Galford, J.E., SPE, Schlumberger Well Services Flaum, C., SPE, Schlumberger Well Services Gilchrist Jr., W.A.,* SPE, Schlumberger Well Services Soran, P.D.,** SPE, Schlumberger Well Services Gardner, J.S.,+ SPE, Schlumberger Well Services
Summary. The basic openhole responses and environmental correction algorithms for compensated neutron logging (CNL(TM)) tools have been updated. The improved processing is based on an extensive set of laboratory formation measurements to which mathematical modeling calculations have been added. In all, the new algorithms include basic responses for the three principal formation matrix types and corrections for seven environmental effects and formation-fluid salinity. A total of 467 laboratory formation measurements have been augmented with 245 data points generated through mathematical modeling. This data base has been used to define more accurately the effects on the tool response of variations in logging conditions from those considered standard in the laboratory. More accurate corrections for the effects of formation pressure, temperature, mudcake, natural or barite mud, and borehole salinity have been defined. Certain other effects depend on more than one parameter. For example, the effect of formation salinity is somewhat matrix-dependent; therefore, the corrections are handled differently for sandstone, limestone, and dolomite. The effect of tool standoff depends on the borehole size; consequently, the standoff correction is larger for larger boreholes. The porosity crossplots and environmental correction charts based on the new algorithms represent a significant evolutionary improvement over previous techniques. They should be an important aid to the use and interpretation of neutron logs.
Introduction The openhole porosity response functions and environmental correction algorithms for the CNL log have been updated. The new algorithms have been field tested and will become available on a routine basis both at computing centers and on field service units. This paper presents a discussion of the new procedures. Environmental correction charts and interpretation crossplots similar to those published by Schlumberger are also included. Finally, log examples are used to illustrate some of the differences between the new algorithms and those they are replacing. Neutron logging has been an important measurement for porosity and lithology determination for more than 20 years. The dual-spacing CNL log was first described in 1971 by Alger et al. and has become a mainstay measurement for formation evaluation. In combination with other services, the CNL log provides information on porosity, lithology, and the presence of gas. The dual-detector design with ratio processing was chosen to reduce environmental effects on the measurement. The downhole device has remained virtually unchanged through the evolution from analog surface instrumentation to the sophisticated computer-based units of today. Even with the reduced environmental effects of the dual-detector design, it is still necessary to apply corrections for certain borehole conditions and for variations in formation-fluid salinity. Correction plots have been published in Ref. 1. Environmental corrections have been routinely applied to field logs before processing at field log interpretation centers. Corrections for a subset of the effects have been used at the wellsite on the Cyber Service Unit (CSU(TM)). Over the years, our understanding of the neutron measurement has continued to improve through the acquisition of additional laboratory measurements and, more recently, through mathematical modeling. This has been reflected from time to time in improvements to response functions and environmental correction algorithms and in periodic upgrades to the log interpretation charts. In addition, several papers have been published that deal with specific areas of CNL response or interpretation.
Laboratory and Modeling Data The new response functions and environmental correction algorithms discussed in this paper are based on an extensive collection of more than 467 laboratory measurements. Measurements were made in limestone, quartz sandstone, and dolomite at porosities ranging from zero to the water point. Two types of laboratory formations were used, homogeneous quarried rock samples and test-tank formations made in the laboratory for specific measurements. The formations used were produced and characterized in the same way as those described by Tittman et al. Each measurement was made for a period of time sufficient to reduce statistical error well below the uncertainty in our knowledge of the porosity of the formations. The estimated uncertainty in the formation porosities was determined as discussed by Tittman et al. Although many of the measurements were made to improve the characterization of the CNL porosity response, most of the data were taken to aid in understanding the environmental effects. Data for variations in hole size, tool standoff, mudcake, mud weight, and borehole salinity were included. Measurements for various formation-fluid salinities were also made. In addition, measurement for certain combinations of effects, such as hole size and standoff, were part of the data base. In addition to the laboratory data, 245 data points were generated by mathematical modeling. Both Monte Carlo and deterministic techniques were used as described in earlier publications. Monte Carlo calculations were used for cases where a three-dimensional treatment was necessary. In some cases, however, it was possible to use a two-dimensional (2D) deterministic model described by Ellis and Case. All modeling data were calculated from the reaction rate of He(n, p) with 100% efficient detectors and no dead time. Consequently, it was necessary to normalize the calculated results so that they could be used with the experimental data. The calculations used for this study were normalized to measurements in the 13.2-p.u. freshwater-filled limestone formation with an 8-in. [20-cm] -diameter freshwater-filled borehole. The uncertainties for the calculated data were determined from the statistical precision of the results. Modeling points were used to extend the laboratory data into areas that were difficult to measure experimentally. Several calculations were used in the critical range between 0 and 15 p.u.
SPEFE
P. 371^