Hydrocarbon Saturation and Viscosity Estimation from ... - OnePetro

Hydrocarbon Saturation and Viscosity Estimation from NMR Logging in the Belridge Diatomite' C. E. Morriss and R. Freedman Schlimberger Wireline and Testing C. Straley Schlumberger-Doll Reseurch M. Johnston Shell Western E&P, Inc. H. J. Vinegar and P. N. Tutunjian Shell Development Co.

Abstract: Nuclear magnetic resonance (NMR) logs have been recorded in Shell's North Belridge diatomite and Brown shale using an experimental logging tool. the CMR' Combinable Magnetic Resonance tool. The CMR tool successfully logged porosity and Tz distributions over 1.500 ft of formation. In the diatomite and Brown shale formations, the CMR porosity is a measure of the total liquid-filled porosity even in this tight lithology because of the high signal-to-noise ratio and short interecho spacing (320 ps). CMR porosity values were in good agreement with core porosity, even in zones with appreciable gas saturation, which indicated complete flushing of the gas within the depth of investigation of the tool. T2 distributions from both the borehole and laboratory measurements are distinctly bimodal, with the shorter T: peak at about 10 ms originating from water in contact with the diatom surface and a longer Tz peak at about 150 ms originating from the light oil. The Tz of the oil peak correlates roughly with oil viscosity. This indicates that the diatomite is predominantly water wet. The water and oil peaks are well separated in the light oil zones, which allows estimation of oil saturation by integrating the T2 distribution beyond a certain cutoff. The assignment of the oil peak and selection of the T? cutoff were determined by laboratory NMR measurements on native-state cores after diffusing in D20 to eliminate the water signal. Because of the tight lithology and good fluid-loss control, estimates of oil saturation from the shallow-reading CMR tool are in good agreement with resistivity logs and core analysis. Tz distributions have been measured on 3 1 crude oil samples from Belridge spanning a range ofviscosities from 2.7 to 4,300 cp. The CMR estimate of oil T2 correctly predicts oil viscosity

'Previously presented at SPWLA 35th Annual Logging Symposium, June 19-22. 1994. 'Mark of Schlumberger 44

and shows that the upper 150 ft of diatomite formation in this well undergoes a transition to heavier oil.

INTRODUCTION

An experimental logging tool, the CMR Combinable Magnetic Resonance tool, was evaluated in Shell's North Belridge field as part of a inultiwell field test campaign. The main objectives of the field tests were to evaluate the downhole sensor design and hardware options under consideration for the commercial version of the tool now under development. Details of the experimental logging tool, signal-processing algorithm, and the log outputs have been discussed elsewhere (Morriss et al., 1993; Freedman, 1994). The sonde is a pad-type device consisting of a directional antenna and only two permanent magnets to generate the static magnetic field. The commercial tool has a similar hardware design but will have a deeper depth of investigation and significantly higher signal-to-noise ratio than the experimental prototype. The borehole and laboratory NMR measurements described here were made to determine the magnetization Mo and transverse relaxation time T2 of hydrogen nuclei contained in bulk and pore fluids. Mo is proportional to the number of hydrogen nuclei in the sensitive region and can be scaled to give NMR porosity $NMR. For bulk fluids, T2 varies inversely with viscosity. For pore fluids, T2 values can be shorter than for the corresponding bulk fluids if the fluids interact with the rock surface. This reduction in T2 is due to the ability of the rock surface to promote NMR relaxation. In the fast diffusion limit, the measured T2 values are given by

The Log Analyst

March-April 1997

Hydrocarbon Saturation and Viscosity Estimation from NMR Logging in the Belridge Diatomite

Figure l a : SEM image of diatomite outcrop sample, showing diatom valve chamber walls.

1 -p2-+-, s --

T2

1

DIATOMITE GEOLOGY AND STANDARD LOG INTERPRETATION

C B

where S is the surface area of the pore, V is the volume of the pore, T2Bis the bulk fluid transverse relaxation time, and p2 is the surface relaxivity, which is a measure of the rock surface’s ability to enhance relaxation. The value of p2 depends on surface mineralogy and pore fluid type. Equation (1) is based on earlier work (Brownstein and Tarr, 1979) that established a link between S/V and the NMR longitudinal relaxation time (Tl) and more recent studies that have shown that Tl and T2 are well correlated at low magnetic fields (Kleinberg et al., 1993). Borehole measurements of T, are preferred over T I ,because T2measurements produce logs that have better precision at faster logging speeds and are free of the bed-boundary effects encountered with Tl logging (Kleinberg et al., 1993). Equation (1) forms the basis of NMR log interpretation: T2is proportional to V/S,which is proportional to pore size. NMR measurements on water-saturated rocks result in not one value of T2,but rather in a distribution of T2values that corresponds to the distribution of pore sizes in the sample. T2distributions have been used to estimate both permeability and producible porosity, two petrophysical parameters that are influenced by pore size (Straley et al., 1991; Kenyon et al., 1992). This study presents NMR measurements on low-permeability rocks that are partially oil saturated. The T2distributions are a summation of the individual contributions from the pore water and in-situ oil. March-April 1997

Figure 1b: SEM image of sample 1,764 showing matrix of fragmented diatoms.

Shell’s Belridge field in Bakersfield, California contains over 3 billion barrels of stock tank oil in place in the diatomite formation, of which only about 15% has been produced. The low recovery is indicative of the intricacies of the diatomite reservoir. The complex structural and diagenetic history in the diatomite has produced variations in lithology, oil saturation, and oil gravity. The diatomite formation consists of very thick (greater than 1,000 ft) high-porosity but low-permeability diatomaceous earth (porosity from 40 to 65 P.u., permeability less than 1 md). Diatomite is a general term for a rock composed mainly of the silica frustules of the diatom, a single celled plant that secretes a siliceous shell. The depositional environment consists of seasonal accumulations of diatoms periodically interrupted by episodes of sedimentation originating from turbidity flows or ash falls. Consequently, the mineralogical composition of the diatomite is mostly biogenic silica but also includes detrital quartz, feldspars, clays, heavy minerals, and carbonates. With increased temperature and pressure, diatomite in the form of opal-A (hydrated X-ray amorphous silica) undergoes a diagenetic change to opal-CT (X-ray crystalline silica) or porcelanite, known locally as the Brown shale. This diagenetic change occurs over a vertical transition zone that is typically tens of feet thick and results in a significant decrease in both porosity and permeability (porosity less than 45 p . ~ . permeability , less than 0.01 md). Scanning electron microscope (SEM) images of diatomite are shown in Figures l a and lb. Formation tops are shown in Figure 2.

The Log Analyst

45

Mnrriss et al.

I

1

Total Porosity

6C.0

(PW

0.0

1200

1300

1400

1500

1600

1700

1800

1900

TRANSITION ZONE 2000

2100

LE 2200

2300

Figure 2: Dean Stark porosity (circles) and oil volume (triangles) compared to openhole log values. Formation tops are shown in track 1. 46

The Log Analyst

March-April 1997


---_-----

CMR Porosity

I

- _ - _ _ _Calipers _ - _1_and_2 _ _ _ _

.-

1m.o

TOM Porosity

115oOFt

1200

1300

1400 ...

1500

1600

.:.:,.:.:.:

1700

1800

....

1900

........

2000

21 00 ..... ....

2200 ....

2300

Figure 3: Two intervals of the CMR log recorded across the diatomite, transition zone, and Brown shale. The CMR porosity, total porosity (from the density log), and Dean Stark porosity (circles) are compared in track 3. Gray scale in track 1 indicates the amplitude of the T2 distribution.

March-April 1997

The Log Analyst

47

Morriss et al.

The formation hydrocarbons consist of crude oil with a small connate gas saturation that is subcritical. The gravity of the crude oil varies from 10” to 35OAP1, with the more viscous oils typically occurring in the upper diatomite. It is estimated that almost one-third of the oil in the diatomite may have oil gravities less than 16” API. The oil gravity is important because it affects both primary and waterflood response. It also determines the production target for future enhanced oil recovery projects. Traditional log analysis at Belridge uses the density and deep resistivity logs supported by core data. The porosity is calculated directly from the density log assuming a constant gas saturation of 7%. Oil saturation is calculated using the Waxman-Smits equation to account for the clay conductivity of the diatomite. This log evaluation model has been verified with core porosity and oil saturation in many locations in the field. In this well, 600 ft of 4-in.-whole diameter core were obtained in four intervals spanning the diatomite, transition zone, and Brown shale. To minimize loss of fluids and wettability changes, the core was kept in sealed and refrigerated containers for transportation and storage. The log-derived porosities and oil volumes are shown in Figure 2, together with core values. Both the core porosities and oil volumes shown on Figure 2 have been corrected for compaction. Oil volumes were determined from the core samples using Dean Stark extraction with toluene. Because of the tightness of the diatomite, this process typically takes several months. Dean Stark oil volumes are considered representative of the virgin formation because the low permeability of the diatomites prevents significant flushing by drilling fluids in the interior of the 4-in. core. Oil viscosity in the diatomite is normally determined from oil samples that have been slowly squeezed from sidewall plugs using compaction presses. Because of the considerable expense and time required for Dean Stark and oil-expulsion analysis, NMR logging and NMR core analysis are evaluated as alternative methods for determining oil viscosity and saturation. CMR LOG DESCRIPTION

Two sections of the CMR log are shown in Figure 3, together with Dean Stark porosity and total porosity from the density log. The depth of investigation of the experimental tool is approximately 1 in. The connate gas appears to be completely flushed within the measurement volume, as the CMR porosity is comparable to the other porosity measurements. The CMR log shown in Figure 3 was recorded using a pulse sequence with a 1.3-s wait period to allow for polarization of the hydrogen nuclei, followed by the acquisition of 600 spin echoes with an echo spacing of 320 ps. The spin-echo sequences are collected in pairs, called “phase alternated pairs” (PAP). Hence, the total acquisition time 48

for 1 PAP is 3 s. The CMR log was run at 600 ft/hr, resulting in a new measurement every 6 in. T2 distributions are presented in track 1 using a color scheme where the gray scale increases with signal amplitude. Long intervals of the formation exhibit T2 distributions that are distinctly bimodal, with peaks at 10 and 150 ms. To provide a quantitative analysis of the borehole T2 distributions, a study of laboratory crude oil samples was initiated. The study and results are discussed below. LABORATORY NMR TESTING AND RESULTS

NMR measurements were made on preserved core samples and crude oil samples using a 2-MHz laboratory spectrometer. The measurements were made at a temperature of 25OC, close to the formation temperature of 40°C. The operating frequency and pulse sequence for the equipment were similar to those of the logging tool. A 5-s wait period was used to allow for complete polarization of the hydrogen nuclei. Immediately following the wait period, 4,095 spinecho amplitudes were measured using a CPMG sequence. The interecho spacing for the laboratory spectrometer was 160 ps, compared to 320 ps for that of the logging tool. The shorter echo spacing will result in increased sensitivity to any very fast decaying components. Studies have shown that the computed T2distributions are otherwise similar for both echo spacings, which indicates that the effect of molecular diffusion is negligible at these frequencies and relatively short echo spacings (Kleinberg et al., 1993). The laboratory NMR measurements differ from the borehole measurements in one important respect: the laboratory measured signal-to-noise ratios are several orders of magnitude higher than the borehole measurement. This results in detailed and robust T2 distributions that can only be approximated with the borehole measurement. To distinguish between water and oil signals in the rock and to evaluate the oil viscosity, NMR measurements were made on produced oils from the Belridge field and crude oils extruded from the core by pressing; cleaned water-saturated cores; diatomite partially saturated with Soltrol; and 28 native state cores before and after deuterium oxide (D,O) diffusion to eliminate the water signal. Table 1 summarizes the laboratory NMR measurements on the core samples and Dean Stark oil volumes and porosities. See the Nomenclature section for the definition of the quantities in Table 1. Bulk Oil Measurements

T2 distributions for the 3 1 Belridge bulk oil samples are displayed in Figure 4 in order of increasing viscosity. In contrast to refined oils that have narrow distributions (essentially single T2values), these crude oils have T2distributions that span several decades as a result of the mixture of

The Log Analyst

March-April 1997

Hydrocarbon Saturation and Viscosity Estimation from N M R Logging in the Belridge Diatomite

Table 1: Summary of Dean Stark and NMR measurements. Samples marked with an asterisk were cleaned and water saturated. All other samples were measured native state and after diffusion with deuterium oxide.

1188 1198 1238 1248 1258 1286 1296 1304* 1306 1324 1334 1344 1352* 1355 1366 1374 1690* 1690 1692 1694 1696 1746* 1760 1764* 1766 1770 1774* 1774 1780 1786 1976* 2008* 2092 2096 2105 2109 21 19* 2472*

50.0 54.5 43.4 60.4 58.3 61.4 60.7 59.9 55.7 ,56.9 53.4 52.9 56.8 60.6 61.5 62.8 58.8 60.4 59.6 57.8 54.0 49.5 49.1 51.5 49.1

47.3 38.5 65.9 57.2 59.2 54.7 52.8 50.1 47.3 53.0 43.6 59.9 58.2 59.0 58.3 59.0 55.0 54.4 56.1 53.4 53.4 57.7 58.2 56.4 58.2 61.8 54.8 54.1 58.6 57.9 58.7 53.8 50.5 49.7 53.3 51.1 49.4 37.3

2.9 8.9 3.8 3.3 1.9 3.1 4.1 0.6 4.4 6.6 8.4 1.6 1.1 2.4 9.8 9.1 4.7 6.8 7.7 7.2 7.3 0.9 8.8 2.2 2.9 11.4 0.4 7.9 13.8 12.8 0.8 0.0 6.1 3.9 4.4 4.3 0.0 0.3

7.2 10.3 6.1 6.0 6.9 6.3 9.2 7.2 8.2 9.0 9.7 9.9 5.0 10.6 9.2 9.9 12.2 16.0 16.5 15.7 16.5 6.6 10.0 9.2 12.9 12.2 7.5 10.2 12.1 11.5 6.8 6.4 5.5 5.1 4.7 4.3 3.4 2.1

hydrocarbon types within each sample. The distribution consists of a longer T2 peak originating from the most mobile hydrogen nuclei and a tail to shorter relaxation times from hydrogen nuclei with more restricted motion. As the hydrocarbon chain length increases, the viscosity increases and the relaxation times shorten. Chemical shift spectroscopy confirms the wider range of hydrocarbon types in the more viscous oils. Figure 5 shows a crossplot of logarithmic mean T2 (T2,10g), computed from the T2 distributions, versus viscosity (q) for the Belridge oils and for an additional 35 samples March-April 1997

9.8 13.5 9.8 11.1 18.6 11.9 12.3

0.5 0.5 0.7 0.9 1.3 0.9 3.0

5.9 5.1 7.5 6.9 7.1 6.6 12.3

9.5 9.6 23.5

12.0 10.2 9.2 18.7

3.4 4.7 4.7 10.5

13.6 23.1 26.1 30.6

20.7 17.4 14.7

18.9 17.7 14.7

10.3 8.9 7.6

29.0 25.8 26.8

22.0 20.6 20.8

21.2 22.7 21.1 21.2

14.8 16.2 14.9 15.8

45.0 48.4 48.1 53.1

15.4

15.3

7.9

26.9

18.3 16.5

18.1 17.5

9.6 10.6

27.2 35.0

9.9 14.6 14.5

13.2 15.9 15.5

6.4 9.3 9.5

26. I 32.2 37.3

10.9 12.1

13.7 12.5 12.9 11.7

5.7 3.8 4.4 4.0

18.9 13.4 15.1 14.9

from international fields and viscosity standards. The bestfit line through the data points, neglecting the four most viscous samples, gives 1200

T2,log

=

-p-

(2)

for T2,10g in milliseconds and q in centipoise. The deviation from this relationship for the four most viscous samples is thought to be due to experimental bias associated with measuring very short T2 values with the laboratory spectrometer. A correlation between crude oil

The Log Analyst

49

Morriss et al.

.l

tO

m.0

100.0

-0

poooo

.l

0

no

m.0

mo.0

I

mouo.0

Figure 4: T2 distributions for bulk oil samples from the Belridge field plotted in order of increasing viscosity, from top left to bottom right. Sample number, logarithmic mean T2 (Tz,lOg)and measured viscosity (in centipoise) are shown for each sample.

50

The Log Analyst

March-April 1997


1J,

viscosities and T I , analogous to the plot in Figure 5, was published previously by Brown (1 96 1). T2 distributions for nine bulk oil samples pressed from the core, together with measured API gravity, are shown in Figure 6 . These samples show a trend that has been established for the Belridge field: the oil undergoes a transition to heavier oil at the top of the diatomite. In the diatomite below about 1,350 ft, the oil is consistently a light crude with roughly a 27"API gravity. Above 1,350 ft the oil becomes more viscous and approaches 16"API gravity at 1.257 ft.

1

Cleaned and Water-Saturated Plugs

T2 distributions were determined for cleaned water-saturated plugs from the diatomite, the transition zone, and the Brown shale. The results are plotted in Figure 7, together with differential mercury capillary pressure curves on nearby samples. The low T2 and high capillary pressures reflect the exceptionally small pore sizes in these rocks. In the diatomite, the peaks in the differential capillary pressure curve at about 500 to 1,000 psi correspond to an

Oepthc 1330 ft

;i

!

at

I

."""I

I

lo

I.1111.1

I

I I 1 1 1 1 1q

I 1111111,

Q

o

l

D

Q

o

I

~

, ~

,1111 ' I

l

O

,': O

.A O

O

Q

Viscosity (cp) Figure 5: Crossplot of logarithmic mean TZ (T2,lOg) versus viscosity for bulk oil samples from the Belridge field (triangles), international oilfield samples, and oil viscosity standards (plus symbols). Neglecting the four most viscous samples, the best fit to the data is given by a power law (i.e., T2,iOg= 1,200/q0.9).

n2lfnMc Qow22AR wlc:5ocP

r--------- r---------

1

Depth: 1338 ft

n55-

m4sllP.c

Depth: 1350 ft

nom#C

Depth: 2010 ft

Qw:25Au

b t c

vbc5ocP

v k

1

I

I

I

1

I

i

h

.l

LO

rO.0

100.0

looo.0

WKKl.0

.1

uw

I

LO

I

m.0

I

m.0

I

w . 0

100(

TAms)

Figure 6: T:! distributions for bulk oil samples pressed from sidewall core plugs. Sample depth and logarithmic mean T2 (T2,lOg)are shown for each sample together with API gravity, if measured. Viscosity is estimated from the API gravity using published tables (Beal, 1946). March-April 1997

The Log Analyst

51

Q

Morriss et al.

."*

1352 ft

2119 ft

I 81

roobo

lodo0

&

MTO

rb

loobo

Mercury Ressure (psi)

lodo0

&

Go

lo

Mercury Ressure (psi)

I

I

I

I

I

I

0.1

1.0

10.0

100.0

rooo.0

0.1

T,(ms)

I

I

1.0

10.0

I

m.0

I

K)o0.0

T,(ms)

Figure 7 : Differential capillary pressure curves (dashed) compared to T2 distributions (solid) on cleaned and water-saturated samples. Sample depths for the capillary pressure measurements are shown on the left in parentheses. Sample depths for the NMR measurements are shown on right, together with NMR porosity (I$NMR), free fluid porosity (4v(33)), and logarithmic mean (T2,log).

52

The Log Analyst

March-April 1997


average 0.2- to 0.4-micron-diameter opening into the wellpreserved diatom bodies. The tail of the curve to higher capillary pressures is due to the distribution of throat sizes in the matrix of fragmented diatom frustules and clays. In the transition zone (1,920 to 2,121 ft), the second peak, which develops at 30,000 to 40,000 psi, is associated with the formation of opal CT. In the deep Brown shale sample (2,493 ft), the diatoms have been largely crushed and recrystallized, suppressing the 500- to 1,000-psi diatomite peaks. Increased clay content in the Brown shale results in a middle peak at about 10,000 psi. Values of T2,10g range from 2 to 12 ms, several orders of magnitude shorter than the T2E of bulk water (approximately 3,500 ms). In this case, the 1/T2B term in Equation (1) is negligible, and the measured T2 is dominated by surface relaxation. In general, NMR measurements are relatively insensitive to very small pores that have short T2 values. However, NMR porosities for the cleaned samples and native state samples are comparable to Dean Stark values, including the Brown shale sample (see the crossplot in Figure 8). This indicates that even the smallest pores are included in the laboratory NMR porosity. Some differences in the shapes of the T2distributions and capillary pressure curves are to be expected, as mercury injection measures the size of pore throats, whereas NMR responds to the size of pore bodies. In addition, NMR T2 measurements are intrinsically low resolution. Therefore, the T2 distributions are comparatively smooth and lack the fine detail exhibited in the capillary pressure curves. The overlay of capillary pressure curves and T2 distributions in Figure 7, which works well for the diatomite samples. is based on a correspondence of

Pnmr Figure 8: Crossplot of NMR porosity ((INMR) versus Dean Stark porosity ($ds) for both cleaned samples and native state samples. Porosity values are also shown in Table 1.

T2 distributions at very high pressures for the transition zone and Brown shale samples suggest there is a different pore body to pore throat ratio for this ultra-fine porosity or that opal-CT has a lower surface relaxivity than opal-A. Partial Saturation with Soltrol

Sample 1776’ was cleaned, water saturated and then injected with Soltrol 130 using a Hassler sleeve apparatus. Soltrol 130 is a refined aliphatic oil with a narrow molecular weight cut that results in a narrow T2 distribution. T2 P = lo4 ms -psi (3) Figure 9 shows T2 distributions for bulk Soltrol, sample This relation allows us to estimate the value of p2 for the 1766’ completely saturated with water, and sample 1766’ diatomite. For a cylindrical pore throat of radius r, the saturated with both water and Soltrol. At partial saturation mercury entry pressure is the distribution is bimodal with a short T2water peak and a long T2 oil peak. This assignment is based on the short T2 p=-, 2Y (4) peak being similar to that for the water-saturated sample, r while the long T2peak is close to the T2of bulk Soltrol. This psi indicates that the Soltrol is not wetting the rock surface. The where y is the surface tension of mercury (7.15 x cm). In a cylindrical pore (i.e., tube-like pore structure), S/V T2 distribution for the partially saturated sample is a linear = 2/r. Therefore, neglecting the bulk relaxation summation of the individual contributions from the water and in-situ oil. Therefore, the oil saturation may be esti1 2 (5) mated from the T2 distribution by dividing the area under -= P 2 7 T2 the oil peak by the area under the entire distribution. This and therefore results in an oil saturation of 28%, which compares favorably to a value of 29% from weight measurements. Similar Y P2 results have been obtained for chalk samples partially saturated with decane and sandstones partially saturated with Then, p2= 0.7 x 10” cm/s. kerosene (Straley et al., 1991 and Howard and Spinler, Discrepancies between the capillary pressure curves and 1993).

=T,P.

March-April 1997

The Log Analyst

53

Morriss et al.

I

I

I

I

I

1

.t

1.0

10.0

100.0

1000.0

10000.0

Figure 9: 72 distributions for sample 1776’: 100% water saturated (long dash) and 28% Soltro1/72% water saturated (solid). 72 distribution for bulk Soltrol 130 also shown (dashed).

Native State and Diffused Measurements

NMR measurements were made on 28 samples in the “as received” condition. The samples were then pressurized with H,0 to ensure that the pores were completely filled with liquid. A second NMR measurement was then made. The difference in NMR porosity between these two nieasurements ranged from 0 to 4P.u., presumably a result ofthe connate gas. For ease of discussion, measurements after pressurization are referred to as native state. The native state samples were then soaked in deuterium oxide (D20).During this process. the D?O diffuses into the sainple and reduces the concentration of H 2 0 .The soaking time and initial volume of D,0 were determined to be suft’icient to reduce the concentration of H 2 0to a negligible level. The high porosity of the diatomite allows the D,O to diffuse rapidly, in the order of one day. After soaking, the NMR signal originates entirely from the in-situ oil, as D,0 does not contribute a signal at the operating frequency ( 2 MHz) of the laboratory spectrometer. Therefore, the measured T, distribution is a property of the in-sihl oil and may be used to estimate oil viscosity and evaluate wettability. In addition, the NMR porosity after diffusion is a direct measure of the oil volume in the sample. T, distributions for the samples in their native state condition and after DzO diffusion are shown in Figures 10 and 11. By comparing these two distributions it is possible to distinguish between the water and in-situ oil signals. All the native state distributions have a water peak at approximately 10 nis, similar to the distributions obtained on the 100% water-saturated samples. From this we infer that the water is in contact with the rock surface and that these rocks are predominantly water wet. This conclusion is supported by centrifuge relative permeability nieasurements that indicate oil displacement is significantly influenced by water imbibition. The logarithmic mean Tz values for some of the in-situ oils (see T2.d2,,values in Figures 10 and 11) are up to 30% $LJ2LI

51

shorter than adjacent pressed oils, suggesting that the oil may be in partial contact with the rock surface. However, these relatively small differences are more likely due to evaporation of light oil components during the 6-month storage period prior to the NMR measurements. Hence, the in-situ oils may have higher viscosities than adjacent bulk samples. Evaporative losses are also indicated by the Dean Stark oil volumes for these samples, which are about 10% lower than for the twin plugs that were measured shortly after the core was cut. Even if the oils interacted with the rock surface, the reduction in oil T, would be small compared to the effect observed with water. This is because Tzs values for these crude oils are already short, and the value of 1/T2Bin Equation (1) is comparable to the surface effect term (i.e., pl- S/V). The small dependence on oil wettability is fortuitous, as it allows reasonable estimation of oil viscosity from the in-situ oil signal. By comparing the native state and D,O diffused distributions (shown in Figures 10 and 1 1 j, the samples can be divided into the following categories: 0 Diatomite samples above 1,324 ft contain heavier oil. The native state distributions are unimodal as the major water and oil signals are coincident. Note that a small water signal also occurs at longer T2 values. 0 Diatomite samples below 1,324 ft contain relatively light crude. The native state distributions are bimodal, and the second peak has significant amplitude centered around 150 ins because of the oil signal. As expected, the amplitude of this peak increases with increasing oil volume. 0 The transition zone samples (2,092 to 2,109 ft) contain intermediate weight crude. The native state distributions are bimodal or are elongated towards high T,. The signal amplitude in the 20- to 100-ms interval originates from the in-situ oil. NMR estimates of oil volume from measurements on the D,O diffused samples are in excellent agreement with Dean Stark oil volumes, as shown in Figure 12. For those samples that contain light oil (i.e., samples 1,334 to 1,786 ft), the volume of in-situ oil can be estimated from the native state distributions using the same method described for the sample injected with Soltrol. However, in this case there is not a distinct T, cutoff separating the oil and water signals. The short T, peak arises froin both water and high-viscosity crude oil; the long T2 peak is predoniinantly from the in-situ oil but may also include 1 or 2 p.u. of water. The trough between the two peaks occurs between 30 and 40 ms, Coincidentally close to the 33-111s cutoff used to estimate producible porosity in water-saturated sandstones (Morriss et al., 1993j. This NMR estimate ofproducible porosity has historically been called the free-fluid porosity ($,133,). Although a 33-ms cutoff was used for the

The Log Analyst

March-April 1997


1198 ft (p-=

38.5

(pa=

135

T2-=

6.9

A

1238 ft

1248 ft = 57.2 (p,(33) = 3.3

A

1374 ft

I

.1

I

I

I

I

1.0

lO.0

100.0

1ooo.o

I

I

I

I

1

.1

1.0

10.0

100.0

1000.0

T,(ms) Figure 10: T2 distributions for samples in the native state (solid) and after soaking in D20 (dashed). Sample depth, NMR porosity ( ~ N M R )and , free-fluid porosity (4f(33)) for the native state measurements are on the left side. NMR porosity (4d20) and logarithmic mean T2 (T2,d20) after soaking in D20 are indicated on the right side. Samples are ordered by increasing depth. March-April 1997

The Log Analyst

55

Morriss et al.

1690 ft 54.4 ( p p ) = 16.8 pn,=

(p-'

21.2

(pm=

22.7

r---------

1692 ft (pf(33)

= 17.7

1694 ft

1760 ft cpn,,=

58.2

2109 ft 51.1

= , p(

(pa==

T2&=

I

I

I

I

1

.1

1.0

10.0

100.0

looo.0

.1

1.0

10.0

100.0

11.7 14.9

1000.0

TAms) Figure 11: T2 distributions for samples in the native state (solid) and after soaking in Dz0 (dashed). Sample depth, NMR porosity ( ~ N M R ) ,and free-fluid porosity (91(333) for the native state measurements are on the left side. NMR porosity (4d20) and logarithmic mean T2 (T2,d20) after soaking in Dz0 are indicated on the right side. Samples are ordered by increasing depth. 56

The Log Analyst

March-April1997


00

/ 0'

0.

0'

0'

Id-

0' 0'

0

0'

N

0'

0

0'

.d

,./'

0

0'

0 0

0

0

9'

u)

P

0 0

/'

0 4 d-

0' 0. 0'

04

6.

0' 0'

0'

10.

16.

20.

26.

0.

3

6.

10.

16.

20.

26.

3

I.

%*o Figure 12: Crossplot of NMR porosity ( 4 ~ 2 after ~ ) diffusion with D 2 0 versus oil volume from Dean Stark (Vds), for samples from 1,334to 1,786ft.

Figure 13: Crossplot of oil volume estimated from the native state T2 distributions versus oil volume from Dean Stark (vds) for samples from 1,334to 1,786ft.

analysis, similar results can be obtained for any value of T2 that lies between the two peaks on the distribution. To estimate oil volume from the native state distributions it is necessary to determine 1) the fraction of the in-situ oil signal with T2 greater than 33 ms (that is, 4r120(33)/4dz0, or 0.59 f 0.08.); and 2) the volume of water with T2 greater than 33 ms (that is, 4 ~ 3 3 -) 4 d 2 4 3 3 ) ; this volume is 2 f 1.1 P.U.). Using the above facts,

shortening of the water T2peak from about I0 to 6 ms. This reduction in T2 reflects the smaller pore sizes associated with the Brown shale. If, as previous studies have shown for sandstone, permeability is proportional to 44(T2,10g)2 matrix permeability for the Brown shale should be at least an order of magnitude lower than that of the diatomite. This is consistent with core analysis. The majority of the diatomite interval from 1,350 to 1,800 ft has a second peak at about 150 ms that is interpreted as the NMR signal from the 27"API gravity oil. The NMR log shows the oil viscosity is constant over this 530-ft interval. In the transition zone, the second peak shortens to about 50 ms reflecting an intermediate oil viscosity. Above 1,350 ft, the T2 of the in-situ oil shortens and eventually merges with the water peak at 1,250 ft. This is due to the gravity of the oil decreasing to 16"API. Oil volumes calculated from the borehole T2 distributions using Equation (8) are displayed in Figure 14 for the interval from 1,650 to 1,850 ft. The oil volumes are comparable to but slightly lower than the core and conventional wireline log estimates. This may be due to a small amount of oil flushing around the wellbore.

(7) and the volume of in-situ oil (V,,,) the native state distributions by

can be estimated from

The oil volumes determined using this technique are compared to Dean Start volumes in Figure 13. The scatter is believed to be primarily due to changes in oil viscosity. CMR LOG AND LABORATORY NMR COMPARISON

The borehole T2 distributions shown in Figure 3 are in good agreement with the native state distributions obtained on the core samples. The bimodal T2 distribution observed over long sections of the log is due to a short T2 component from the water and a longer T2 component from the in-situ oil. The transition to the Brown shale is clearly marked by a decrease in CMR porosity from about 55 to 40 p.u. and a March-April 1997

CONCLUSIONS

CMR log results gave good estimates of oil viscosity, porosity, and oil saturation. The CMR log will be extremely useful for delineating the light and heavy oil reserves in the Belridge field. Similar results can be expected in other low-permeability formations containing light crude, for which the water and oil T2 signals are well separated.

The Log Analyst

51

Morriss et al.

. ..

...... ... ...... .:.:.:.:. ..... ........ .........

Figure 14: Section of CMR log from 1,650to 1,850ft. Oil volume estimates from the CMR Tz distributions (solid) are compared to openhole values (dash) and Dean Stark values (triangle).

Laboratory NMR measurements on samples soaked in D 2 0 successfully isolated the in-situ oil signal and therefore allowed estimation of oil viscosities and volumes. This methodology should work in other oil-saturated cores. An analogous technique for logging would be to add Mn EDTA to the mud system (Brown and Neuman, 1980), which would selectively kill the formation water signal. The resulting porosity log would then be a direct measure of the residual oil volume in the invaded zone. 58

The similarity between laboratory and borehole measurements is an attractive feature of NMR. Petrophysical relationships can be established in the laboratory and applied to the interpretation of NMR logs. NOMENCLATURE T2

T?B

The Log Analyst

NMR transverse relaxation time NMR transverse relaxation time for bulk fluid March-April 1997


Logarithmic mean T2 for native state and cleaned samples Logarithmic mean T, after diffusion with

deuterium oxide surface relaxivity pore surface area pore surface volume N M R porosity N M R free-fluid porosity using a 33-nis cutoff N M R porosity after diffusion with deuterium oxide; equal to the volume o f oil in the sample N M R free-fluid porosity after diffusion with deuterium oxide Dean Stark porosity Dean Stark oil volume Volume of oiI estimated from native state T2 distribution Fluid viscosity

ACKNOWLEDGMENTS We thank D. Rossini for performing the laboratory NMR measurements and J. Ferris, N. Patterson, and L. Bielamowicz for assistance with sample preparation and petrophysical measurements. We gratefully acknowledge usefuI discussions with R. Kleinberg.

REFERENCES Beal, C., 1946, The viscosity of air, water, natural gas, crude oil and its associated gases at oil field temperature and pressures: Transactions AIME, v. 165, no. 94.

March-April 1997

Brown, R. J. S., 1961, Proton relaxation in crude oils: Nature, v. 189, no. 4762, p. 387-388. Brown, R. J. S. and Neuman, C. H., 1980, Processing and display of nuclear magnetism logging signals: application to residual oil determination, paper K, in 21st Annual Logging Symposium Transactions: Society of Professional Well Log Analysts. Brownstein, K. R. and Tarr, C. E., 1979, Importance of classical diffusion in NMR studies of water in biological cells: Phys. Rev. A , v. 19, no. 6, p. 2,4462,453. Freedman, R., 1994, Processing method and apparatus for processing spin echo in-phase and quadrature amplitudes from a pulsed nuclear magnetism logging tool and producing new output data to be recorded on an output record: US Patent No. 5,291,137, issued March 1, 1994. Howard, J. J. and Spinler, E. A., 1993, NMR measurements of wettability and fluid saturations in chalk, SPE-26471: Society of Petroleum Engineers, Richardson, Texas. Kenyon, W. E., 1992, Nuclear magnetic resonance as a petrophysical measurement: Nuclear Geophvsics, v. 6, p. 153-17 I . Kleinberg, R. L., Straley, C., Kenyon, W. E., Akkurt, R.. Farooqui, S. A., 1993, Nuclear magnetic resonance of rocks: TI vs T2, SPE-26470: Society of Petroleum Engineers, Richardson, Texas. Morris, C.E., Macinnis. J., Freedman, R., Smaardyk. J., Straley, C., Kenyon, W. E., Vinegar, H. J., Tutunjian, P. N., 1993, Field test of an experimental pulsed nuclear magnetism tool, paper GGG, in 34th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts. Straley, C., Momss, C. E., Kenyon, W. E.. Howard, J. J., 1991, NMR in partially saturated sandstones: laboratory insights into free fluid index, and comparison with borehole logs, paper C, in 32nd Annual Logging Symposium Transactions: Society of Professional Well Log Analysts.

The Log Analyst

59

Effective Porosity, Producible Fluid, and Permeability in Carbonates from NMR Logging' Dahai Chang and Harold Vinegar Shell Development C0tnpan.y Chris Morriss Schlumberger Wireline and Testing Chris Straley Schlutnberger-Doll Resenrcli

Abstract: Producibility estimates in carbonate formations have always been a challenge for log interpretation. Broad pore size distributions in carbonates, from microcrystalline to large vugs, have a large effect on productivity, permeability, and estimation of hydrocarbon saturation from resistivity logs. In mixed complex carbonates it is even difficult to obtain accurate porosities from conventional wireline logs without calibration against core. An experimental logging tool, the CMR2 Combinable Magnetic Resonance tool, has been evaluated in the Glorieta and Clearfork carbonates in West Texas. The logged interval is complicated by significant amounts of separated vuggy porosity. The lithologies consist of dolomite. limestone, anhydrite, and clastic material (silt containing quartz, feldspar, and clay). The CMR porosity is derived independently of formation lithology. In the clean mixed carbonates, the CMR porosity is equal to total porosity. In the silt zones, the CMR porosity measures the effective porosity because it is insensitive to microporosity associated with the clastic material. Using the CMR free-fluid porosity, the producing oil-water contact, which was difficult to determine using the conventional wireline logs alone, was identified in this well. Twenty-seven core plugs from the well were analyzed by low-field nuclear magnetic resonance (NMR) and conventional core analysis. Estimates of producible fluid obtained from the T2 distributions using a free-fluid cutoff of 92 ms agreed with centrifugeable fluid from the plugs. Petrophysical properties depending on the relative amounts of intergranular and vuggy porosity were found to be well correlated with the long end of the NMR T, distributions. Permeability estimation from the porosity with T, < 750 ms using the relation k -@4 T i was superior to that based on total porosity. Finally, the cementation exponent m was found to increase with the fraction of long T, porosity.

INTRODUCTION

Numerous Glorieta and Clearfork reservoirs on the Central basin platform have been produced for many years. Many of these fields are in a mature stage of production through primary recovery and waterflood. Some of the fields are being CO? flooded and others are being studied as potential C 0 2 flood targets. Because of the complexity of the reservoirs, understanding the remaining hydrocarbon distribution and reservoir flow characteristics is essential for achieving advanced reservoir management. The main lithology of the Glorieta and Clearfork formations in this field is dolomite with textures varying from mudstone dolomite to grainstone dolomite. Various amounts of anhydrite are distributed through the formation in a range of size scales and modes from pore filling and matrix replacement to nodular and fracture filling. Quartz, plagioclase, K-feldspar, muscovite, and illite, collectively called silt because of their fine particle sizes, appear frequently. These silty zones are usually continuous laterally, though variable vertically, thus providing good markers for geological correlation. In addition, limestone and calcite cement are occasionally observed in the lower section of the Clearfork formation. The Glorieta and Clearfork (Permian) formations are low-porosity and low-permeability reservoirs. The porosity typically ranges from nonporous to 25 p.u. The air pernieability varies from approximately 0.01 md to more than 1 darcy, illustrative of the heterogeneity of the formation. The pore geometry varies through the entire formation and many zones have a large amount of vuggy porosity. The productive interval is carbonate (dolomite or limestone). The presence of a small matrix of silt is closely associated 'Previously presented at SPWLA 35th Annual Logging Symposium. with low-permeability zones, which act as good vertical flow barriers. June 19-22, 1994. Formation evaluation in the Glorieta and Clearfork res"Mark of Schluniberger 60

The Log Analyst

March-April 1997

Effective Porosity, Producible Fluid, and Permeability in Carbonates from NMR Logging

ervoirs has been difficult. The main challenges are lithology identification, such as silt-zone delineation, and porosity evaluation for determining net pay. The fluid distribution and flow characteristics such as permeability and producible fluid volumes are also important as many of these reservoirs are in secondary or tertiary recovery. To assist in formation evaluation, the CMR logging tool was evaluated in a Shell Western Exploration and Production Inc. (SWEPI) well in West Texas across the Glorieta and Clearfork formations. In addition to the CMR log, a complete logging suite was run including DLL' Dual Laterolog Resistivity log, MicroSFL' log, CNL' Compensated Neutron Log, Litho-Density' log, NGS' Natural Gamma Ray Spectrometry log, and EPT' Electromagnetic Propagation log. In addition, 100 ft of conventional whole core was taken in the lower section of the Clearfork reservoir. The CMR tool presents a promising evaluation technique for this complex carbonate reservoir because it provides a lithology-independent porosity measurement, as well as an estimate of producible fluid (free-fluid porosity). Combined with other logs, it was hoped the CMR log would be able to determine the oil-water contact (OWC), which is difficult with conventional wireline logs. Laboratory NMR measurements on the core were directed towards obtaining petrophysical properties in a section of particularly vuggy dolomite. NMR measurements before and after centrifugation of the core were used for determining the producible fluid. Point counting thin sections showed up to 50% vuggy porosity in this zone. Because the vugs in this section are separated, they contribute only weakly to fluid or electrical flow. Thus, it was expected that eliminating the vuggy porosity in the petrophysical correlations for permeability and cementation factor would improve the fit.

the pore, T2Bis the bulk fluid transverse relaxation time, and p2 is the surface relaxivity, which is a measure of the rock surface's ability to enhance relaxation (Kenyon, 1992). The value of p2 depends on surface mineralogy and pore fluid type. Previous investigations have shown p2 is weaker in carbonates than in sandstones (Timur, 1972). Laboratory data suggest p2 0.0005 c d s for carbonates compared to -0.001 5 cmls for sandstones. Equation (1) shows that T2 is proportional to V/S which, in turn, is proportional to pore size. For example, for a spherical pore, SIV = 3/r, where r is the radius of the pore. In a carbonate, if the bulk fluid is mud filtrate with a T2Bof 2 s, and if p2 = 0.0005 cmh, then a pore diameter of 60 microns would have a T2 of 1 s. NMR measurements on water-saturated cores result in a distribution of T2 values that corresponds to the distribution of pore sizes. T2 distributions have been used to estimate both permeability and producible porosity, two petrophysical parameters that are influenced by pore size (Straley et al., 1991; Kenyon, 1992). In sandstones the T2 cutoff for producible fluid has been shown to be about 33 ms; however, the weaker surface relaxivity in carbonates results in a longer value for T2.One of the purposes of the laboratory NMR study reported here is to establish the T2cutoff for the free-fluid index in a dolomite lithology.

-

CMR TOOL AND OPERATION

The CMR tool is an experimental pad-type pulsed magnetic resonance logging tool employing two permanent magnets to generate the static magnetic field. The details of the experimental logging tool have been outlined in an earlier publication (Morriss et al., 1993). The depth of investigation is about 1 in. The commercial tool has a similar hardware design, but it will have deeper depth of investigation and significantly higher signal-to-noise ratio NMR IN POROUS MEDIA than the experimental prototype. Pulsed NMR measures the magnetization (M,) and transThe CMR log was run using a pulse sequence with a verse relaxation time (T2)of hydrogen nuclei contained in 1.3-s wait time followed by the acquisition of 600 spin the pore fluids. Mo is proportional to the number of hydroechoes with an echo spacing of 320 pus. The spin-echo gen nuclei in the sensitive region and can be scaled to give sequences are collected in pairs, called phase-alternated an NMR porosity (@NMR). For fluids confined in pores, T2 pairs (PAPS). Hence, the total acquisition time for one PAP values can be shorter than for the bulk fluid if the fluids was 3 s. A maximum logging speed of 600 ft/hr is required interact with the rock surface, which promotes NMR reto ensure that a new PAP is acquired during each 6-in. laxation. In the fast diffusion limit, the measured T2values sample interval. are given by A bland saltwater drilling mud was used in this well. Laboratory NMR measurements of the mud filtrate show a T2 of 2.06 s. The bottomhole temperature is 110°F. The produced oil is an intermediate gravity crude with a 30"API where S is the surface area of the pore, V is the volume of gravity. The logarithmic mean T2for the crude oil at 110°F is about 120 ms. Residual oil saturations in the core range from 10% to 30%. As a result of its shallow depth of investigation, the CMR tool is assumed to be measuring a zone flushed with mud filtrate. 'Mark of Schlumberger March-April 1997

The Log Analyst

61

Chang et al.

“DEPTH F

5900

5950

6000

6050

6100

6150

6200

Figure 1 : Comparison of CMR and conventional porosity logs from Glorieta formation (5,900-6,200ft) showing lithologic variations.The N M R porosity agrees with the conventional porosity interpretation except in the silty zones. In t h e silt zones all the N M R porosity appears to be bound fluid.

LOG INTERPRETATION

Figures 1 and 2 show logs and interpretations near the top of the Glorieta formation and at the lower section of the Clearfork reservoir. The interpretation is based on a corecalibrated evaluation technique developed by SWEPI for the Glorieta and Clearfork formation (Clerke, 1993). This evaluation scheme uses NGS/Litho-Density/DLL logs and solves for lithology, porosity. and water saturation. As illustrated in track 2 of both Figures 1 and 2, significant lithology variation occurs across the Glorieta and Clearfork formations. In Figure 1, large amounts of silt are present in the top section from 5.900 to 5,960 ft. Moderate amounts of anhydrite, ranging from 20% to 40%, are observed from 6,140 to 6,200 ft. In Figure 2, dolomite is the dominant lithology with small amounts ofanhydrite. Below 7,260 ft, however, both liniestone and dolomite are present. 62

This lithology interpretation is confirmed by core in this and nearby wells. One of the main difficulties in formation evaluation in this cotnplex lithology is deriving lithology-independent porosity for accurate hydrocarbon saturation and volunietric estimation. Track 3 of Figure 2 shows three different porosity curves. $+,MR is the CMR porosity. @CON is the porosity based on SWEPI’s multilog evaluation technique, and is the porosity measured from 2-in.-OD x 2-in.long core plugs. The plugs were regularly sampled at 1-ft spacing and the measured porosity was smoothed over 1.5 ft to approximate logging tool resolution. A similar presentation is given in track 3 of Figure 1. Considering the heterogeneity of the reservoir, there is good agreement among porosity logs even though the amounts of dolomite, anhydrite, and limestone vary considerably. Thus, the CMR log provides a good lithology-independent porosity for this complex carbonate reservoir.

The Log Analyst

March-April 1997

Effective Porosity, Producible Fluid, and Permeability in Carbonates from N M R Logging

I M ........ E .................. ........................ PH I NMR ..._. ....................L.......... .....................................

0 .2

1

CFlL I -__-----------15 5 1 "

O E P lH

F

CR

"

0

RNHY

_ - - _ _ - - - _ _ - -SILT ---. 200 0

0

M O U O 1L ____-_---------.

_ L _ _ _ _P _ _H- -1- _CORE ---.

0 .2

0

0 SW

P H I CON

0 0

1 .2

2 MOUWRT

1 .2

0

7150

7200

7250

7300

Figure 2: Comparison of CMR and conventional logs from Clearfork formation (7,150-7,300ft) showing movable oil and water. Based on the CMR log, the oil-water contact is believed to be at 7,240 ft.

The only disagreement occurs in the silty zones indicated in track 2 of Figure 1, where $NMR is lower than $CON. Through core calibration and production data, it is known that the silty zones are nonproductive. Conventional porosity logs such as density, neutron, and sonic overestimate the formation porosity in these zones because of associated microporosity and different matrix properties. The CMR log is insensitive to the microporosity in the silty zones not only because of the very small pore size but also because of the enhanced T2 surface relaxivity of the silt matrix. Since the CMR log is insensitive to very short relaxation times (T2 < -1 ms), it fails to detect this microporosity. Hence, the CMR porosity can be interpreted as an effective porosity. Although no core was obtained in the silty intervals in this well, these low values for effective porosity are consistent with core in the silt zone in nearby wells. CMR free-fluid porosity measurements provide reasonable estimates for the permeable zones. Based on NMR laboratory data on the core discussed below, free-fluid index is obtained using a T2 cutoff of 92 ms. As illustrated in

($r>

March-April1997

track 4 of Figure 1, permeable zones are indicated by a large

$f In this example, they correspond to nonsilty zones with porosity above 3 to 4 P.u., which is consistent with the porosity-permeability relationship in this field. In silty zones, however, $j. is low. This suggests nonpermeable zones with very small effective porosity, as discussed above. In addition, a small $F in conjunction with high apparent porosity from conventional porosity logs can be used as an indicator of silt, Our drilling and production experience in this field has suggested a possible producing OWC at approximately 7,250-ft in Figure 2. Lateral variation of this oil-water contact OWC has also been observed. Our previous evaluation method based on conventional wireline logs fails, however, to detect any OWC in this well-track 4 indicates zones with water saturation less than 60%, an empirical water saturation cutoff based on two-phase relative permeability measurements. Utilizing the free-fluid porosity, a movable water volume, MOVWAT, is computed. As shown in track 5 , the CMR log indicates an increasing

The Log Analyst

63

Chang et al.

frequency and pulse sequence for the laboratory equipment are similar to the logging tool. For the laboratory measurements, a 10-s wait period is used to allow for complete polarization of the hydrogen nuclei. Immediately following the wait period, 4,095 spin-echo amplitudes are measured using a CPMG sequence with a 400 ps interecho spacing. As in the logging tool, the spin-echo sequences are collected in PAPS.

,/< 0 0'

,/' 0

/' 0

,/'

,/' 0

0.0

Porosity

/

,/'

0.

6.

10.

16.

20.

1

Flgure 3: Comparison of porosity by weight (Gbuoy) with N M R porosity (GNMR). The NMR porosity is on average 1.4 p.u. higher, suggesting incomplete extraction of t h e residual hydrocarbons from the samples by the Dean Stark method.

volume of movable water below 7,240 ft. From the EPT log, residual oil saturation is evaluated assuming the EPT log measures the flushed zone. A comparison to in-situ oil saturation derived from deep resistivity logs leads to an estimate of movable oil volume, MOVOIL, as shown in track S . Combination of both movable water volume and movable oil volume clearly suggests that a significant water production would be encountered below 7,240 ft. This was consistent with field experience and was later confirmed by production data from this well. In addition, the CMR interpretation suggests a possible correlation between the OWC and the lithology transition from dolomite/anhydrite to doloniite/limestone matrix. LABORATORY NMR TESTING

Twenty-seven core samples were cut parallel to bedding and cleaned in a Dean Stark extraction with toluene. Even after several months of extraction, some color remained in the effluent, indicative of a small amount of oil not fully extracted from the core. Porosity and grain density were determined by hydrostatic weighing in toluene. The samples were dried in a vacuum oven and unstressed Klinkenberg-corrected air permeability measurements were made. Optical photographs were taken to establish qualitatively the sample texture and the degree of vuggy porosity. The samples were then brine saturated and measured by NMR before and after centrifugation. NMR measurements were made on 1-in.-OD x 1.5-in.long core samples using a 2-MHz laboratory spectrometer. The NMR measurements were made at 25°C. The operating 64

Figure 3 compares the porosity by weight with the NMR porosity. The agreement is very good except the NMR porosity is on average about 1.4 p.u. higher. This is believed to be due to a small amount of unextracted heavier hydrocarbons in the cores. NMR measurements on a core sample vacuum-dried at 110°C continued to show this amount of residual signal. Producible Fluid Porosity

We want to establish a T2 cutoff in the T, distribution that will provide a good estimate of centrifugeable water in this carbonate lithology. Previous studies in sandstone showed a T2 cutoff of 3 3 ms on the logs agrees well with centrifugeable water measured in the laboratory. Figure 4 shows T, distributions for the 27 core plugs before and after centrifugation for three days at a rotational speed equivalent to 100 psi air-brine equivalent capillary pressure. As can be seen, the largest pores (predominantly vugs) have been drained by centrifugation. In order to achieve this result, the samples were centrifuged for three days. Figure 5 shows a time-lapse study of the drainage of the largest pores in three samples that indicates a long spinning time is required for complete drainage in these tight rocks. Figure 6 compares the NMR free-fluid porosity with the centrifugeable water for the Clearfork samples. Analyzing the laboratory NMR ineasurements shows that the optimal T2 cutoff for these carbonate samples is 92 nis, about three times longer than the T, cutoff for sandstones. This conclusion is based on Figure 7, which shows the error in estiniating the volume of centrifugeable water, averaged over 70 samples from the Clearfork and two other suites of carbonate samples from international oil fields, as a function of the T2 cutoff value. Vuggy Porosity

Carbonates often contain vuggy porosity in addition to intergranular and fracture porosity. Vugs are cavities formed in the matrix by diagenesis, typically by dissolution processes, and can range in size from -100 microns to cavern size. The distinguishing feature is an enlargement in

The Log Analyst

March-April 1997


# 48

16.3 pu 150 rnd

50.05 .P

----.__._. U --.- O

---

i

P 37

I

-

I

1.0

I

I

m.0

100.0 T2

I

looo.0

1

looo0.0

(ms)

Figure 4: T2 distributions for 27 water-saturated samples before (line) and after (dash) centrifugation at 100 psi air-brine capillary pressure. Sample number, porosity, and permeability are indicated to left of the distributions. Samples are ordered by decreasing permeability, from top left to bottom right. March-April 1997

The Log Analyst

65

Chang et al.

# 47 7.1 pu 1.1 md i.

Figure 6: Comparison between free-fluid porosity (@f(92)), using a cutoff of 92 ms, and the volume of centrifugeable water (@cent)'

I

1

I

I

I

1.0

10.0

100.0

1000.0

10000.0

T2

d

v

8c

(ms)

al

Figure 5: Time-lapse NMR study of water drainage by centrifugation in samples 35, 47, and 50. T2 distributions are for the samples water saturated (line), and after centrifugation for 1.5 hr (short dash), 6 hr (long dash), and 3 days (short-long dash). al

the pore geometry relative to the average intergranular pore size. Vugs may often be detected by a second peak at the long end of the T, distribution as pores larger than 100 microns in diameter have T, values greater than -1 s. However, if the intergranular porosity has large pore sizes, there may be insufficient separation between the intergranular porosity and the vugs, which results in a unimodal distribution. Figure 8 shows photographs of six core plugs and their associated T, distributions. It is clear there is a good correlation between the physical appearance of the plugs and the i", distributions. Plugs with large visible vugs have long T, components in the T, distribution; i.e., Tz > -1 s: Sample 43 has no visible vugs and unimodal T2 values less than 100 nis. indicating relatively small pore sizes. The matrix of the sample appears to have a smooth texture, which indicates fine grained matrix. Also visible on the photograph is a throughgoing fracture in the plug. 66

rn

2

aP Free Fluid Cutoff Time (ms) Figure 7: The error in estimating centrifugeable water versus the T2 cutoff used to calculate free-fluid porosity. The minimum error occurs at 92 ms. The data set includes 70 carbonate samples from the Clearfork and two international oil fields.

Sample 39 has few visible vugs but appears to be coarse grained. In this case the long T, values are associated with the intergranular porosity. The T, distribution is not bimodal. Sample 38 has intermediate values of T, with no bimodality. No vugs are visible and the texture is intermediate between samples 43 and 39.

The Log Analyst

March-April 1997


Three of the sampl s (12A, 22, and 44) have visible vugs on the photographs and also have bimodal T2 distributions. The vugs appear to be imbedded into a finegrained matrix. The low T2values correlate well with the low T2values observed in sample 43. The long T2values (1 to 3 s) are associated with the vuggy porosity. These samples have comparable porosities to the nonvuggy samples but much lower permeabilities. As shown in Table 1, if one compares the six samples with porosities in the range 8.6 to 10.9 p.u. (12a, 19,22,38, 39,41), the permeabilities vary by a factor of 65, from 0.2 md for sample 12a to 13 md for sample 39. The poor correlation between porosity and permeability is presumably due to the presence of large, separated vugs. Hence, we would expect permeability to correlate better with pore volumes that do not include the vuggy porosity. This reduced pore volume can be approximated from the T2distribution by including only those pore sizes below a certain T2 cutoff as described below.

Permeability Estimation from N M R

Permeability estimation by NMR is based on the fact that permeability has dimensions of length squared and uses the pore size obtained from NMR (Seevers, 1966; Timur, 1968, 1969a, 1969b; Kenyon, 1988). Previous studies in sandstones with intergranular porosity have used the following estimate:

-

k $4T22 ,

(2) where T2 is the mean logarithmic value of the T2 distribution. In carbonates we expect a similar expression to Equation (2) for the intergranular portion of the porosity. For these samples the intergranular pore volume may be approximated by excluding components with T2 values above a certain cutoff. That is, both the pore volume and logarithmic mean T2 are calculated from that fraction of the T2 distribution below a cutoff value. The error in the estimated permeability was calculated for various values of the cutoff. The results are shown in Figure 9. The best correlation with Equation (2) occurs with a T2 cutoff of 750 ms.

Table 1: Summary of laboratory measurements: Sample

Depth

@buoy

@NMR

@cat

@fl92)

12 18 19 20 22 29 35 36 37 38 39 40 41 43 44 45 46 47 48 49 50 51 52 53 54 55 66

7265 7259 7258 7255 7257 7248 7242 7241 7240 7239 7238 7237 7236 7234 7233 7232 723 1 7230 7229 7228 7227 7226 7225 7224 7223 7222 721 1

10.9 13.6 8.6 11.8 8.7 3.9 5.9 7.1 4.0 10.2 11.9 3.7 8.5 6.1 6.6 6.0 2.8 ‘7.1 16.3 5.4 12.2 5.0 3.5 1.9 1.1 1.5 11.2

11.0 14.4 9.3 12.6 9.9 4.6 6.9 8.0 5.3 11.5 12.9 4.6 9.5 6.5 7.8 7.2 4.7 8.6 18.2 6.6 13.1 6.0 4.3 3.6 3.1 3.5 12.4

5.5 12.0 3.7 9.4 5.7 n.a 4.0 4.8 1.8 8.8 10.7 2.0 7.2 1.7 4.9 n.a 1.7 5.3 16.0 4.1 10.0 2.5 2.1 0.7 0.6 0.6 9.3

6.4 11.5 5.9 9.3 7.9 1.9 4.2 5.1 2.3 6.5 11.4 2.0 7.2 0.6 5.3 4.4 2.3 6.4 16.3 5.0 10.9 2.8 2.6 2.0 1.8 1.3 7.6

March-April 1997

The Log Analyst

T2,105?

156.5 235.9 219.7 220.6 527.6 61.9 143.8 220.4 71.9 91.4 534.2 62.1 251.5 15.7 310.8 137.2 76.8 314.9 346.1 328.0 304.8 75.1 145.5 84.9 119.3 50.5 122.9

kair

0.240 32.0 2.10 20.0 0.690 fracture 0.230 0.063 0.040 12.0 13.0 fracture 4.30 0.480 0.069 0.890 0.038 1.10 150.0 0.064 7.50 0.045 fracture 0.027 0.041 0.026 5.60

m

m-NMR

2.11

2.02

2.10 2.00 2.19

2.07 1.98 2.17

2.05

2.05

2.02

2.04

1.73 2.09 1.95

1.92 2.10 2.00

2.13 2.03

2.07 1.99

1.93

1.97

1.95

1.92

67

Chang et al.

i 39

11.9 pu 15 md I

10.9 pu

02 md #r2

dCLLL

0.7 pu 0.7 md

Figure 8: Optical photographs and T2 distributions for six samples give a qualitative estimate of the vuggy porosity and coarseness of the grain structure. 6X

The Log Analyst

March-April 1997

Effective Porosity, Producible Fluid, and Permeability in Carbonates from N M R Logging

0,

I

I

L

2

15

-1.1 -1.8

2

i'-a

-'.9 -2 -2.1 -2.2

-2.2

-2.1

-2

-1.9

-1.8

-1.7

m Measurement

Figure 12: Correlation between measured cementation exponent on 13 samples and values computed from the T2 distributions using pore combination modeling.

correlation between permeability and Equation (2) is better than with total porosity alone. (The standard deviation decreases from a factor of 3.5 to 2.6.) Whereas Equation (2) has a fixed exponent of 4,the best fit with porosity alone occurs for a porosity exponent near 5 . The prefactor in

Figure 11: Crossplot of NMR porosity (@NMR) versus measured air permeability (kair).The best fit (i.e., the dashed line) has a slope of 4.9. The standard deviation is a factor of 3.5.

Equation (2) is 4.75, which is similar to results in sandstones.

Figure 10 shows the best correlation of air permeability versus Equation (2), while Figure 11 shows the best comelation of air permeability versus total porosity. Four Samples with fracture porosity have been excluded. Clearly the

Although the permeability correlation with total porosity iS relatively good for this Suite O f cores, the porosity exponent will in general not be known unless core is available. This suggests another advantage in using Equation (2) since the exponents of both porosity and T, are predetermined.

March-April 1997

The Log Analyst

69

-

Chang et al.

K,

1.E-01

Porosity (p.u)

(md)

1.€+00 l.E+01

l.E+02

G 0

-

3 ........

.................................... L

....... .....

L Figure 13: Log showing CMR porosity (solid line) and the pore volume with T2 greater than 750 ms (dashed line). The CMR permeability was calculated using 4.75 ( ~ N M R , ~ ~ O ) ~ (Cored T ~ , ~ interval ~ O ) * . is indicated by the sketch to the right. Diagonal lines on the sketch indicate intervals for which vugs were observed in the core.

70

The Log Analyst

March-April 1997

Effective Porosity, Producible Fluid, and Permeability in Carbonates from NMR Logghg

Cementation Factor m

Another petrophysical parameter that depends on the fraction of vuggy porosity is the cementation factor m in Archie's relation,

F=#",

(3)

since the vuggy porosity contributes weakly to electrical flow. Thus one expects m to increase with the relative amount of vuggy porosity. Previous studies (pore combination modeling, PCM) have established a relationship between m, intergranular porosity, and total porosity (Myers, 1991): (4)

This relation is used by SWEPI in conjunction with point count estimates of intergranular porosity. Electrical resistivity measurements were made on a subset of 13 brine-saturated plugs at 25°C. Three-point salinity measurements were made and a straight line fit through the origin was used to measure the formation resistivity factor. Figure 12 shows there is a good correlation between tn from resistivity and nz from the pore combination model (PCM) using the fraction of vuggy porosity computed from either a 1-s T2 cutoff for samples with unimodal T, distributions or a 642-ms cutoff for samples with bimodal T2 distributions. The values of m range from 1.74 to 2.21, whereas conventionally a value of m = 2 would be assumed for this reservoir. In particular, the sample with no visible vugs (#43) and no long T2 components has the lowest m value, and the sample with the most visible vugs (#22) has the highest nz value. The anomalously low ni value in sample 43 is due to the throughgoing fracture.

accurate determination of pay. In the presence of microporosity in nonproductive silts, the CMR porosity estimates effective porosity, a direct link to the productivity in this reservoir. Silty zones can be identified by utilizing the NMR free-fluid index in conjunction with high apparent porosity from conventional porosity logs. Producible water volume can be computed based on the bound-fluid volume and the water saturation from deep resistivity logs. A T2cutoff of 92 ms is found to be optimal for this dolomite lithology as well as for two international oil reservoirs. The CMR tool enabled the OWC to be detected in this well, which was not possible with the conventional log interpretation. Improved correlations for permeability and cementation factor have been evaluated using the NMR Tz distributions for carbonate samples containing significant amounts of separated vuggy porosity. NOMENCLATURE

F It1

Mo S T2 T2B T2,750

v 4 #buoy @cent

#CON h 9 2 )

@in,

@NMR

Core-Log Synthesis

@NMR,750

In Figure 13, laboratory data are used to synthesize a log of permeability and vuggy porosity. The left track in Figure 13 shows the CMR permeability calculated using 4.75 (@Nh4r/750)4(T2,750)2 .

The right track shows the CMR porosity (solid line) and the pore volume with T2 > 750 ms shows a high degree of correlation with vugs observed in the core. The average fraction of vuggy porosity from point counting the vuggy intervals (26%) agrees with the log average over this interval. CONCLUSIONS

In complex carbonates, the CMR tool provides a lithology-independent porosity measurement and allows a more March-April 1997

P 2

formation resistivity factor cementation factor magnetization of hydrogen nuclei pore surface area NMR transverse relaxation time NMR transverse relaxation time for bulk fluid log mean T, calculated from that portion of the distribution with T2 < 750 ms pore surface volume total porosity core porosity by weight centrifugeable porosity conventional porosity based on SWEPT'S multilog evaluation technique free-fluid porosity using cutoff of 92 ms intergranular porosity NMR porosity NMR porosity with T2 < 750 ms surface relaxivity ACKNOWLEDGMENTS

We thank D. Rossini of Schlumberger-Doll Research for performing the laboratory NMR measurements; M. Myers of Shell Development Co. and P. Dryden of Schlumberger-Doll Research for resistivity measurements; R. Shew of Shell Development Co. for point counting; and J. Rohan of Shell Development Co. for centrifugation. REFERENCES

Clerke, E. A., Williams, K. W., and Pearce, L. A., 1993, The DAK formation evaluation model for the Permian basin Clearfork, SPE-26264: Society of Petroleum Engineers, Richardson, Texas. Howard, J. J. and Spinler, E. A., 1993, NMR measurements of

The Log Analyst

71

Chang et al.

wettability and fluid saturations in chalk, SPE-2647 1 : Society of Petroleum Engineers, Richardson, Texas. Kenyon, W. E., Day, P. I., Straley, C.. and Willemsen, J. F., 1988, A three-part study of NMR longitudinal relaxation studies of water-saturated sandstones: SPE Forniatiori Evalimtion, v. 3. p. 622436. Kenyon. W. E., 1992, Nuclear magnetic resonance as a petrophysical measurement: Nirclenr. Geophvsics, v. 6, p. 153-1 7 I . Kleinberg, R. L.. Straley. C.. Kenyon, W. E.. Akkurt,R., Farooqui, S. A,, 1993. Nuclear magnetic resonance of rocks: T , vs T?. SPE-26470: Society of Petroleum Engineers, Richardson, Texas. Myers, M. T., 1991. Pore combination modeling: A technique for modeling the permeability and resistivity properties of complex pore systems, SPE-22662: Society of Petroleum Engineers, Richardson. Texas. Morriss, C. E., Macinnis, J., Freedman. R., Smaardyk, J., Straley. C., Kenyon, W. E, Vinegar, H. J., Tutunjian, P. N., 1993, Field test of an experimental pulsed nuclear magnetism tool, paper GGG. in 34th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts.

72

Seevers, D. 0.. 1966, A nuclear magnetic method for determining the permeability of sandstones. paper L, in 7th Annual Logging Symposium Transactions: Society of Professional Well Log Analysts. Straley, C.. Morriss, C. E., Kenyon. W. E., Howard, J. J.. 1991, NMR in partially saturated sandstones: Laboratory insights Into free fluid index, and comparison with borehole logs, paper C, in 32nd Annual Logging Symposium Transactions: Society of Professional Well Log Analysts. Timur, A., 1968, An investigation of permeability, porosity and residual water saturation relationships for sandstone reservoirs: The Log Analyst. July-August. Timur, A., 1969a, Producible porosity and permeability of sandstones investigated through nuclear magnetic resonance principles: The Log Ana(i~st,January-February, p. 3-1 1 . Timur, A,, 1969b. Pulsed nuclear magnetic resonance studies of porosity, movable tluid and Permeability of sandstones: Jouriial of Petroleum Technology, v. 246, p. 775-786. Timur. A,, 1972, Nuclear magnetic resonance study of carbonate rocks: The Log ..lnuIysr, v. 13, no. 5. p. 3-1 1 .

The Log Analyst

March-April 1997