Dennis F. Wyatt Jr.Halliburton Logging Services Inc.Houston
Recent improvements in detector designs are allowing new carbon/oxygen logging tools to give reliable measurements at substantially higher logging speeds.
New detectors on the tools provide about a 100% increase in the dynamic range in the carbon/oxygen ratio, and much better statistical precision than previous systems. These new tools can eliminate the need for station measurements.
Carbon/oxygen (C/O) logging tools are cased-hole pulsed neutron devices that use induced gamma spectroscopy to determine oil saturation in reservoirs having low (less than 30,000 ppm NaCi) or unknown water salinity. Additionally, elemental analysis of the measured spectra determines lithology independent of other data. Although recent and ongoing work has expanded the use of these tools into complex lithology and geochemical analysis, their primary application remains the determination of oil saturation in cased holes.
PULSED NEUTRON DEVICES
Because casing is electrically conductive, the electromagnetic tools which are run in open holes to determine water saturation cannot be used in cased wells. Therefore, pulsed neutron tools are employed.
The two main types are the pulsed neutron capture tools and the induced gamma ray spectroscopy tools.
Pulsed neutron research dates back to the early 1960s; however, commercial tools were not available until the mid-1970s.
Pulsed neutron capture (PNC) tools generate fast, high-energy neutrons. The generator is pulsed at a rapid rate to produce large numbers of neutrons in short bursts.
Two interactions that occur between the neutrons and the materials through which the neutrons travel are inelastic scattering and thermal neutron capture. Each interaction produces gamma rays that subsequently undergo Compton scattering.
During inelastic scattering, which essentially occurs only during the neutron burst, a fast neutron collides with the nucleus of an atom and loses energy to that nucleus.
The nucleus releases some of its excess energy in the form of gamma rays. These energies characterize the particular element involved in the interaction.
After a number of collisions with atomic nuclei, the neutron is slowed to thermal energy, 0.025 ev (electron volts) at which thermal neutron capture can occur.
In this interaction, the neutron is captured by the nucleus of an element, and a new isotope of the element is formed. Any excess energy possessed by the new isotope is usually released in the form of gamma rays. These energies also characterize the particular element involved.
A measure of the probability that a material will capture thermal neutrons is the thermal neutron macroscopic capture cross section (S). S is expressed in capture units (c.u.). Typical values are 20-22 c.u. for oil, 22 c.u. for freshwater, and 59 c.u. for 100,000 ppm NaCl water.
S for gas depends upon gas temperature and pressure and commonly ranges from 5 to 10 c.u..
Common pure matrix capture cross sections are 4.6 c.u. for quartz, 7.1 c.u. for calcite, 4.7 c.u. for dolomite, 12.6 c.u. for anhydrite, 18.6 c.u. for gypsum, and 761 c.u. for halite. In general, most reservoir rocks have somewhat higher cross sections because of the presence of impurities and trace elements.
A gamma ray spectrum is a plot of gamma ray intensity-vs.-gamma ray energy. Were it not for Compton scattering, each of the gamma ray spectra from neutron inelastic scattering and thermal neutron capture would be only a finite set of points.
Each point would correspond to the intensity of the gamma ray energy characteristic of one of the elements involved in the neutron interactions. However, Compton scattering causes the gamma ray energies to be continually reduced. This, together with the energy "smearing" in the detector, causes the intensity values at the characteristic energy points to decrease and also the spectrum to become continuous.
The characteristic energy points become spectral peaks that, if Compton scattering is substantial, are difficult to discern from the surrounding Compton-scattered energies, or noise. Hence, for measuring gamma ray spectra, the detector and tool should minimize Compton-scattering effects and produce prominent spectral peaks.
TOOL COMPARISON
Detectors for some pulsed neutron tools are designed to sense and count gamma rays arriving in selected time intervals and time gates, and thereby can produce a decay curve of gamma ray intensity-vs.-time.
Other detectors are able to distinguish, in each time gate, the gamma rays arriving in selected energy ranges and energy windows. These detectors produce gamma ray spectra.
PNC tools produce decay curves from which the rate of thermal neutron population decays in the downhole environment determine the formation capture cross section (Sfm) (Figs. 1 and 2).
If the formation water is saline, Sfm can distinguish water from hydrocarbon.
From S, it is evident that a saltwater zone with moderate-to-high porosity will have a significantly higher Sfm than an identical zone containing oil.
In low-porosity zones, where fluid volume is relatively small, the difference in Sfm values between saltwater and hydrocarbon zones may be very small and difficult to reliably discern on a PNC log. Also, a freshwater zone and an otherwise identical zone containing oil will have virtually the same Sfm and most likely will be indistinguishable on a PNC log.
Pulsed neutron spectral (PNS) or C/O tools produce inelastic and capture gamma ray spectra as illustrated in Fig. 3. Spectral peaks related to such elements as carbon, oxygen, calcium, silicon, iron, and hydrogen that are present in the formation can be identified and analyzed to determine hydrocarbon content and lithology.
During inelastic scattering, a ratio, RC/O, can be formed of gamma ray count rates in energy ranges characteristic of gamma rays scattered from carbon and oxygen, respectively.
This carbon/oxygen ratio is related to the relative abundances of carbon and oxygen downhole. Because oil contains carbon but no oxygen, and water contains oxygen but no carbon, an oil zone of moderate-to-high porosity will have a higher RC/O than an identical zone containing water.
Because RC/O is unaffected by water salinity, C/O logs can distinguish an oil zone from a water zone, regardless of water salinity. However, because gas contains only a small amount of carbon per unit volume, RC/0 cannot reliably distinguish gas from water zones.
RESPONSE CHARACTERISTICS
The left-hand chart in Fig. 4 shows how RC/0 varies with porosity, lithology, and oil saturation. The response curves were generated from measurements using 7-in. OD casing that was freshwater-cemented in a 10-in. borehole.
Here, the tool's RC/O dynamic range is defined based on the measured response in a 36-p.u., oil-saturated sandstone and in a 35-p.u., water-saturated sandstone.
Under these conditions, the minimum RC/O value, about 0.485, is in the water-saturated sandstone, while the maximum value, about 0.575, is in the oil-saturated sandstone. This then gives a dynamic range of about 18.6%.
Recent algorithm upgrades have improved the dynamic range to 22%.
Also note from Fig. 4 that RC/O is much higher in limestone than sandstone because of the carbon in the rock matrix. Therefore, increases in RC/O can be caused either by increased oil saturation or carbonate lithology.
A supplemental method to distinguish these two situations measures the ratio of calcium to silicon.
Because silicon and calcium also have substantial inelastic gamma rays, a ratio RCa/Si (the calcium/silicon inelastic ratio curve) can be produced from these spectra. This ratio will help distinguish lithology changes without being sensitive to oil saturation changes.
The right-hand chart in Fig. 4 shows how RCa/Si varies with lithology and porosity. Again, the response curves assume the same borehole conditions.
The RCa/Si dynamic range is defined based on the response in a 26 p.u. limestone and in a 36-p.u. sandstone. Under these conditions, that the maximum RCa/Si, about 1.81, is in the limestone (calcium volume (VCa) = 1), while the minimum value, about 1.54, is.in the sandstone (Vca = 0).
Again, this is a dynamic range of about 17.5% that now with new algorithms is nearly 22%.
PRIMARY LOG CURVES
The RC/O curve (carbon/oxygen inelastic ratio) is probably the most important curve on the PNS log. In the simplest situations, RC/O increases as oil saturation increases and as the volume of limestone increases relative to the volume of sandstone.
The RCa/Si curve is a good lithology indicator and helps to differentiate oil-saturation effects from lithology effects on the RC/O curve. RCa/Si increases as the volume of limestone increases relative to the volume of sandstone.
The ratio of net inelastic count rate to total capture count rate (RIC) in the 1.5-to-8.5 mev range is a good neutron porosity indicator.
RIC increased as porosity increases. A drawback, however, is that this curve is somewhat salinity and borehole sensitive.
A number of elemental indices are available from the processing of the capture spectra that can be used for lithology determination.
One such curve is a potassium indicator that aids in identifying zones containing day. Potassium values generally increase in shales and decrease in clean formations.
Other quantities measured are silicon, calcium, hydrogen, chlorine, iron, and sulfur. These real time indices are only pseudo-yields, and not calibrated. Therefore, these indices are useful only for qualitative interpretation.
To allow quantitative lithology analysis, elemental yields computed from fitting pure element spectra to the measured spectra with a weighted-least-squares procedure is required and can be provided as a post-log computed product.
With elemental yields from the capture spectra, an interpretation model can determine formation lithology.
Previously published articles have given general forms for this conversion, but most of the algorithms remain proprietary.
C/O LOGGING LIMITATIONS
Historically, C/O tools have been run at very slow logging speeds, 1-2 fpm, and combined with stationary readings to reduce statistical variations. Even then, results were frequently marginal.
This poor repeatability was primarily because of the small dynamic range of the C/o measurement, typically on the order of 10% or less. In the past, this limited C/O logs to reservoirs with porosities exceeding 20% and with good borehole conditions.
Care must be taken in interpreting RC/0 because changes in the carbon content of both the formation matrix and the fluid in the pores affect the ratio. For this reason, inelastic RCa/Si and similar ratios are included on C/O logs to identify lithology changes.
A limitation is dolomite (CaMg(CO3)2). Here, the RC/0 will respond to the carbon in the matrix but the RCa/Si will not completely normalize for the carbonate mineralogy because of low calcium content.
NEW PNS TOOLS
After several years of research and development, new PNS tools entered commercial service in late 1990. These new tools employ high-density scintillation detectors.
This new detector provides about a 100% increase in the dynamic range of RC/0, and much better statistical precision than previous systems.
As shown in Fig. 5, one such tool string is 33.5 ft long and includes collar locator, telemetry, and gamma ray assemblies.
This tool's detector/generator and electronics sections both are 33/8-in. OD. The collar locator, telemetry, and gamma ray sections are 31/sin. OD, but pressure housings for these sections are available in a 3/8-in. OD size for use in 4-in. ID tubing.
Tool string temperature and pressure ratings are 300 F. and 15,000 psi.
A 5 1/2-in. OD borehole fluid excluder device is available that can improve the capture measurements in large-diameter boreholes. This device is a borated rubber sleeve which fits around the tool and minimizes borehole effects.
DETECTOR CHARACTERISTICS
This particular tool uses a single bismuth germinate (BGO) detector. Because of its high density and high atomic number, BGO provides substantially better gamma ray detection efficiency than the sodium iodide (NaI) detector in previous tools.
With a BGO crystal, a larger fraction of the total detected gamma rays are from the full-energy peak, and less from the lower energies that also include Compton-scattered gammas. The result is an energy spectrum that is both sharper and more statistically precise than older systems.
Studies indicate that compared to NaI systems, this measurement should provide a statistical improvement by a factor of more than two.
In other words, at traditional logging speeds, the tool can achieve more repeatable logs than the other systems, or attain comparable results at substantially higher logging speeds.
BGO does have a deficiency, although that can be easily overcome. BGO detectors perform poorly at high-temperature, and this requires thermal protection of the detector with a Dewar.
The holding time for modem Dewars exceeds 12-15 hr at a maximum temperature of 300' F. This is a minimal constraint, and the tool can easily maintain proper detector temperatures during normal logging operations.
A positive side is that a Dewar leads to a more constant measurement resolution because of the smaller and slower thermal drifts of the detector system. This reduces gain-stabilization problems.
PRIMARY MEASUREMENTS
For each of the inelastic and capture spectra, the data are reduced to windows which are processed to allow calculation of the C/O and Ca/Si ratios from the inelastic spectra, and several elemental indicators from the capture spectra.
In post log processing, calibrated elemental yields calculated by a least-squares fitting technique can be provided. Formation capture cross section measurements are obtained from the decay measured during the background pause.
Several curves on the PNS log monitor log quality. One of these curves is the C/O ratio statistical uncertainty (STUN). The curve is calculated from the logging speed, count rate filter length, and the carbon and oxygen window count rates.
Two other quality indicators, iron edge ratio (FER) and hydrogen peak location (HPL), monitor PNS system downhole gain stabilization.
These two curves are very sensitive to shifts in the gain or zero offset of the system. Monitoring both is critical to assuring good quality logs. Their values should remain essentially constant.
LOG INTERPRETATION
Because RC/0 increases with increases in oil saturation or matrix carbonate content, and RCa/Si increases with increases in carbonate content but not hydrocarbon content, a "quick-look" overlay of these ratios, properly scaled, will indicate oil saturation directly.
A word of caution is that when using the RC/O-vs.-RCa/Si quick-look display, RC/0 is a very poor gas indicator.
On the field quick look, the RC/0 and RCa/Si curves tend to overlay in a gas zone just as in a wet reservoir or a shale zone. Computing-center interpretation software also does not detect the gas.
The only indications of gas on the PNS log are increased capture count rates and a low hydrogen index.
With this in mind, do not be alarmed when, in calculating oil saturations, an apparent oil zone is found with water on top. This, unfortunately, is the gas-cap characteristic of C/O log interpretation.
To calculate oil saturation DELTA C/0 is first determined by:
See Chart
where: is porosity and k is an offset.
DELTA C/O is essentially the difference between the measured RC/0 and the RC/O response when oil saturation is 0% under the same porosity and lithology (RCa/Si) conditions.
The offset k takes into account such factors as borehole size and tool variances. Porosity can be obtained from open hole neutron, density, sonic, or from cased-hole neutron, sonic, or calibrated RIC.
To determine k, first find log values of RC/O, RCa/Si, and f in a 0% oil-saturated zone. Then substitute these values into the equation, set DELTA C/O = 0, and solve the resulting equation for k.
After finding DELTA C/O, the following equation can determine oil saturation (S0).
See Chart
Oil saturation can also be obtained graphically from Fig. 6. For example, if DELTA C/0 = 0.043 in an 8-in. borehole in a 24-p.u. formation, then so = 60%.
STATION MEASUREMENTS
The logs in Fig. 7 are from Alaska's North Slope. This is an older well that has been producing for several years. The PNS log was run to determine the current oil saturations of the various reservoir sands and to evaluate the effectiveness of the new generation PNS logs. Open hole logs confirmed the range of the oil saturation calculations and the mineralogy indicated by the PNS elemental yields.
The borehole size was 8/2 in., and casing OD is 7 in.
Three passes were run at 5 fpm and averaged to produce a final log that would have only minimal statistical variations.
The quick-look overlay shows a large, moderately clean, oil sand from X0780 to X1000 ft, bounded above by a shaly limestone. The iron and sulfur elemental indicators show a pyrite streak at X0790-X0800 ft and suggest dispersed glauconitic clays throughout the bed.
This log was run shut in and the low, steady activation curves on the quality log show that no crossflow was taking place.
As a check on the repeatability of the continuous logs, 5 min station readings were taken at selected depths. The stations are plotted on the averaged continuous log as circles located at X0823, X0906, and X0922 ft. The excellent agreement between the stations and the continuous runs backs up the claim that repeatability of the new generation PNS logs has improved to where time-consuming stations are no longer required.
SHALY SAND
The log example of Fig. 8 is from the Niger Delta of Nigeria. The well was drilled in 1964, and open hole logs consisted of ES (short normal and long lateral resistivities), IES (deep induction and short normal resistivities), and microlog.
Bit size was 91/s in., and casing diameter is 7 in.
A series of cased-hole logs, November 1990, evaluated the condition of the well bore tubulars and the reservoir. Because of the brackish formation waters, a PNS log (rather than a PNC log) was run to determine oil saturation in the various sands. These data were used in a comprehensive reservoir evaluation study.
The PNS logged three passes at 5 fpm that were then averaged. The overlay of the averaged RC/O and RCa/Si ratios in Track 2 serves as a quick-look oil indicator and predicts a good oil zone from X160 to X220 ft, a transition zone from X220 to X270 ft, and water below X270 ft. The overlay also suggests potential production in the smaller sand at X340 ft.
The computing center interpretation, at the right in the figure, generally confirms the conclusions derived from the quick-look display on the field log. It also closely matches the reserves that the operating company estimated from material balance calculations.
COMPLEX LITHOLOGY
In Fig. 9, the log is from the North Slope of Alaska. The hole size was 81/, in. with a 7-in. casing.
This example compares open and cased-hole fluid saturations and illustrates lithology determination from elemental yields analysis. To minimize statistical variations, this log is from a straight average of three individual logging passes, each recorded at 5 fpm.
The log was run a few months after casing was set to evaluate the accuracy of PNS logs. There was also hope that the elemental yields could be used to determine lithology accurately enough to reduce the open hole logging program in high-angle development wells where measurement while-drilling (MWD) would be used.
The quick-look overlay of the inelastic C/O and Ca/Si ratios on the primary log indicate oil or carbon-bearing minerals throughout most of the displayed section.
The capture elemental yields for calcium and silicon (YCa and YSi) indicate sand with a small amount of calcite possible. The iron yield (YFe) shown on the quality log indicates two areas of high iron concentration in the reservoir at X130-Xl55 ft and X220-X230 ft.
The computing center interpretation used open hole density, neutron, and gamma ray logs for effective porosity and shale volume calculations. The PNS elemental yields were used for the lithology analysis and the inelastic C/O ratio was used to compute oil saturation.
The results show the lithology to be generally shaly sand with small amounts of calcite cement. Siderite was determined to be present in the high iron-bearing zones. The open-hole water saturation is displayed along with the cased-hole oil saturation for comparison.
This close agreement is to be expected since the well has not yet been put on production. Core analysis confirms this interpretation by describing the lithology as sandstone with varying amounts of calcite and glauconite cement and large amounts of siderite in the zones at X130-X153 ft and X220-X230 ft.
ACKNOWLEDGMENT
The author wishes to thank the operators of the various wells for their permission to publish the data in the log examples.
BIBLIOGRAPHY
Gartner, M.L., and Jacobson, L.A., "Detector Design and Data Processing Effects on Elemental Yield Variance," SPWLA 13th Annual European Formation Evaluation Symposium, October 1990, Budapest.
Jacobson, L.A., Beals, R., Wyatt, D.F. Jr., and Hrametz, A., "Response Characterization of an Induced Gamma Spectrometry Tool Using a Bismuth Germinate Scintillator," SPWLA 32nd Annual Logging Symposium, June 1991, Midland, Tex.
McWhirter, V.C., "Introduction to Carbon Logging," SPE Rocky Mountain Regional Meeting, May 1976, Casper, Wyo.
Pulsed Spectral Gamma Log, HLS Technical Brochure EL-1069, February 1992.
Scott, H.D., Stoller, C., Roscoe, B.A., Plasek, R.E., and Adolph, R.A., "A New Compensated Through-tubing Carbon/Oxygen Tool for Use in Flowing Wells," SPWLA 32nd Annual Logging Symposium, June 1991, Midland, Tex.
Smith, H.D., Jr., "Finding and Monitoring Hydrocarbons Behind Casing," SPE Distinguished Lecture Series, 1986.
Woodhouse, R., and Kerr, S.A., "The Evaluation of Oil Saturation Through Casing Using Carbon/Oxygen Logs, The Log Analyst, January-February 1992, pp. 1-11.
Wyatt, D.F. Jr., Jacobson, L.A., Durbin, D.P., and Lasseter, E.A., "Logging Experience With a New Induced Gamma Spectrometry Tool," SPWLA 33rd Annual Logging Symposium, June 1992, Oklahoma City.
Copyright 1993 Oil & Gas Journal. All Rights Reserved.
