CALIBRATING OLD, NEW LOGS TO CORES NEEDED FOR PLANNING CO2 FLOOD

Dennis W. Dull
Texaco Exploration & Production Inc.
Midland, Tex.

As part of reservoir characterization and for calculating original oil-in-place, correcting porosity logs to core data was necessary to predict the effectiveness of CO2 injection into the Mabee field.

Of the Mabee field's more than 800 logs, a majority are old gamma ray-neutron logs.

The modern porosity logs were calibrated to core porosity by crossplotting log against core porosity. Linear regressions were constructed which are defined by the slope and the y-intercept.

The old neutron logs demonstrated a good inverse linear relationship between core porosity and the log10 of the neutron deflection.

MABEE FIELD

Texaco Exploration & Production Inc. is targeting the Mabee field as one of three fields for enhanced oil recovery with CO2- Mabee field is located in the Permian basin (Fig. 1).

The field study provided an accurate reservoir description that would not only support the past history of the field but would predict future reservoir performance and recoveries and help monitor the CO2 miscible flood.

One of the major tasks in the reservoir description was to determine the original oil-in-place with all available log and core data. Obtaining reservoir height (PHIxH) required calibrating modern porosity logs to core porosity and normalizing old neutron logs.

Of the more than 650 old neutron logs in the Mabee field, about 75 cannot be used at all because these wells were stimulated with nitroglycerine.

The Mabee field also has about 150 modern porosity logs. In addition, approximately 85 wells have been cored at the Mabee field, of which 35 have the core report but no actual core.

All of the logs and core data have been digitized. Before the logs were sent out for digitization, all pertinent information such as logging company, tool model number, hole size, casing point, casing size and weight, source to detector spacing, etc. were recorded and entered on to a spreadsheet to be used in calibrating the logs to core porosity.

The log analysis, mapping, and data base management necessary to obtain PHI x H were done on a personal computer. Without personal computers, the study could not have been finished within the time constraints.

GEOLOGY

Discovered in October 1943, the Mabee field covers an area of 12,800 acres and is located east of the Central basin platform in the central portion of the Midland basin. (Fig. 1).

The Mabee field produces from the San Andres formation of Permian Guadalupian age. Although isolated from similar San Andres production, the favorable reservoir facies was draped over paleostructure/topography of Early Pennsylvanian age.

The Mabee field has produced more than 90 million bbl of oil and is currently producing about 8,000 bo/d.

The San Andres production is from a dolomite reservoir, a time-transgressive sequence that prograded from southeast New Mexico southward across the Midland basin.1

The San Andres of the Mabee field is composed of six distinct facies typical of a sabkha-type environment such as found in the present day Persian Gulf. The six facies are:

Supratidal (anhydrite rich, permeability barrier responsible for trapping the oil)
Oncolites/pisolites
Subtidal
Ooid
Sandstone
Open marine.

The productive sequence is almost exclusively confined to the subtidal and ooid facies. The reservoir at the Mabee field has been divided into three zones.

Zone 1 is capped by a very thin clay-rich stratigraphic marker known as the B. This marker is easily identified on the logs by its characteristic high-radioactive gamma ray response.

Below the B marker is the supratidal facies, composed of dolomite, nodular anhydrite, and stromatolitic lamina.

Below the supratidal facies is a mixture of subtidal mudstone to wackestone to peloid packstones and subtidal oolite packstone to grainstones.

Zone 2 is composed of primarily a sandstone and ooid facies. The sandstone facies, except on rare occasions when porosities reach 15%, is impermeable, nonreservoir rock. The sandstone facies can be generally identified on the logs by their associated high gamma ray response when compared to the clean, low gamma ray of the ooids.

Zone 3 is dominated by the ooid facies, vuggy porosity, solutioning, fractures, and high porosities and permeabilities.

This zone typically produces high volumes of water with significant H2S.

Zone 3 has produced considerable amounts of oil but because of the high porosities and permeabilities will not be flooded because of the potential for thieving of the CO2.

The interval to be flooded, gross pay, as used in this article averages 115 ft in thickness and consists of Zones 1 and 2 and excludes the sandstones.

LOG ANALYSIS

The log analysis was completed in two steps. The first step was the analysis of the modern porosity log vs. core porosity. The second was to establish a relationship between core porosity and old neutron-log deflection.

NEUTRON-DENSITY LOGS

Log-analysis software was used to crossplot core porosity against neutron-density crossplot porosity (PND). The regression work indicated that a first degree polynomial fit the data best (Fig. 2a).

In other words, there was a linear relationship between neutron-density crossplot porosity and core porosity.

Individual plots of core porosity vs. PND were made for 16 wells over gross pay. Equations of the line, slope, and y-intercept, along with correlation coefficients were generated with the log-analysis software. (Note: For statistical purposes, it is extremely important that the interval be large enough to be significant and correlative from,well to well.)

The results of the linear regressions are shown in Table 1. All of the wells exhibit a high correlation coefficient. (A correlation coefficient of 1.00 would indicate a perfect linear correlation.)

With the exception of Well A-1 No. 483, all wells were used in calibrating the PND curves to core. Well A-1 No. 483 has an anomalously low slope but high correlation coefficient. This is believed to be the result of drilling with oil-base mud. All other wells were drilled with brine.

Logging Company A used the same neutron and density tools, with the exception of Well A-1 No. 574 which had a slightly different neutron tool. The linear regression slopes varied from 0.708 to 0.996. The y-intercepts varied from 0.009 to 0.026.

Company B also demonstrated similar variability even though using the same logging tools.

Despite the variability in slope and y-intercept, the linear regressions had a high correlation coefficient. This variability in slopes and y-intercepts is attributed to the changes in geology (lithology, porosity types, and percentages) and changes in salinity due to waterflooding with freshwater.

In other words, the slopes and y-intercepts are more of a function of where the wells are drilled than the logging company.

An example of this is Well A-4 No. 69 logged by Companies A and B. The linear regressions generated slopes and y-intercepts that are very close (Table 1).

If geology or location is the controlling factor, then mapping of the slopes and y-intercepts should reflect a gradual change across the field when contoured. In addition, slopes and y-intercepts should be predictable.

Fig. 3 illustrates the maps of the slopes and y-intercepts of the linear regression of core porosity vs. neutron-density crossplot porosity.

Well A-1 No. 648 was cored and logged after the map was constructed. The map predicted a slope of 0.86 and a y-intercept of 0.014. Table 1 shows the actual slope and y-intercept to be 0.88 and 0.018, respectively.

DENSITY VS. CORE POROSITY

Linear regression analysis used log-analysis software for density porosity on a dolomite matrix of 2.87 g/cu cm vs. the core porosity. This was done for two reasons.

First, because the San Andres at the Mabee field is a known dolomite reservoir, the density porosity was verified to have a good correlation with core porosity .

Second, the logging tools were stacked with the neutron tool on top. This left the bottom portion of the pay section with only the density porosity.

No rathole was obtained for logging because of the high water volumes encountered when drilling into Zone 3 and its high H2S content.

Fig. 2b shows the crossplot of the density porosity (Pddol) against core porosity.

Table 2 lists the slopes and y-intercepts of the linear regressions and their associated correlation coefficients for density porosity vs. core porosity for 16 wells.

The linear regressions produced a good correlation between density porosity when crossplotted with core porosity.

As with neutron-density crossplot porosity vs. core porosity, the density porosity vs. core porosity showed a gradual change of slope and y-intercept of the linear regressions across the field.

The slopes and y-intercepts are controlled more by where the well was drilled or geology than logging company. Again, A-4 No. 69 had similar slopes and y-intercepts for both logging companies A and B (Table 2).

In addition, as with the slopes and y-intercepts of the A-1 No. 648 of the neutron-density crossplot porosity vs. core porosity, the linear regression of density porosity crossplotted against the core porosity had a slope and y-intercept very close to the predicted value from the maps. The predicted values of slope and y-intercept from the contoured values were 0.725 and 0.01 1 with the actual being 0.765 and 0.012.

NEUTRON VS. CORE POROSITY

The cased-hole neutron-porosity analysis did not exhibit the same relationship as the open hole porosity logs. Well A-4 No. 69 was logged by four different logging companies.

Linear regressions of log porosity on a dolomite matrix (Pndolch) vs. core porosity were done. The slopes and y-intercepts show a significant difference (Table 3).

Notice that all companies have a high correlation coefficient indicating a good linear response for each company's calculation of porosity. It appears from this that the logging company does make a significant difference in the relationship between core and cased-hole neutron log porosity.

Therefore, mapping of slopes and y-intercepts regardless of logging company to convert log porosity to core porosity would be impossible. However, mapping slopes and y-intercepts by logging company would be a solution providing there is enough core and wells logged by a specific company.

TRANSFORMING POROSITY

The log porosities were transformed to core porosity with the slope and y-intercept for the contoured values and application of that transform to that specific well.

In other words, instead of one transform for all the wells logged by a specific logging company, there would be a different transform for every well. To verify the accuracy of the transform, the pseudocore porosity was compared to the actual core porosity for all wells used in the analysis (Fig. 4a).

Once this relationship had been established, the transforms were obtained from the maps and used to convert the log porosity to pseudocore porosity of any well in the field.

In regards to the cased-hole porosity logs, there was only one well, logged by Company D, that went through gross pay. The maps of y-intercept and slope of Company D were the only ones necessary to convert the well's log porosity to pseudocore porosity.

Table 4 shows the porosity times gross thickness (PHI x H) of the cored wells to their core transforms.

OLD NEUTRON LOGS

The conversion of log porosity to pseudocore porosity was necessary to convert accurately the old neutron logs to porosity. The more core data, the better the control of porosity that could be applied to the old neutron logs.

The ideal way to transform old neutron logs to core porosity is to have a core in every well, obviously that situation usually does not exist. However, there were 13 wells with cores and old neutron logs over gross pay.

The relationship between neutron log deflection and porosity was demonstrated by Brown and Bowers.2 They discovered that there is an inverse linear relationship between porosity and the log10 of the neutron deflection measured from neutron zero.

Fig. 2c shows an example of this relationship. In calibrating neutron logs at Sacroc, Swulius3 discovered that the same transform could be obtained with statistical descriptors in place of the entire core. Those statistical descriptors were the maximum, minimum, and mean of core porosities vs. log10 deflections.

The most unreliable descriptor was the relationship of the log10 of the minimum neutron deflection to maximum porosity. Probably this is in part due to the low count rates in the high porosities.

Fig. 5 shows the linear regressions of two wells using:

Core porosity vs. log10 deflection
Maximum, minimum, and mean values of core vs. log10 deflection
Mean and minimum of the log porosity vs. mean and maximum of log10 deflection
Mean and 0.015 (field minimum) porosity vs. mean and maximum of log10 deflection.

The two examples demonstrate that using field minimum porosity or minimum porosity and mean porosity vs. the mean and maximum of the logs neutron deflection nets nearly the same result as using all the core data vs. the log data.

In other words, the statistical descriptors worked as well as if all the data had been used. The significance of this, providing there is ample core data and that the neutron log is over gross pay, is that the mapping of the mean porosity across the field would allow the calibration of any old neutron log to core regardless of logging company, tool model number, hole size, cased or open hole, etc.

Table 5 presents the results of the regression of the 13 wells of core porosity vs. logs deflection. Table 5 demonstrates as Fig. 5 illustrates that using the statistical descriptors of core (mean and minimum) is sufficient for obtaining the slope of the line, therefore, the transform for converting log10 deflection to porosity providing that these logs cover the entire gross pay.

In Mabee there were 29 cores and 28 porosity logs employed in generating the mean porosity map. Of the 29 cores, 16 wells had both core and modern open hole porosity logs and 15 had cased-hole neutron-porosity logs.

Thirteen wells had core porosity over gross pay with old neutron deflection curves. All neutron logs over gross pay were transformed to porosity by crossplotting mean and maximum of log10 neutron deflection against the mean (obtained from contoured value on the map of the mean porosity) and field minimum porosity (0.015).

This generated a regression equation that then was applied to the log10 of the neutron deflection curve to transform it to porosity.

Fig. 4b illustrates the transforms for Well No. 105. This log compares core porosity, core transform porosity, and pseudocore transform porosity (with maximum and mean log10 neutron deflection vs. 0.015 and mean porosity of the core data to generate an algorithm for neutron log transformation to porosity).

Well No. 105 shows excellent agreement between transform porosities.

RESULTS

This study demonstrated that:

The neutron-density and density porosity demonstrated an excellent linear correlation to core porosity that depended more on where the well was drilled than the logging company.
The relationship of neutron-density and density porosity to core porosity for any one logging company varies in the Mabee field reflecting changes in geology.
The cased hole neutron porosity log response displayed a good linear response to core porosity but indicated a dependence on logging company.
The linear correlation of the cased hole neutron porosity log to core porosity for any one logging company varied across the field as did the neutron-density logs mirroring changes in lithology.
The neutron-density, density, and cased hole neutron porosity logs were transformed to pseudocore porosity utilizing the maps of the slopes and y-intercepts of the linear regressions of log porosity crossplotted against core porosity.
The old neutron logs exhibited an inverse linear response of the log10 neutron deflection when crossplotted against core porosity.
The statistical descriptions of mean and field minimum porosity (0.015) crossplotted vs. the mean and maximum logs neutron deflection generated nearly the same slope and y-intercept of the linear regression as applying all the core and log data.
The mapping of the mean porosity from the cores and the transformed porosity logs would enable the generation of a transform to convert log10 neutron deflection over gross pay to porosity.

ACKNOWLEDGMENT

The author is grateful to Texaco for permission to publish this article and to the CO2 department, Midland producing division, for the encouragement and support in assembling it. Thanks are also due Lois Folger for her assistance in the analysis of the old neutron logs and Tekla Dupuis for her help in producing the maps and log data base.

REFERENCES

Todd, R.G, "Oolite-bar Progradation, San Andres Formation, Midland Basin, Texas," AAPG Bulletin, Vol. 60, 1976, pp. 907-25.
Brown, A.A., and Bowers, B., "The Relationship Between Neutron Log Deflection and Porosity," CWLS, Canadian Symposium Papers 1, No. 39, 1957, pp. 39-43.
Swulius, T.M., "Porosity Calibration of Neutron Logs, Sacroc Unit," Journal of Petroleum Technology, April 1986, pp. 468-76.

BIBLIOGRAPHY

Bebout, D.G., and Harris, P.M., editors, "Geologic and Engineering Approaches in Evaluation of San Andres/Grayburg Hydrocarbon Reservoirs - Permian Basin," Bureau of Economic Geology, 1990.
Ghosh, S.K., and Friedman, G.M., "Petrophysics of a Dolostone Reservoir: San Andres Formation (Permian)," West Texas, Carbonates, and Evaporites, Vol. 4, 1989, pp. 45-119.