Transcript Document

Southwest Research Institute
J.O. Parra and C.L. Hackert.
Southwest Research Institute
P.C. Xu, Datatrends Research
H.A. Collier,Collier Consulting
M. Jervis, TomoSeis
Acoustic and CT Images to
Characterize Vuggy Carbonates In
South Florida Aquifers: From the
Pore to the Field Scales
Summary
Based on standard core measurements and image analyses, we
determined that the carbonate rock from the Ocala Limestone, South
Florida, is formed by moldic, vuggy, intergranular, intraparticle and
intercrystalline pores. Thin section analysis provided information on the
matrix porosity, and X-ray computed tomography (CT) provided
information on the vuggy porosity. We constructed vuggy porosity models to
calculate synthetic ultrasonic responses based on the finite difference
method, and we compared the synthetic with the ultrasonic data. The results
explained velocity changes associated with vugs in the carbonate rock at
the core scale. We cross plotted ultrasonic velocities with permeability
based on whole core measurements for different flow units of the carbonate
aquifer and found that velocity correlates well with permeability for each
carbonate flow unit. To determine whether we could assess the degree of
connectivity between vugs or between the matrix and the vugs based on
acoustic data, we calculated the squirt-flow lengths using a poroelastic
model. The results showed a good correlation between squirt-flow lengths
and the increase in permeability for each flow unit.
Summary, con’t
To predict relations between velocities and permeability at the borehole
scale, we determined the sonic velocities from full waveform monopole data
and permeability and porosity from NMR well log data. Saturated and
desaturated NMR core plug measurements aid in estimates of the irreducible
water in the rock and the variable T2 cut-offs for the NMR well log
calibration. The measurements established empirical equations to extract
permeability from NMR well logs. In addition, different empirical relations
were determined between the sonic P-wave velocities and the NMR-derived
permeability and porosity logs. Based on crosswell seismic measurements
conducted between two wells in the formation, we determined velocity and
reflection images at the field scale. Both images correlate well with the
lithology. Empirical equations developed at the borehole scale were applied
to predict permeability and porosity images from velocity tomography data
recorded between the two wells. The permeability images provided
information on three flow units; in particular, a high-permeability zone
characterized by interconnected vugs. This is consistent with the hydrological
results of high water production being monitored in the interwell region.
Figure 1. A 3D view of an X-ray CT scan image of the diameter of core 41, showing
an interconnected vug system typical in carbonate rock from the Ocala Limestone
Formation, Palm Beach, South Florida. This core is from a limestone with wackestone
texture taken from a depth of 1138 feet in the Ocala, an Upper Eocene formation.
The high permeability of the sample is controlled by interconnected vugs, and a
stochastic analysis of the CT image predicts two length scales: 2.2 and 17 mm in the
z-direction, and 2.4 and 7.2 mm in the xy-plane.
Figure 2. (a) Calculated density from the X-ray CT data for a slice from core 41.
Data plotted is dry density in g/cc. (b) A photograph of the end of core 41. This
core has extensive cavities that are captured by the CT images.
Figure 3. (a) 2D density images of cores 7 and 41 show small separated
and large interconnected vugs, respectively. Permeability of core 7, also
from the Ocala Formation, is 7 millidarcies; permeability of core 41 is >
4000 millidarcies.
Figure 3. In (b) and (c), the black line is the waveform for a homogeneous saturated
core, the green line is the waveform for a heterogeneous saturated core, and the red
line is the waveform for a heterogeneous dry core. (b) The synthetic waveform for core
7 is almost identical to the waveform of a uniform, equivalent core, although reduced in
amplitude. Core 7 has few cavities and is much more uniform than core 41. (c) The
waveforms associated with core 41 reflect the heterogeneity of the carbonate sample.
Figure 4. Cross plots of 24 measured and calculated P-wave velocities of whole
core samples versus permeability. The curves represent six flow units. The numbers
in the plot are squirt-flow lengths in millimeters. Each curve represents a flow zone
indicator (FZI) in terms of porosity and permeability. The curves show that P-wave
velocity decreases as permeability increases. Also, as permeability increases,
squirt-flow length increases as well, i.e., a large permeability correlates with large
flow lengths. The plots suggest that differences between observed and calculated
velocities are due to the presence of vuggy porosity in the cores. We expect that
the increase in permeability with squirt-flow length reflects the presence of
connected vugs.
Core #4
Core #21
Core #6
Core #32
Core #49
Figure 5 Comparison of NMR core signatures with OM thin sections for cores 4, 6, 21, 32
and 49. The NMR plots contain the T2 relaxation distributions for both desaturated and
fully saturated samples overlain on one plot. The NMR measurements were made for small
plugs having different permeabilities and porosities than the whole cores.
Core #4 is a limestone of grainstone texture with intergranular pores and vugs. The
saturated T2 distribution exhibits two pore size peaks at 40 ms and 1000 ms. The
photomicrograph show localized vugs that are separated. The T2 distribution reflects the
presence of the separate vugs (at 1000 ms) and the intergranualar pore size (40 ms). The
plug permeability is 158 md, and the whole core permeability is 1235 md. A
heterogeneous sample.
Core #6 is a limestone of grainstone texture with no vugs and with
poorly interconnected moldic pores. The T2 exhibits a peak at 12ms,
which corresponds to the matrix porosity of the grainstone sample.
The permeability of this plug is close to zero, and the permeability of
the whole core is 0.7 md. A homogenous sample.
Core #21 is a sandstone with an average grain size of 0.095 mm.
Intergranular pores (igp) with dolomite cement dominate the pore system.
The T2 distribution peak at 12 ms reflects the matrix porosity of the sample.
A low permeability, homogenous sample (average permeability of hte plug
is 31 md and of the whole core is 86 md).
Core #32 is a limestone of poorly-preserved packstone texture. Moldic and
intraparticle pores can be identified. The T2 distribution has a peak at about 400 ms,
which corresponds to large pore size (vuggy porosity). The sample is heterogeneous.
The plug permeability is 220 md, and the whole core permeability is 1805 md.
Core #49 is limestone of packstone texture. The pore system compromises common
intraparticle and intergranular pores, with moderate amounts of intercrystalline
pores. The T2 peak is at 80 ms, which corresponds to moderate pore sizes. The
permeability of hte plug is 13 md, and the permeability of the whole core is 98 md.
The sample is relatively homogeneous.
Figure 6 NMR measurements performed on 18 saturated and desaturated core
samples provided the information to produce T2 relaxation cut-off times (T2c) for use in
log calibration. Different relaxation rates are observed in the data associated with the
different rock groups in the well, which provides the appropriate bound-volume index
(BVI) for each core plug. Models based on flow zone indicator (FZI), NMR derived
porosity, and the BVI parameter provide the relationship to calculate permeability at
the core scale.
Figure 7. Four T2c, selected from cores and the permeability equation given in Figure
6, provide the calibration parameters to produce NMR-derived permeability in the
middle track above. Here the far right plot shows the T2 distribution in the color scale
and overlain red lines. The white log is the median T2 value, which is used with porosity
to discriminate among relaxivity units. The thick red line is the variable T2 cut-off. For
comparison, this figure shows the lithology, porosity and the Vp and Vs logs.
Figure 8. The Velocity Image, the sonic
log for well PBF-10, and gamma ray
logs for both wells are plotted in
measured depth from ground level (GL)
of well PBF-10. The data available for
quality control includes sonic and density
logs in well PBF-10 and gamma ray and
induction logs in both wells. Gamma ray
and induction logs show that the
structure is almost one dimensional, with
vertical but no lateral variations. A sonic
log from PBF-13 shows considerable
disagreement with that of PBF-10, so it
was not used as a quality control aid.
The velocity image shows good
agreement with trends in the sonic and
density log variations with depth.
Vertical velocity changes are fairly well
resolved and correlate with the sonic
and density logs as well as formation
changes. Since there are no receiver
locations below 1105 feet in well PBF10, the velocities below this interval are
constrained only by the travel time picks
from source locations below 1100 feet
in well PBF-13.
Figure 9. Location of crosswell seismic experiments at the western Hillsboro Aquifer
Storage Recovery (ASR) site near Boca Raton, Florida. One profile was acquired
between wells PBF10 and PBF13. At the surface, the well heads are approximately
333 feet apart.
Figure 10. The figure shows the wiggle traces
superimposed over the velocity tomogram. The
reflection image, velocity image, sonic log, and
tomographic velocity are plotted in measured
depth referenced from ground level for well PBF10 about 10 feet from the well (no velocity logs
are available for well PBF-13). Actual reflection
coverage below the total depth for each well is
limited by well spacing as well as the deepest
source and receiver locations in each well.
The resolution of the reflection data for
this profile is very good (to about 2 feet vertical
resolution). The reflection images show good
agreement with the acoustic and density log
variations with depth and hence with the synthetic
seismogram. Vertical velocity variations agree
very well with the reflection events, with little or
no lateral structural variations. Lateral reflection
variations probably result from imperfect
amplitude corrections applied to address large
amplitude variations in the data due to
differences in well casings between the wells. In
addition, there was no access to receiver locations
below 1110 feet due to an obstruction in the
well. This limited access further reduced reflection
coverage in the deeper part of the section and
thus the amount of reflection data available for
mapping.
Figure 11. Since there is a good correlation between P-wave velocity and permeability
at the core scale, as shown in Figure 4, we correlate the Vp with the NMR-derived
permeability and with the total porosity log. The computed relations are shown in Figure
11. The figure shows nonlinear relationships between the Vp and the permeability in the
permeable Ocala carbonates (1070-1200 ft) and in the Arcadia Formation at depth
intervals of 950-1010 ft and 1025-1045 ft. Figure 11 also shows linear relationships
between Vp and the total porosity in both regions.
Figure 12. The comparison of the permeability log with the lithology shows a very low
permeability in the sandstone, and a moderated permeability low in the chalky carbonate.
The high permeability values correlate with the carbonate with interparticle porosity, which
is consistent with the core data. The permeability and porosity equations are applied to
the velocity pixels to produce the permeability and porosity images. Here we compare the
lithology with the velocity, porosity and permeability images between 950-1200 ft. A
permeable zone between 1100-1200 ft in the interwell region of wells PBF10 and PBF13
is developed at well PBF10 and decreases toward well PBF13.
Conclusion
Ultrasonic responses based on CT models is a practical technique to
determine the heterogeneous conditions of the rock fabric and the applicability
of acoustics to describe the aquifer formation. The results of the analysis of the
observed and calculated ultrasonic responses leads to correlations of P-wave
velocity and permeability at the core scale. This suggests that the velocity can
be related to permeability at the borehole and interwell scales. The excellent
correlations between P-wave velocity and porosity and P-wave velocity and
permeability at the borehole scale demonstrate that fluid flow effects can
indeed be related to velocity tomography data. Thus, the velocity images at the
interwell scale were converted to permeability and porosity distributions
between wells PBF10 and PBF13. The permeability images delineate three
flow units in the 330-foot interwell region. The upper two flow units are thin
continuous connected beds, less than or equal to about 10 feet thick. The
deeper flow unit is approximately 60 feet thick and more heterogeneous than
the upper two flow units. The velocity, porosity and permeability images
correlate with the gamma ray and velocity logs.
Conclusion, con’t.
Comparisons of the plug and whole core permeability measurements imply the
presence of permeability heterogeneity at these two scales. This condition suggests
that NMR whole core measurements will provide more representative permeability
data than that from plugs. Thin section and core analyses show that the NMR well
log-derived permeability represents the matrix permeability. Although NMR well
log calibration based on plugs has apparently underestimated the true
permeability of the vuggy formation, the relative permeability correlates with the
lithology in the region between 1000 and 1200 feet.
This study demonstrated that acoustic and CT image techniques, supported by
core analyses and NMR measurements, can be used to characterize vuggy
carbonate formations at the pore and borehole scales. The final results
demonstrate that integration of well logs with tomography and reflection images is
a practical approach for mapping flow units in the carbonate aquifer of the Ocala
Limestone Formation in Palm Beach County, South Florida. In particular, the crosswell
seismic data provided high-resolution images, beyond the resolution of surface
seismic data, that are useful in characterizing the formations in the region of wells
PBF10 and PBF13.
Acknowledgements
This work was supported by Contract DE-AC26-99BC15203 from the U.S.
Department of Energy, National Petroleum Technology Office. Assistance of P.
Halder (DOE) is gratefully acknowledged.