Vous êtes sur la page 1sur 9

CHAPTER 3

F U L L I N T E R P R E T A T I O N

The quicklook analyses presented in Chapter 2 will be sufficient for oper-


ational decisions on the well. Usually the results are presented by making
a clear print of the evaluated logs at scales of 1:200 and 1:500 with the
sums and averages marked on the logs and the porefluids marked using
appropriate colors. All companies will use blue for water, but some prefer
red for oil and green for gas, while others prefer red for gas and green for
oil.
Once the final data and prints have been received from the logging
contractor, the digital data should be stored within a corporate database.
Normally at this point the petrophysicist will do a full interpretation,
which might be revised as further core analysis or information from offset
wells becomes available.
In some cases the quicklook Archie model might be completely set
aside in favor of a more advanced model, as described later in this book.
In other cases it is sufficient merely to refine the conventional Archie
interpretation. In this chapter the ways the Archie model may be refined
will be discussed.

3.1 NET SAND DEFINITION

If core data have been acquired, it is essential that the petrophysicist


pay a visit to the core shed at the earliest opportunity to inspect the slabbed
core. This will provide a check that there are not anomalous zones that
have been wrongly allocated to reservoir or nonreservoir status in the
interpretation. Where reservoir can be easily identified, one should make
measurements of the core to ascertain the exact net sand footage that can
be checked against the calculations made on the logs.
In order to match the net sand footage calculated from the logs with
that seen on the core, the shale volume (Vsh) cutoff may be varied. Core
photographs will be taken under both normal and UV light, which can
also assist in the determination of net reservoir. Once the conventional
core analyses have been completed, one will have regular measurements
of core porosity, grain density, and permeability.
If measurements at overburden conditions have been performed, then
the conversion factors to convert porosity and permeability to in-situ con-
ditions should be established. If they are not available, one should assume
values based on regional data until special core analyses (SCALs) are
completed.
In-situ porosity vs. logarithm of permeability should be plotted, if nec-
essary dividing the data according to facies and/or formation such that a
single line can be fitted to the data with reasonable accuracy. This yields
the so-called poroperm relationship, which is usually of the form (in
millidarcies [md], porosity as fraction):

k = lOA(ka+kb*ty) (3.1.1)

where k = permeability of the reservoir. Typical values of ka and kb are


- 2 and 20, respectively.
Using the Vsh cutoff chosen, it should be the case that the net sands
should not contain porosities much below a level corresponding to 1 md
permeability in oil zones and 0.1 md in gas zones. If this is not true, then
it may be necessary to apply an additional porosity cutoff to exclude tight
zones, which are not picked up purely by a Vsh cutoff.
Where core data are not available, it is sometimes helpful to plot the
gamma ray (GR) vs. the density log to help to establish the best point to
discriminate net from non-net from the GR log. Typically the plot will
show a behavior as shown in Figure 3.1.1.
As shale becomes dispersed in the pore space (increasing GR), the
density will rise until the point at which the pore space available for free
fluids becomes zero. Beyond this point, the amount of shale may still
increase until the formation becomes 100% shale, but the density will
change only slightly (depending on variation in density between quartz
and shale). The correct cutoff point is therefore the point at which the
gradient changes, corresponding to zero effective porosity.
If radioactive minerals are present in the sands, deriving Vsh from the
GR alone will not be appropriate. In such formations it is recommended
to use purely a porosity cutoff. In the case of thinly laminated sands, it is
log data
Density (g/cc)

model
GRsa
GRsh
cutoff point

GR (API)
Figure 3.1.1 Determining Reservoir Cutoffs Using a GR-Density Crossplot

possible that the entire formation interval will be designated as nonreser-


voir using a Vsh or porosity cutoff. In this situation it is recommended to
not apply any cutoffs whatsoever. The Archie approach will no longer be
appropriate, and advanced techniques should be adopted.

3.2 POROSITY CALCULATION

In most cases the density porosity, with an appropriate choice of fluid


density, is still recommended. However, a calibration against the conven-
tional core analysis, corrected to in-situ conditions, should be made. The
core data should be depth-shifted to match the logs and plotted together
with the calculated porosity. A histogram should be made of the core grain
density measurements to determine the appropriate value to use in the
sands. Note that it is not appropriate to include plugs taken in clearly non-
reservoir sections within the analysis. The histogram should provide the
mean grain density, as well as give an indication of the likely possible
spread of values that could be encountered.
The next step is to make a crossplot of the log density against the in-
situ core porosity values, as shown in Figure 3.2.1:
When the core porosity is zero, the density should be equivalent to the
core grain density. Also, when the core porosity is unity, the density should
be equivalent to the fluid density. The normal procedure is to fix the line
core porosity

Core porosity
rhog=2.66, rhof =
1.02

log density (g/cc)

Figure 3.2.1 Core Calibration of Porosity

through the data so that the core grain density is honored, and then to
extrapolate the line to the point at which core porosity is unity to deter-
mine the appropriate fluid density. Note that this has to be done separately
in any gas, oil, and water legs. In theory the fluid densities thus derived
should be close to those assumed during the quicklook analyses. Differ-
ences might occur due to:

1. Slight miscalibration of the density log


2. Effect of certain mud chemicals (e.g., barite) on the density log
3. Invasion being less or more than previously assumed
4. Problems with the core plug measurements or conversion to in-situ

Whatever the reason for the apparent fluid densities being what they
are, the combination of the assumed grain and fluid densities, when
applied to the density log, will at least ensure that the log porosities match
the core densities. Where the fluid densities are anomalous, one would
probably want to use them in only the current well, and possibly only over
the cored interval. If, however, the densities agree with the values
expected, then they may also be applied with confidence in other wells
drilled using similar drilling parameters.

3.3 ARCHIE SATURATION

SCAL data measurements of cementation (m) and saturation (n) expo-


nents should be incorporated into the Archie model. In m measurements,
the plugs will have been flushed with a brine of an equivalent salinity to
that expected in the reservoir and the resistivity measured. By plotting the
logarithm of formation factor, given by log(F) = \og(Ro/Rw), against
log(porosity), according to Archie:

(3.3.1)

Therefore, the gradient of the line gives m. Note that the higher the m
value used, the higher the water saturations, Sw, that will be calculated,
and vica versa.
In n measurements, the plugs will have been flushed with brine, then
desaturated (either with air or kerosene) to yield measurements of true
resistivity, Rh vs Sw. By plotting the logarithm of the resistivity index,
given by log(7) = log(Rt/Ro), against 1Og(SJ, according to Archie:

(3.3.2)

Therefore, the gradient of the line gives n. Note that the higher the n value
used, the higher the Sw that will be calculated, and vica versa. Values of
n that are anomalously high (above 2.5) may be indicative of a mixed or
oil-wet system and require further investigation. Low values of n corre-
spond to good-quality water-wet permeable rock.
Having set m and n, there is no longer complete freedom to choose
Rw if one is required to calculate Sw - 100% in known water sands. If
formation-water salinity is well known from produced water samples, one
is sometimes faced with a dilemma of whether to honor m or Rw. In many
cases, the true cause of this discrepancy is actually an error in the por-
osity calculation. However, where the porosities are robust, one has to
make a choice whether to change m or Rw. It is always worth looking again
closely at the cementation-exponent measurements to see how much
scatter in the data there is and whether or not the m value chosen is really
reliable. If the measurements do not come from the water leg at all, it is
possible that diagenetic effects in the reservoir mean that values from the
oil leg are not representative. However, it may also be the case that the
Rw value is not robust for reasons highlighted in Chapter 2 (Section 2.5).
With respect to the value of Rt to be used, one needs to decide whether
invasion or shoulder-bed effects are significantly affecting the deepest
reading resistivity tool. For a well drilled with OBM (oil-based mud) that
encounters thick sands, I would recommend simply using the deep resis-
tivity as it is. Where there are significant effects of invasion or shoulders,
then generally I would always recommend going with a saturation/height
approach in favor of Archie.
If it is decided to still try to correct the resistivity for such effects, then
the contractor's chart books may be used for making the appropriate cor-
rections, or computer-based algorithms may be applied. Remember that
such corrections apply equally in the water leg if one is using a Pickett
plot to determine m and Rw.

3.4 PERMEABILITY

For the final evaluation, a permeability log, as well as zonal averages,


will usually be required for input to the static and dynamic models. Using
the poroperm relationship described in Section 3.1, it is relatively simple
to derive a permeability log using the porosity log. However, once the log
has been derived, it is important to scrutinize it for any intervals for which
the permeability goes to an anomalously high value. Most sandstones do
not exceed about 1500md, although top-quality sands with porosities
above 35% may have permeabilities up to about 4000 md. If necessary,
apply a cutoff to cap the permeability at a level that is supportable by the
core data. In the nonreservoir sections, the permeability should usually be
set to a very low value (e.g., 0.001 md). Permeabilities calculated should
be roughly in line with those calculated from other sources, such as a for-
mation pressure tool, NMR (nuclear magnetic resonance), or production
tests.
For making zonal averages of the permeability, it should be noted that
three types of average are possible: arithmetic, geometric, and harmonic.
The arithmetic average is given by:

(3.4.1)

Hence, if, say, the zone were 50 ft, comprising 100 £ values at 0.5-ft
spacing, the average would simply be the sum of the 100 values divided
by 100. This average is appropriate to use if the flow in the reservoir is
in the direction of the bedding plane. Small, impermeable streaks will
have only very little effect on the average.
The geometric average is given by:

(3.4.2)

In effect, one takes the average of the logarithms of the individual £'s,
then takes the exponential at the end. This average is appropriate to use
if the flow in the reservoir is partially in the direction of the bedding plane
and partly normal to it. Impermeable streaks will have some influence but
not completely kill off the zonal average.
The harmonic average is given by:

(3.4.3)

In effect, one takes the average of the inverse of the individual &'s, then
inverts the result at the end. This average is appropriate to use if the flow
in the reservoir is normal to the direction of the bedding plane. Imper-
meable streaks will completely dominate the zonal average.
Depending on which method is used, the petrophysicist can get widely
different results. Typically the arithmetic average will be at least 10
times higher than the harmonic, with the geometric lying somewhere in
between.
Note that in horizontal wells there is an additional effect due to the fact
that kv I kh on the microscopic scale is usually less than 1. The effect of
this may be estimated as follows. Let a = kv/ kh, where kv = permeability
of the vertical well and kh = that of the horizontal. It may be shown that
the average permeability (kav) seen by the wellbore, which will be par-
tially influenced by kv and partly by kh, is given by:

The result, for various values of a, is shown in Figure 3.4.1.


The parameter of kv / kh will generally be assumed over an entire reser-
voir within a dynamic model. Typical values are between 0.1 and 0.3.
From Figure 3.4.1, it may be seen that the permeabilities, as determined
from a poroperm relationship, need to be adjusted in a horizontal well,
even if the formation appears homogeneous throughout.
When giving zonal averages, it is usual to also include the product
k*h, where h is the thickness of the zone, since it is this which can be
^hor^arith

KlK

Figure 3.4.1 Effect on Average Permeability of kv / kh < 1 in horizontal well

related to the flow generated in a production test. Where log-derived


values offc*h are compared with production tests, it is often the case that
the result derived from the arithmetic average will be higher than that seen
in the production test. Reasons for this are:

• Not all the perforated zone contributes to the flow, and the actual h is
smaller than that assumed in the petrophysical calculation.
• Some of the flow is not parallel to the bedding plane.
• Formation damage (called "skin") has occurred between the openhole
logging and the testing operation. While the test analysis seeks to iden-
tify this as a separate term from k*h, it may still be partly incorporated
into the calculated kh quoted.
• Relative permeability effects, such as gas blocking, may be occurring,
making the lab-calibrated in-situ brine permeability inappropriate.

Differences between the log-derived and test permeabilities are a fact


of life in real reservoirs and do not necessarily invalidate the log-derived
permeabilities that find their way into the static and dynamic models.
What happens in practice is that during the history-matching process in
the simulator, the permeabilities may be adjusted either globally or near
certain wells in order to make the predicted flow rates match the produc-
tion data.

Use the data in Appendix 2 to:

1. Revise your net sand discrimination criteria if necessary


2. Calibrate the density log against core porosity. Assume a net effective
stress of 2000 psi.
3. Derive a poroperm relationship
4. Derive revised values of m and n to use
5. Recalculate the sums and averages and additionally calculate average
permeabilities (using arithmetic, geometric, and harmonic averaging)
and k*h
6. Calculate the total equivalent hydraulic conductivity (EHC) (thickness
* porosity * Sh) and compare with your quicklook results.

Vous aimerez peut-être aussi