g., core plug) can then be
quantitatively reconcile these two seemingly inconsistent expressed with heterogeneity index envelopes, which are
observations. discussed in the next section.
The heterogeneity index is useful when comparing the
Heterogeneity Index vs. Measurement Uncertainty porosity measured by two sensors with different volumes of
Uncertainty is defined as a “Parameter, associated with the investigation; e.g., a core plug (small volume) and a porosity
result of a measurement that characterizes the dispersion of the log (larger volume). The volume investigated by the porosity
values that could reasonably be attributed to the measurand.”2 log is taken as the reference volume and the range of potential
In petrophysics the measurand is the sensor(s) (such as core porosity values measured by a hypothetical set of plugs
radiation detectors or electronics in a porosity-logging tool; taken from within that volume is estimated and displayed.
helium pycnometer or calipers in a core laboratory). The Note that the “normalized heterogeneity index” is defined
uncertainty associated with the measurement is caused entirely as:
by inherent dispersion, whether random (nuclear counting σ hi (v)
statistics, for instance) or systematic (imperfect calibration, a σ hi (v) = ........................................................... (3)
design defect, or wear and tear), in the sensor. For clarity in Φ
this document, we call this source of uncertainty the
“measurement error.” where Φ is the mean porosity.
Although all porous rock samples are heterogeneous at The remainder of this paper will cover a more detailed
some scale (e.g., at the scale of grains and pores), when analysis of the results of applying this technique to a particular
well in a Cretaceous carbonate reservoir in Abu Dhabi: Well
AD-1.
SPE 88788 3
Quantitative Heterogeneity Analysis from Image Data Generation of the heterogeneity logs involves the
To understand heterogeneity, the technique presented here following steps7:
takes advantage of the high-resolution data provided by 1. Begin with a borehole image—a 2D array of pixels
resistivity imaging tools within the borehole. However, on the borehole wall.
because we are interested in understanding the heterogeneity 2. Compute the 3D Cartesian coordinates of each pixel
observed in porosity variability, we must transform the from hole geometry, tool geometry, and tool
resistivity image to a porosity image. orientation.
The recognition that borehole images can be mapped to 3. For a given frame (depth), identify all possible two-
petrophysical properties is well understood.3–4 Perhaps the best point pairs of pixels and the chord lengths between
established transformation is that from resistivity to porosity them (lags).
space using the Archie saturation equation.5 4. Assign pixel pairs to bins representing similar lags.
5. For each frame, compute the experimental
aRmf semivariogram by computing the Φ variance for
S wn = , …………………………………………(4)
Φ m Rxo each lag bin.
6. Derive a model semivariogram from the
where S w is the saturation of the wetting phase (0.00 to 1.00), experimental semivariogram for each frame, thereby
determining the modeled geostatistical parameters
n is the saturation exponent, a depends on the tortuosity and making them available for further processing or
(0.35 to 4.78), Rmf is the resistivity of the mud filtrate, Φ is display.
the porosity (0.00 to 1.00), m is the cementation exponent 7. Using the geostatistical parameters that have been
modeled at the scale of the input measurement,
(1.14 to 2.52), and Rxo is the resistivity of the flushed zone. compute the same parameters for measurement with
This relationship can be rearranged as follows: larger volumes of investigation using upscaling
1 approaches.
⎛ aR ⎞ m 8. The upscaled geostatistical model parameters can
Φ = ⎜ n mf ⎟ then be used to generate heterogeneity index logs for
⎝ S w Rxo ⎠ ..........................................................(5) corresponding core and logging measurements.
A carbonate porosity solution application transforms the Examples of semivariograms generated for different
resistivity image data into porosity ( Φ ) using Eq. 5, in which intervals in the well are illustrated in Fig. 1 and Fig. 2. In
addition to borehole microimaging and core-plug
the remaining parameters are either input by the user or
measurement volumes, in this analysis we also considered two
derived from other logs. Applying a 1.2-in. vertical window to
logging tool measurement volumes: the high-resolution and
these data generates and analyzes a porosity distribution
low-resolution porosity tools. The modeled dimensions of
histogram for every depth (0.1-in. vertical spacing).4 This
these various measurement volumes are listed in Table 1, and
technique has proven very powerful in identifying some
conceptually illustrated in Fig. 3.
aspects of porosity distribution such as dual porosity. Note,
however, that this technique effectively collapses the image
Heterogeneity Index Logs for Porosity Measurements
data into a histogram, discarding spatial information (other
In the model semivariogram, the total sill for the porosity
than depth). It does not consider the 3D geometry of the
electrical image sensors in the borehole. measurement volume v is equal to the sample porosity
variance ( σ v ) (see Clark8); namely:
2
In the technique presented here, borehole images are
treated as 3D measurements of the geological and
petrophysical properties of the rock. The technique then uses σ v2 = Cv0 + Cv . .............................................................. (6)
the established approaches of geostatistics6 to model the 3D
geological and petrophysical heterogeneity of the rock in the From this, the standard deviation ( σ v ) of the porosity
region of the borehole. measurement related to the heterogeneity of the porosity is
These geostatistical parameters, which represent the given by:
statistical variability at the scale of the borehole image
measurements, are then used to model the statistical variability σ v = Cv0 + Cv ............................................................ (7)
at measurement scales representing larger volumes of
investigation (e.g., core plug, porosity log, and resistivity log). Hence, the heterogeneity index (Eq. 1) is defined as:
These modeled parameters can then be used to describe the
impact of heterogeneity on these larger-scale measurements in σ hi ( v ) = 2σ v . .............................................................. (8)
representing the mean formation properties and differences
expected between them caused by their differing volumes of Eq. 8 can then be combined with Eq. 2 to generate
investigation. heterogeneity index logs of the porosity measured by the
As discussed above, one product of this analysis is a series porosity logging tool ( Φ ); namely,
of single-channel logs describing the geological and
petrophysical heterogeneity of the borehole. Φ ± σ Φ ( v ) ................................................................ (9)
4 SPE 88788
Heterogeneity Analysis Results for Well AD-1 We can use the computed porosity heterogeneity envelope
The computations of heterogeneity index were performed on a to compute a heterogeneity envelope for permeability. The
study well from Abu Dhabi. The results are summarized in minimum permeability is computed by substituting porosity
Figs. 4 and 5. The measured heterogeneity at the electrical minus the porosity heterogeneity index for porosity. The
image scale is in the range of 4 to 6 p.u.—in excess of 20% of maximum permeability is the same, using porosity plus the
the porosity—for a large number of depths. Higher values are porosity heterogeneity index. The other terms in Eq. 10 are
also observed, some of which are errors in the data sampling held at their nominal value for the depth. Since these
caused by bed-boundary effects in which data in the analysis parameters may vary as well, this computation yields a lower
are from different beds. The histogram of the correlation limit of the heterogeneity index of the permeability. In Fig. 11
lengths shows that the vast majority of the correlation lengths we used the heterogeneity index for porosity at the image scale
are below 1 in. The upscaled histograms of the heterogeneity to compute a range of permeabilities within which we expect
index show significant variability at the scale of core plugs the minipermeameter readings (taken every inch) to fall if they
and very little impact on the wireline log volume. This reflects are consistent with the log-derived permeability.
a correlation length that is close to the size of the core plug but
Conclusions
much smaller than the logging tool’s volume of investigation.
We have demonstrated that geostatistical analysis can be
Track 3 of Fig. 4 contains the porosity log (red) and the
applied to borehole image data to generate an improved
modeled heterogeneity index (blue). The heterogeneity index
understanding of formation porosity heterogeneity.
is combined with the porosity log to generate a heterogeneity
Specifically, we have demonstrated how to:
envelope (grey region) for the log porosity. In contrast, at the
scale of the core plugs (Track 2), the heterogeneity index • Generate an experimental variogram from high-resolution
(blue) is much larger, and mirrors the greater “scatter” in the borehole images.
core plug measurements (magenta circles and black diamonds • Extract model parameters from experimental variograms.
in Track 1). • Upscale the geostatistical model parameters to larger-
The plot in Fig. 6 combines the core-plug scale measurement volumes.
heterogeneity index with the log-scale heterogeneity index to • Transform the borehole resistivity images into porosity
predict the maximum expected scatter in plug porosity values. values.
Statistically, 95% of the core-plug measurements will fall This technique can be applied to complex carbonate
within this envelope if the deviation of plug porosity from log reservoirs to:
porosity is caused only by heterogeneity. Fig. 6 illustrates the • Characterize different intervals in terms of their
transition between a homogeneous carbonate and the most heterogeneity.
heterogeneous interval in the well—a rudist debris rock type • Explain apparent differences between porosity
(RRT 8). Note how the heterogeneity envelope precisely measurements derived from core plugs and logs.
reconciles the scatter of the core-plug measurements with the To contain small-scale formation heterogeneities within
smoothly varying log measurements. the core-plug volume, we recommend using larger-diameter,
Analysis of the geostatistical parameters becomes very 1 1/2-in. core plugs to measure petrophysical parameters in the
interesting when the data are classified by reservoir rock type.1 study reservoir.
Figs. 7 to 9 illustrate the range in mean porosity and References
heterogeneity index for the RRTs encountered in the study 1. Russell, S.D., Akbar, M., Vissapragada, B. and Walkden, G.M.,
well. The variation in the heterogeneity index between “Rock types and permeability prediction from dipmeter and
different rock types hints at a potential application for this image logs: Shuaiba reservoir (Aptian), Abu Dhabi,” AAPG
technique as a new input to RRT classification. Bulletin (2002) 86, 1709.
2. ISO/IEC/OIML/BIPM. "Guide to the expression of uncertainty in
Heterogeneity in Permeability measurement," 1992. 1-91.
Minipermeameter measurements were performed on selected 3. Delhomme, J.P., Bedford, J., Colley, N.M., Kennedy, M.C.:
core slabs from the subject well.9 The measurements were “Permeability and Porosity Upscaling in the Near-Wellbore
Domain: The Contribution of Borehole Electrical Images,”
performed on a 1-cm grid on 4-in.-wide core slabs that were 6
(1996) paper SPE 36822.
to 12 in. long. We computed experimental variograms for 4. Newberry, B.M., Grace, L.M., Stief, D.D.: “Analysis of Carbonate
these data and compared the results with the porosity Dual Porosity Systems from Borehole Electrical Images,” (1996)
variograms from the image log over the same intervals SPE 35158.
(Fig. 10). The two are related because the length scale is the 5. Fanchi, J.R.: “Shared Earth Modeling,” (2002) Elsevier 306pp.
same and the sill values correlate. 6. Bloomfield, P.: “Fourier Analysis of Time Series: An
Hassall et al.9 propose an equation to compute permeability: Introduction,” 2nd Edition (2000) Wiley 261pp.
7. Tilke, P., Allen, D., Gyllensten, A.: ”Quantitative Analysis of
k = 0.5 Φ2 (ρ T2LM)2 (Φ / (Φ - Vmacro)) 2 ........................(10) Porosity Heterogeneity: Application of Geostatistics to Borehole
in which: Images”, Mathematical Geology (in press)
8. Clark, I.: “Practical Geostatistics,” (1984) Elsevier 129pp
k = permeability (md) 9. Hassall, J.K.; Ferraris, P., Al-Raisi, M., Hurley, N., Boyd, A., Allen,
Φ =Porosity (fraction) D.F: “Comparison of Permeability Predictors from NMR,
Vmacro = Volume of macroporosity (fraction) Formation Image and other Logs in a Carbonate Reservoir,” paper
ρ = Surface relaxivity (µm/sec) SPE 88683 to be presented at the 2004 Abu Dhabi International
T2LM = Log mean average of T2 distribution (msec). Petroleum Exhibition & Conference, October 10–13.
SPE 88788 5
*
6 SPE 88788
Φ = 50%
Φ =11% Φ = 22%
Core Plug
Core Plug
Wellbore
Density
Log
Fig. 4 Depth log display of the computed heterogeneity index (blue lines) and heterogeneity envelope (grey areas) for
measurements with the volume of a 1” core plug (center track) and the low resolution logging sensor (table 1) in the
right track. Core plug porosity measured every foot is plotted in the left track so that its scatter may be compared with
the prediction in the center track. The correlation length at the scale of the electrical image is shown in the depth track.
The colored band on the right differentiates two different reservoir rock types, the one at the top, above 155 ft., (red) is a
rudist debris and is more heterogeneous.
8 SPE 88788
Electrical Electrical
image image
0 5 10 15 0 1 2 3 4
Fig. 5a Histogram of Heterogeneity Index at Fig. 5b Histogram of the correlation length at
electrical image scale for the 700’ zone of the electrical image scale for the 700’ zone of
interest in well AD-1. (2 x standard deviation) interest in well AD-1. In inches
p.u.
Core Plug
Density
Log
0 5 10 15 0 5 10 15
Fig. 5b Histogram of Heterogeneity Index at Fig. 5d Histogram of Heterogeneity Index at
core plug scale for the 700’ zone of interest in low resolution log (density) scale for the 700’
well AD-1. (2 x standard deviation) p.u. zone of interest in well AD-1. (2 x standard
deviation) p.u.
SPE 88788 9
0.4
Core Plug Porosity (v/v)
0.35
Core Plug
0.3
0.4 Tw in Core Plug 0.0
0.25
Porosity (%)
Heterogeneity Envelope (v/v)
0.2
0.15
0.4 0.0 0
RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT
1 1a 2 2s 3 4 5 5a 8 11 12 13 17
RRT
0.9
0.8
0.6
0.5
0.4
0.3
0.2
0.1
0
RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT
1 1a 2 2s 3 4 5 5a 8 11 12 13 17
2.5
2
Range (in)
1.5
0.5
0
RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT RRT
1 1a 2 2s 3 4 5 5a 8 11 12 13 17
RRT
3 0.002
) 0.0016
2
) )
Variance (k
K)
2
2
Variance ( (Φ
10
2
0.0012
Variance (log
Variance
1 0.0008
0.0004
0
0 2 4 6 8 0
Lag (in) 0 2 4 6 8
Lag (in)
0.002
3
0.0016
) )
(k2)
2
Variance (log K )
Variance ( (Φ
2
2
0.0012
10
Variance
Variance
0.0008
1
0.0004
0 0
0 2 4 6 8 0 2 4 6 8
Lag (in) Lag (in)
3 0.002
0.0016
Variance K(k) 2)
) )
2
2
Variance ( (Φ
2
2
10
0.0012
Variance (log
Variance
0.0008
1
0.0004
0 0
0 2 4 6 8 0 2 4 6 8
3 0.002
0.0016
(k2)
Variance (log K )
))
2
2
2
10
Variance(Φ
0.0012
Variance
Variance (
0.0008
1
0.0004
0 0
0 2 4 6 8 0 2 4 6 8
3
0.002
Variance (k2)
0.0016
)
Variance (log 10K2)
2
Variance ( (Φ
2
2
)
0.0012
Variance
1 0.0008
0.0004
0
0
0 2 4 6 8
0 2 4 6 8
Lag (in)
Lag (in)
Fig. 11 Predicted core permeability range based on the heterogeneity in porosity compared with core plug
(black) and minipermeameter (red) measurements from well AD-1.