Response of a Multidetector Pulsed Neutron Porosity Tool

Transcription

1 Response of a Multidetector Pulsed Neutron Porosity Tool H. D. Scott, P. D. Wraight, J. L. Thornton, J-R. Olesen, R. C. Hertzog, D. C. McKeon, T. DasGupta, I. J. Albertin Schlumberger Wireline & Testing Houston, Texas ABSTRACT The development of a neutron porosity tool using a pulsed source of 14-MeV neutrons and multiple detector spacings provides significant improvement in the determination of formation hydrogen index. Optimizing the source-to-detector spacing has substantially reduced the unwanted effects of increased formation atom density exhibited by clay minerals in shaly formations, and has also reduced the lithology effect. The detector system has five neutron detectors: four epithermal and one thermal. By operating the source in a pulsed mode, the detector system can measure both neutron count rates and neutron arrival times. To reduce borehole effects, the detectors are backshielded and the tool is run eccentered in the borehole. The epithermal neutron porosity measurement is corrected in real time for tool standoff from the borehole wall by combining count rate ratios with the epithermal neutron slowing-down time measurement. The use of epithermal neutron detection removes the influence of thermal neutron absorbers commonly encountered in shaly formations. In addition, fluid salinity and temperature effects are significantly reduced. The improved vertical resolution of the measurement plus the lower sensitivity to clay makes it easier to identify and evaluate thin beds. The single thermal neutron detector is used to measure the capture cross section (sigma) of the formation close to the borehole. This extra measurement has good bed resolution and is valuable for formation evaluation and gas indication. The response of the tool to a wide range of formation and borehole parameters has been determined using a combination of laboratory measurements and Monte Carlo modeling. In carbonate formations, the responses to limestone and dolomite are almost identical. The use of an electronically controlled pulsed neutron source eliminates the need for the conventional radioactive AmBe neutron source for this type of measurement, improving radiation safety. This technology has been found particularly useful for gas exploration in shaly formations. INTRODUCTION Neutron porosity logging plays an important role in the evaluation of newly drilled wells when used in combination with resistivity and density logging. For many years, this triple combination has been the industry standard for quantifying new oil and gas reserves. Evaluations have been improved recently as a result of better resistivity tool designs, but the basic design of commercial neutron tools used for openhole logging has remained unchanged for many years. Conventional neutron tools of this type typically have a continuously emitting source of neutrons and use either one or two neutron detectors. The performance of such devices has been summarized by several authors (Alger et al., 1972; Arnold and Smith Jr, 1981; Smith, 1986; Tittman, 1986; Ellis, 1987; Galford et al., 1988; Mickael and Gilchrist, 1993). An improved neutron porosity measurement has been designed using an electronically controlled miniature (minitron) neutron source which reduces radiation hazards. The basic tool design, known as the Accelerator Porosity Sonde (APS), was described in an earlier paper (Flanagan et al., 1991). Since that time, several innovations have been implemented. The APS measurement has the following advantages over conventional compensated neutron porosity logs: The porosity response is affected primarily by the hydrogen index of the formation and is relatively insensitive to changes in formation atom density. The vertical resolution of the measurement is improved. In carbonates, the response to limestone and dolomite is almost identical. Combining appropriately spaced measurements allows gas detection without the use of other logs. Environmental effects are reduced. An epithermal neutron slowing-down time measurement is provided. A thermal neutron capture cross-section measurement (sigma) is also provided. This paper gives a detailed description of the tool, results of Monte Carlo modeling of the tool response to various conditions, results of laboratory measurements, -1-

2 and a log example showing the present tool performance. APS SONDE DESIGN The APS sonde is an important component of the new modular IPL * Integrated Porosity Lithology nuclear measurement tool string. This string consists of a single electronic cartridge to control and process all the nuclear measurements performed by three sondes: the Litho-Density * Sonde (LDS), the APS sonde, and the Hostile Environment Natural Gamma Ray Sonde (HNGS). The arrangement of the tool string is shown in Figure 1. The APS sonde has a diameter of in., a temperature and pressure rating of 175 C and 20 kpsi, and an eccentralizing device. The major components of the APS sonde are shown in Figure 2. There are two fundamental differences between the electronic source of the APS sonde and the chemical source of previous generation dual-detector compensated neutron tools: the neutrons have 3 times the energy (14 MeV instead of an average of 4.5 MeV) eight times as many neutrons are emitted. The lower response to porosity because of the higher energy of the neutrons is more than offset by the ability to pulse the electronic neutron source at different frequencies. The higher source output allows epithermal neutron detection to be used instead of thermal neutron detection without increasing statistical variations. This allows measurements which are free from the effects of unknown thermal neutron absorbers in the formation and borehole regions. Neutron Detectors To improve the formation-to-borehole signal ratio, the helium-3 neutron detectors are eccentered, shielded from the borehole, and focused toward the formation. For the porosity measurement, the detector section uses four epithermal neutron detectors placed at three different spacings from the source. A single thermal neutron detector is also used. Below the source is the near epithermal detector (Fig. 2) which is centered in the tool and has much lower porosity sensitivity than the other detectors. This detector provides normalization of the neutron source output and partial compensation of environmental effects. * Mark of Schlumberger The two array epithermal detectors are backshielded to achieve focusing and to minimize borehole effects. The two detectors used together provide improved count rate statistics and redundancy. Measurements include the average count rates and time distributions of the neutrons detected in each array detector. The ratio of near/array epithermal count rates is the principal porosity measurement provided by the APS sonde. The time distributions are used to measure the epithermal neutron slowing-down time. The single thermal neutron detector provides thermal neutron decay time measurements. It is different from a conventional pulsed neutron capture (PNC) measurement since it detects neutrons rather than gamma rays and is shielded from the borehole. As a result, this shallow device measures the formation capture cross-section of the invaded zone. The detector backshielding allows the range of formation sigma measurements to extend over a wider range than conventional PNC devices. This is a useful shaliness indicator with a vertical resolution better than the gamma ray log. The far detector, which is also eccentered and backshielded, provides the longest spacing measurement. Its source-to-detector spacing is similar to that of a compensated neutron device. In terms of sensitivity to formation atom density and gas effect, porosity computed from the ratio of the near-to-far count rates shows characteristics similar to the CNL * Compensated Neutron Log measurements. Tool Operational Check The APS sonde uses a novel method to check the tool is operational. Since the neutron source is not switched on until the tool is well below the surface, a method is needed to confirm that the tool is working correctly before placement in the borehole. Previous methods required an auxiliary neutron source at the wellsite. However, a major reason for developing the APS sonde was to completely eliminate the need for neutron sources. To solve this dilemma, a special design was developed for the helium-3 neutron detectors. Each detector is augmented with a very weak internal alphaparticle source. The alpha sources have extremely low activity. All radiation is completely contained within the detector since the alpha particles have insignificant penetrating power. These particles create a low background count rate of ionizing pulses in each detector which are counted by the tool electronics. This count rate is compared with that measured during tool calibration to confirm the tool is operating correctly prior to logging the well. This procedure eliminates the need for an auxiliary neutron source and the complete -2-

3 logging operation can be performed with enhanced safety and environmental protection. Neutron Pulsing Scheme The pulsed neutron source allows the measurement of both total count rates and neutron time distributions. To measure the epithermal neutron slowing-down time, short neutron bursts of 10 µsec are followed by measurement periods of 30 µsec. This contrasts with the sigma measurement with neutron bursts of 100 µsec followed by a measurement period of 700 µsec. A fully interleaved pulsing scheme accommodates both measurements simultaneously with a series of 30 slowing-down time cycles followed by one sigma cycle. A neutron burst-off period is incorporated to allow the measurement of the detector background count rates. This permits detector performance to be monitored continuously, even while logging. MONTE CARLO MODEL An MCNP model of the APS sonde has been developed for the determination of tool response algorithms. The MCNP-3A code (Briesmeister, 1986) is a general purpose Monte Carlo code for simulating the transport of neutrons and/or gamma rays in complex threedimensional geometries. The code, developed at Los Alamos National Laboratory in the 1970s, has been used for modeling the response of well-logging tools for over a decade. Objectives were to have a model sufficiently accurate for developing tool response algorithms while maintaining the flexibility to handle a wide range of problems. The use of laboratory measurements and a large MCNP database has proved to be a powerful combination. Even with the Environmental Effects Calibration Facility (EECF) in Houston, only a limited set of variations in wellbore conditions such as lithology, porosity, salinity, gas saturation, standoff, borehole fluid and mudcake may be practically studied. By combining this limited range of laboratory measurements with a larger set of MCNP results, a better understanding of tool response can be obtained. The APS MCNP model was designed to compute the count rates in the helium-3 detectors. Since the absolute source strength is unknown, the model is not typically used to calculate absolute count rates. Instead, count rate ratios are used to normalize out the source strength. The model was designed also to simulate the slowing-down time measurement of epithermal neutrons. The count rate time-profiles in the array detectors were computed with the code and processed to compute the slowing-down time. The results were benchmarked using EECF data for various lithologies, porosities and standoffs. The benchmarked model has been used to predict the slowing-down time response in conditions of heavy mudcake and standoff that could not be duplicated easily in the laboratory. Detector Physics The helium-3 detectors in the APS detect neutrons when a neutron is absorbed by a helium-3 nucleus. The reaction causes a proton to be ejected from the excited nucleus. The recoil nucleus and the proton lose energy in the gas by stripping bound electrons from atoms. The electrons are accelerated to an anode wire by a high electrical potential. Pulses on the anode wire correspond to neutron interactions within the detector gas. To accurately model the active volume of the helium-3 detectors, simulations were made of the electrostatic potential for the detectors. Only reactions in the active volume of the detectors were counted. An additional reaction in a helium-3 detector is elastic scattering. Since the tool electronics eliminates pulses below 100 kev in energy, the code was modified to create a special elastic scattering tally to detect elastic scattering events depositing over 100 kev in the helium gas. Benchmarking The purpose of benchmarking is to prove that the model can calculate the tool response in environments that cannot be easily measured. A comparison of normalized count rates involves determining a normalization constant so that the MCNP count rates (or ratios) match the measured values. In some cases, a normalization function is required to match the MCNP values with the measured values. The best situation is when a single normalization constant can be used for any combination of wellbore conditions such as hole size, lithology, porosity and salinity. Comparisons of MCNP calculated ratios with EECF measured values have shown that single normalizations for the near/array ratio and the near/far ratio are sufficient. These normalization constants were determined for three lithologies (sand, lime and dolomite) and several different porosities: Near/Array : Ratio APS = Ratio MCNP Near/Far : Ratio APS = Ratio MCNP Crossplots of the measured ratios versus the normalized MCNP ratios are shown in Figure 3. EPITHERMAL NEUTRON RATIO One of the most important factors to consider when designing traditional neutron logging tools is the source-to-detector spacing. Since the objective is to measure neutrons that have traveled through the formation rather than the borehole, the emphasis is to use as large a source-to-detector spacing as possible. -3-

4 This is balanced by the need to have good counting rate precision, thin-bed resolution and a neutron source strength within safe limits. The present dual-spaced detector designs meet these requirements and, by using the ratio of detector count rates, partially compensate for borehole environmental effects. The measurement principle of neutron porosity logging is based on the fact that hydrogen is very efficient in the slowing down of fast neutrons. A measurement of the spatial distribution of neutrons resulting from the interaction of high energy neutrons with the formation can be related to its hydrogen content. If the hydrogen, in the form of water or hydrocarbons, is contained within the pore space, the measurement correlates with porosity. If the formation also contains clay minerals containing bound hydrogen, the tool responds to the total hydrogen content. Hydrogen index (HI) is defined as the formation hydrogen content relative to that of water at standard conditions. Thus, at standard conditions, the HI of water is defined as 1.0. The transport of neutrons through the formation is controlled by the various atomic nuclei present and their respective neutron scattering and absorption cross sections (Tittman, 1986; Ellis, 1987). Predictive models of compensated neutron log behavior make use of the concept of neutron slowing-down length to epithermal energy (Allen et al., 1967; Edmundson and Raymer, 1979; Scott et al., 1982). This is a measure of the size of the epithermal neutron cloud surrounding a source emitting neutrons into the formation. Although this length is dominated by the amount of hydrogen present, it is also affected to a lesser extent by all other atoms in the formation. Porosity The lower half of Figure 5 shows the epithermal neutron population plotted against source-to-detector spacing for three porosities with constant matrix density, such as limestone. The three curves show the decrease of the neutron population with distance from the neutron source, and the relative sensitivity to HI, or formation porosity. At a fixed distance far from the source, the detectable neutron population and, therefore, the detector count rate decreases when the formation porosity increases. Moving the detector position closer to the source decreases the sensitivity to formation porosity. Formation Atom Density Before describing the effect of atom density on neutron logging, some new terminology will be defined. The convenient term atom density is used in the following discussion instead of the more correct term for neutron scattering, density of atomic nuclei. For a typical well with a density log run in combination with a neutron log, two measurements are produced: bulk density and apparent neutron porosity. If the apparent neutron porosity is a correct measure of the formation HI, combining the bulk density allows the deduction of an additional property of the formation. This property has been named the dehydrated formation density (g/cm 3 ) and is defined below: ( ρ HI b ) Dehydrated formation density = ( 1 HI) (1) where ρ is formation bulk density and HI is the b formation hydrogen index. This is the formation matrix density that would yield the same bulk density with the formation hydrogen index entirely related to water-filled porosity. Values cover a wide range for minerals typically found in sedimentary formations. For example, an illite clay, with a bulk density of 2.52 g/cm 3 and HI of 0.35 corresponds to a dehydrated formation density of 3.34 g/cm 3. Kaolinite and iron-chlorite, with bulk densities of 2.54 and 3.42 g/cm 3 respectively, correspond to 3.61 and 4.72 g/cm 3 dehydrated formation densities. All clays have dehydrated formation densities well above sand or lime densities (2.65 or 2.71 g/cm 3 ), for which the porosity response of conventional neutron devices is calibrated. It can be shown (Fig. 4) that for alumino-silicate dominated clay minerals there is an approximately linear relationship between dehydrated formation density and dehydrated atom density (atoms/cm 3 ), defined below: Dehydrated atom density = ( ) N t 1.5N H ( 1 HI) (2) where N t is the formation total atom density and N H is the formation hydrogen atom density. The factor 1.5 is used to subtract one oxygen atom for every two hydrogen atoms since the dehydration step removes molecules of H 2 O. From Figure 4, it can be seen that the use of dehydrated formation density is a convenient way to estimate the effect of atom density. Notable exceptions to this approximation are anhydrite and minerals containing significant quantities of heavy elements, such as iron. Returning now to Figure 5, the upper half shows how the neutron population depends on the distance from the neutron source and the value of the atom density when the HI is held constant. At the far position from the source, the detectable neutron population decreases when the atom density increases. This increase in the number of atoms (nuclei) per unit -4-

5 volume causes increased neutron scattering and absorption, limiting the number of neutrons arriving at the far detector. Increasing or decreasing the atom density has an effect similar to increasing or decreasing the formation porosity. Moving the detector closer to the source decreases the sensitivity to atom density. By moving the detector still closer to the source, the backscattering of the neutrons toward the detector becomes the dominant factor influencing the count rate. In this region, increasing the atom density causes the detector count rate to increase since the additional scattering maintains the neutrons closer to the source. At the crossover zone (at an intermediate source-to-detector spacing), the detector has negligible count rate sensitivity to changes in the atom density with HI held constant. In other words, for formations containing most clay minerals, there is no sensitivity to the dehydrated formation density. The lower part of Figure 5 shows the approximate position of the APS near-, medium- and far-spaced detectors. The medium-spaced detector consists of an array of two small detectors labeled 'array'. The nearspaced detector exhibits little sensitivity to porosity and provides the means to normalize the neutron source yield and partially compensate for environmental effects. The optimized spacing of the medium-spaced detector minimizes the effect from dehydrated formation density while exhibiting reasonable sensitivity to formation porosity. The far-spaced detector is highly sensitive to porosity with a sensitivity to atom density similar to that of conventional compensated neutron devices. This includes the excavation effect response to gas since this is equivalent to a low dehydrated formation density effect. Recognition of the importance of atom density on the far-spaced detector measurement and inclusion of a closer-spaced detector that minimizes this effect while retaining HI sensitivity is a step forward in neutron porosity logging tool design. Modeling Results The Monte Carlo code was used to study in detail the response of the APS detectors to changes in porosity and dehydrated formation density. Figure 6 shows the results of some of these computations. The APS sketch on the left indicates relative detector spacings. The three plots on the right indicate near-, array- and fardetector count rate sensitivities to formation hydrogen index and to dehydrated formation density for a water sand (solid line), a gas-bearing sand (dotted line) and a dense shale (dashed line). Note that these plots use reciprocal count rates on the vertical axis. Sensitivities to formation HI and to dehydrated formation density behave as illustrated in Figure 5. The sensitivity to HI increases with detector spacing; smallest in the near detector and greatest in the far detector. The far detector is most sensitive to dehydrated formation density. With HI constant, count rate decreases with dehydrated formation density. The near detector is less sensitive to dehydrated formation density and shows an increase in count rate with an increase in dehydrated formation density. The array detector is insensitive to dehydrated formation density while retaining a reasonable sensitivity to formation hydrogen index. It should be noted that the far detector, which is similar to the far detector of conventional compensated neutron tools, is sensitive to both formation HI and dehydrated formation density. Therefore, using the near/far ratio retains the traditional sensitivity to porosity and also the excavation effect in gas zones. The above modeling results were obtained for clastic formation mixtures of quartz sand and clays. Results for carbonates show a small systematic difference from clastics which appears to be related to the nuclear cross sections for the dominant elements in carbonates, rather than to a difference in atom density. This difference is discussed later as part of the basic tool response. Laboratory Results Measurements in laboratory formations were used to compare the APS near/array and near/far porosity responses with the CNL response. The results are summarized in Table 1. The tool calibrations were confirmed in a 35-p.u. freshwater sandstone formation. An extremely dense shale formation was simulated using aluminum oxide grains with a grain density of 3.97 g/cm 3. This formation had a true porosity of 47 ±2 p.u. and a bulk density of 2.57 g/cm 3 with fresh water in the pore space. A gas-bearing sand was simulated using a new approach employing grains of fused quartz with a grain density of 2.18 g/cm 3. A formation porosity of 20 ±1 p.u. was obtained using particles of different sizes. The advantage of using fused quartz is that it contains exactly the same atoms - silicon and oxygen - as a normal quartz formation but has lower bulk density, similar to a gas sand. When saturated with fresh water, the fused quartz formation had a bulk density of 1.94 g/cm 3, exactly equivalent, in terms of atom density, to a quartz sand formation with a porosity of 34 p.u. with a water saturation of 58%. Results shown in Table 1 use the sandstone porosity scale. In the simulated dense shale, the APS near/array porosity gave a reading of 51 p.u., only slightly higher than the true porosity value of 47 ±2 p.u. despite the very high grain density of 3.97 g/cc. This small -5-

6 deviation is caused by a small atom density effect introduced by the near detector. In contrast, the APS near/far porosity was 85 p.u. and the CNL porosity was 81 p.u. The high APS near/far porosity reading shows a strong sensitivity to the dehydrated formation density effect. This is true for the CNL reading also, but included in the CNL shift is about 4 p.u. induced by the presence of neutron absorbers in the aluminum oxide formation, such as boron. The APS sigma value of 25 c.u. for this formation indicates an equivalent boron content of 84 ppm. When the CNL tool was first introduced, its thermal neutron response did not reflect the true formation HI in shales. The increase in apparent porosity was attributed to the possible presence of thermal neutron absorbers in clays. Log results with the epithermal CNL tool show that only about one third of the apparent porosity increase in shales is due to such absorbers. Sensitivity to the dehydrated formation density is responsible for the remaining effect. In the simulated gas-bearing formation, the near/array response was 19 p.u., very close to the correct porosity (HI) of 20 ±1 p.u. The APS near/far and CNL porosities exhibited the classic excavation effect, both reading the significantly lower value of 14 p.u. These experiments confirm that the near/array porosity measurement is practically independent of dehydrated formation density in alumino-silicate formations, measuring close to the true HI of the formation. RESPONSE TO POROSITY AND SIGMA The basic responses of the APS tool fall into three categories: Integrated count rates from the four epithermal detectors for porosity determination from count rate ratios Neutron counts as a function of time for the two epithermal array detectors to determine slowingdown time, also for porosity Neutron counts as a function of time for the thermal array detector to determine formation capture cross section (sigma) Count Rate Ratios Figure 7 shows the results of near/array count rate ratio measurements made to establish the response of the tool in freshwater limestone formations. These measurements were made with the tool eccentered in freshwater boreholes with diameters in the range to 8.5 in. The data include measurements in the API limestones at the University of Houston (Belknap et al., 1959; API, 1974), the EUROPA formations in Aberdeen (Locke and Butler, 1993) and the EECF facility. For the EUROPA data set, the HI of each formation was used. This was computed from core porosity and core analysis of trace bound hydrogen (Locke and Butler, 1993). EECF and EUROPA measurements in sandstone and dolomite formations show that the limestone and dolomite responses are almost identical, differing by less than 1 p.u., and the difference between sandstone and limestone is significantly less than that for the CNL measurement (Galford et al., 1988). These results have been confirmed with Monte Carlo modeling. Limestone data are shown in Figure 8 for the near/far ratio. This ratio measurement has larger sandstone and dolomite effects, relative to limestone, similar in magnitude to the CNL measurement but different in detail. A comparison of these lithology effects is shown in detail in Figures 9a and 9b. Here, the magnitude of the porosity correction to obtain the true porosity in sandstone or dolomite is shown. Monte Carlo modeling indicates that in terms of total hydrogen content (HI), the near/array ratio response for quartz sand, and quartz sand containing typical clay minerals (shale) fall on the same line. Even though the near/array ratio has a smaller lithology effect than the near/far or CNL measurement, it is still possible to obtain good lithology identification using the neutron-density cross plot, as shown in Figure 10, since the density log provides excellent lithology separation. Slowing-Down Time The epithermal neutron slowing-down time (SDT) is determined from an analysis of the decay rate of the epithermal neutron population (Mills et al., 1988, 1989; Flanagan et al., 1991). Hydrogen is the most important element in this slowing-down process. Figure 11 shows the epithermal neutron time distributions acquired during 5-min periods in laboratory formations of zero-, medium-, and high-porosity freshwater limestone with the tool fully eccentered in an 8-in. freshwater borehole. Also shown is the result for an infinite water tank representing 100 p.u. The total epithermal neutron population decreases with increasing porosity. To facilitate comparison of the decay rates, the measurements shown in Figure 11 were normalized to the count rate immediately after the 10-µsec neutron burst in the zero-porosity formation. Sensitivity of the decay rate to formation porosity is excellent at low to medium porosities. At higher porosities, the sensitivity is reduced and counting rates -6-

7 become lower. The slowing-down time processing at present is a simple exponential fit to the data in the time window 2 to 13 µsec after the neutron burst. The response to porosity (HI) in the API, EUROPA and EECF freshwater limestone formations is shown in Figure 12. Sigma The APS sigma measurement is a direct measurement of the decay rate of the thermal neutron population. This is different from other thermal neutron decay time tools that measure capture gamma rays produced in the formation by capture of thermal neutrons. The APS sigma measurement has a shallower depth of investigation since the neutrons must return to the tool to be detected. Typical thermal neutron decay curves have been presented previously (Flanagan et al., 1991). At present, sigma is computed by fitting the data to a simple single-exponential function over the time period 160 to 700 µsec after the 100 µsec neutron burst. The results obtained in the well-characterized EUROPA formations are shown in Figure 13. Two of these limestone formations are saturated with salt water having a salinity of 200 kppm NaCl, with the same salinity salt water also present in the borehole. Two similar limestone formations have a salinity of 100 kppm. These four formations provide a range of sigma values from 15 to 30 c.u. The measured values in Figure 13 are mostly within 1 c.u. of the formation sigma values computed from core and known fluid salinity data. BED RESOLUTION AND DEPTH OF INVESTIGATION The bed resolution of the neutron measurements has been determined empirically by an analysis of the tool porosity and sigma responses in a specially designed test pit. The formations were made from stacked uniform formation blocks of porosities alternating from nearly 0 p.u. to 18 to 24 p.u. and thicknesses of 1 ft or 3 ft. The high resolution near/array (APLC) and near/far (FPLC) porosities and sigma (SIGF) response were measured along with the LDS density porosity (DPO). The working definition used for thin-bed response was the vertical distance required for the neutron measurement to go from 10% to 90% of the total change from one equilibrium value to the next as the tool moved past a step change of the formation porosity. The results are shown in Table 2. Depending on the actual porosities across the step, the results will vary to some extent. Since the vertical resolution of the near/array neutron and the density logs are now more compatible than previously, and since on the IPL string both measurements are focused in the same direction, formation analysis of thin beds is dramatically improved (Olesen et al., 1994). The APS neutron radial spatial responses and depths of investigation were investigated by MCNP modeling studies (Couet and Watson, 1993). The working definition of depth of investigation (DOI) was defined as the radial distance that contributes the first 90% of the cumulative (or saturated) tool response into the formation. As with the vertical resolution, the DOI is dependent on the formation porosity. The results of the integrated radial porosity sensitivity study for a porosity of 15 p.u. are shown in Table 3. All the neutron measurements have increasing DOI in lower porosities. ENVIRONMENTAL CORRECTIONS IN OPENHOLE The standard reference conditions for the APS laboratory measurements are as follows: 8-in. borehole diameter Freshwater in borehole and formation No standoff or mudcake 75 F temperature Atmospheric pressure Tool eccentered in hole Corrections to the uncorrected log data are required to account for the difference between actual logging conditions and the standard conditions. To the degree that the inputs to these corrections may be poorly known, an important design criterion for any new neutron porosity tool is to minimize the need for such corrections. In part this is accomplished by using the ratio of count rates to derive the porosity since borehole conditions that cause each detector count rate to change by the same factor cancel in the ratio and hence result in no net effect. However, to the extent that one detector count rate changes more than the other, a correction becomes necessary. The environmental corrections for the APS near/array and near/far porosities along with those for the CNL are compared in Table 4 for a 15-p.u. limestone formation with a typical selection of borehole environmental conditions. A mid-porosity example was chosen both as representative of typical logging conditions and because mid-porosities tend to be the hardest cases. Since the APS detects epithermal neutrons, corrections for temperature and salinity effects are reduced significantly compared with CNL corrections. The APS corrections for borehole size are also smaller. The APS mud weight corrections are larger than for CNL corrections in the 8-in. borehole example chosen, but in a 16-in. borehole the corrections for the near/array and near/far porosities reduce to

8 and 1.0 p.u., respectively. The effect for tool standoff is somewhat larger for the APS than for the CNL tool. However, for APS a technique has been developed that both measures the effective tool standoff during logging and makes the necessary corrections at each depth. Thus, the standoff-corrected APS logs are more accurate than CNL logs since for the CNL logs the engineer can only estimate the average tool standoff over the logging interval. To determine the APS environmental corrections the tool response has been measured in roughly 1000 different combinations of lithology, porosity, mud weight, borehole size, standoff, and formation and borehole salinity. These measurements and computations of the pressure and temperature effect were used to derive the individual environmental corrections as well as their principle interdependencies, such as the borehole size and standoff effect dependence on mud weight and salinity. All these corrections are available in the standard wellsite product. Each of these effects is discussed below. Mud Weight Increasing the mud weight with additives such as barite decreases the water volume fraction of the mud. Therefore, the mud hydrogen density (HI) is decreased. This allows more neutron transmission through the borehole fluid, resulting in an increase in count rate at the detectors. When changing from fresh water to 12 lbm/gal barite mud, the increase is less than 7% for the APS detectors in an 8-in. borehole with no tool standoff. Each detector behaves slightly differently depending on the amount of backshielding from the borehole fluid. Using detector count rate ratios to compute apparent porosity reduces the overall effect of mud weight. Measurements have been made in laboratory formations using 12- and 16-lbm/gal freshwater barite muds and in a 15 lbm/gal oil-base mud. The HI of these muds is 0.87, 0.73 and 0.69, respectively. The corrections required for both the near/array and near/far ratio porosities in an 8-in. borehole are shown in Figure 14 for the freshwater barite mud case. The results for oil-base muds are similar, although reduced somewhat from those for a freshwater barite mud with the same HI. The sign of the APS mud weight correction is opposite that for the CNL correction due to the backshielding of APS array and far detectors. Since the APS near detector is centered and only partially backshielded, its count rate increases most as the mud weight is increased. This results in a net positive mud weight effect. For CNL measurements, the far detector count rate increases most, resulting in a mud weight correction of the opposite sign. Borehole Size The borehole size effect has been determined with boreholes varying from 6 to 16 in. diameter and with borehole fluids of water and each of the three muds described above. The effect for the near/array porosity in water and in a 16 lbm/gal freshwater barite mud are shown in Figure 15. Figure 16 shows the same effect for the near/far porosity. Since the array and far detectors are backshielded and focused toward the formation, the borehole size effect is controlled by the crescent of borehole fluid between the tool face and the borehole wall. Therefore, bit size rather than the caliper log is used to characterize the curvature of the borehole wall opposite the tool and to compute the borehole corrections. As expected, the uncorrected porosity increases with borehole size as the crescent gets larger and the tool sees more borehole fluid. As can be seen from the figures, the borehole size effect decreases with mud weight as heavier muds contain less hydrogen. Standoff The tool standoff effect for the APS is larger than that for the CNL tool since the APS is a focused device. The unfocused CNL tool, even though it is usually run eccentered, does receive some signal from the formation behind the tool when the tool stands off from the borehole wall. However, an estimate of the actual tool standoff during logging is not easy to determine. A series of APS measurements have been made with standoffs from 0 to 2 in. in laboratory boreholes ranging from 6 to 16 in. filled with water or any of the three muds described above. Since the effect is larger than that for CNL measurements, an independent measurement of the effective tool standoff has been developed using the epithermal neutron slowing-down time measurement. The response of the slowing-down time measurement to formation porosity was shown previously in Figure 11. Since this measurement has a shallow depth of investigation, it is also sensitive to tool standoff. Figure 17 illustrates the effect of standoff on the neutron time distributions with the tool in a 0- p.u. limestone formation and an 8-in. borehole as the standoff is progressively increased. Each time distribution consists of a slow formation decay component, more visible at later decay times, and an early borehole component whose amplitude increases with increasing standoff. The single exponential fit to the data in the time window shown, discussed previously, provides a good formation porosity measurement when there is no tool standoff. With tool standoff, the result is a good standoff indicator. For this application the slowing-down time apparent porosity has been named standoff porosity. -8-

9 In perfect borehole conditions, there is very little difference between the porosity values derived from slowing-down time and the near/array ratio. Both measurements respond primarily to the formation HI, and both are insensitive to thermal neutron absorbers and the dehydrated formation density. They differ slightly in the sensitivity to lithology but mostly in the sensitivity to tool standoff. Figure 18 is a crossplot of the two porosities, with the uncorrected near/array ratio porosity on the horizontal axis and the slowing-down time apparent porosity on the vertical axis. Data shown are for three laboratory formations of 0, 17 and 44 p.u. Without tool standoff, both porosity measurements agree. With increased tool standoff, both apparent porosities increase but the increase is several times faster for the standoff porosity than for the ratio porosity. Crossplotting these two independent measurements provides formation porosity and tool standoff. To accomplish the above standoff correction procedure during logging, an algorithm has been developed to correct for standoff using the difference between the uncorrected near/array porosity and the slowing-down time porosity, the bit size and the mud weight. The effect in water was shown in Figure 18. As the mud weight increases, both the near/array and the slowing-down time porosities increase less with standoff since the mud HI is less and the overall effect is as shown in Figure 19 for a 16-lbm/gal mud. A similar standoff correction is made for the near/far ratio porosity using the standoff information derived from the array detector. Salinity With previous generation dual-spaced thermal neutron porosity tools, corrections for salinity are a complex combination of hydrogen displacement by NaCl and thermal neutron absorption by the chlorine. For the epithermal APS measurement, the correction is simplified to that for hydrogen displacement only. The correction for saltwater in the borehole is computed in the same fashion as the mud weight correction for a light mud. The correction for formation saltwater is a simple HI correction computed knowing the salinity, temperature, pressure and water saturation. A series of APS measurements for a range of saltwater salinities in the borehole and formation has been acquired to characterize the effect on apparent neutron porosity and sigma. The results for formation saltwater in Figure 20 clearly show the effect of decreasing hydrogen index with increasing water salinity. Relative to fresh water with a HI of 1.0, saltwater with a salinity of 200-kppm NaCl has a hydrogen index of Temperature and Pressure The choice of epithermal neutron detection for the APS simplifies the needed corrections for temperature and pressure to a single correction for borehole HI. A major benefit is the much smaller temperature correction, about a factor of seven less than for thermal neutron systems. Dual-spaced thermal neutron porosity tools have a very large temperature correction resulting mainly from the choice of using thermal neutron detection. As shown in Table 4, corrections of the order of 8 p.u. are often required (Galford et al, 1988; Mickael and Gilchrist, 1993). The offsetting correction for increased pressure with depth is relatively small. Mudcake The effect of mudcake thickness is the combined effect of reduced borehole size and tool standoff. The changes to the individual detector count rates and the slowingdown time caused by the presence of barite mudcake have been computed using the MCNP model. For a typical 1-in. thick mudcake with a HI of 0.42, the tool behaves in low- and medium- porosity formations as if it stood off from the formation at a distance of about 0.35 in., approximately the product of true thickness and HI. Thus, the standoff correction processing, described previously, is also effective for correcting for the presence of mudcake. RESPONSE TO MINERALS As shown previously from both modeling and experimental data, the APS near/array ratio measurement provides a result which is very close to the true HI for typical reservoir formations. To develop a formation evaluation procedure for complex lithologies, it is necessary to know the specific response to individual hydrogenous minerals in the formation. The Monte Carlo model has been used for this purpose. Minerals Containing Hydrogen Since minerals containing hydrogen, such as clays, commonly exist as a component of the formation mixture, modeling was done for a volumetric mixture of 50% of the mineral with 50% of quartz sand having a porosity of 30 p.u. The results obtained for several clay minerals are shown in Table 5, together with results for muscovite and iron chlorite. For this study, clays containing clay bound and interlayer water were used to simulate typical formation conditions. Similar results using dry clay formation mixtures have been reported separately (La Vigne et al., 1994). Listed in the table for each mineral mixture are the HI, bulk density and dehydrated formation density from Eq. (1). It can be seen that for all these minerals, the dehydrated formation density is substantially greater than that for quartz sand (2.65 g/cm 3 ). The modeling -9-

10 results are the apparent neutron porosity on the sandstone scale for near/array ratio (APSC) and near/far ratio (FPSC). Results show that for the five clay minerals listed, APSC is very close to the true HI of the mixture. This is also true for muscovite. The agreement for iron chlorite is not as good but is acceptable considering the very high dehydrated formation density and the fact that in nature a 50% volume of iron chlorite is unlikely. Typical formation volumes of iron chlorite are close to 5%. These results indicate that quantitative evaluation of the APS near/array apparent neutron porosity log (APSC) should be possible using the mineral HI for the interpretation. The results for the near/far ratio porosity show significantly increased values caused by the high values of dehydrated formation density. Halite The APS near/array ratio response has been optimized for the typical formation conditions found in oil and gas reservoirs: sandstones, limestones and dolomites with varying amounts of clays and some fraction of heavy minerals. One particular formation type that does not fit into this framework is halite (NaCl). Halite often exists as a bed of 100% pure mineral with a density of g/cm 3. Modeling results for the apparent neutron porosity on a limestone scale (APLC) gave a value of 21 ±2 p.u., in agreement with well log values of 21 to 24 p.u.. The correctness of the model was confirmed, also, using a laboratory dry halite formation. The reason for the high apparent porosity in halite is the very low atom density and the large neutron slowingdown length (43.5 cm.) that exceeds the source to array detector spacing. Modeling of the near/far ratio apparent porosity gave a value of 0.5 ±0.1 p.u., close to the correct HI value of zero. This was in good agreement with a value of 0.3 p.u. measured in the laboratory formation and well log values close to zero. Anhydrite Modeling results for anhydrite with a bulk density of 2.96 g/cm 3 gave a value of 1.8 ±0.5 p.u. for the near/array ratio porosity (APLC). Field tests indicate an average value of 1.5 p.u. For the near/far ratio, modeling gave a value of 0 ±0.2 p.u., also in agreement with field tests. EXAMPLE A typical APS log run in the Amoco Catoosa field, Rogers County, Oklahoma, is shown in Figure 21. The section shown is a series of shales and shaly sands over the bottom interval leading to an almost clean sand at the top of the sequence. The thin zone from 617 ft to 609 ft is an isolated limestone bed indicated by the high value of PEF and the lowest GR and sigma readings in this section. In this relatively shallow well, the APS was run at a speed of 800 ft/hr to accommodate other tools on the string. The standard logging speed is 1800 ft/hr, and 900 ft/hr if the high-resolution sampling mode is required. Shown with the APS log are companion logs from the IPL tool string: the LDS density and photoelectric factor logs and the HNGS uranium-free GR. Also shown is the CNL log from a separate pass run at 900 ft/hr. The APS logs shown are the following: APS epithermal array ratio porosity corrected for borehole conditions and tool standoff presented on a sandstone scale (APSC) APS epithermal far ratio porosity, also on a sandstone scale (FPSC) Sigma formation (SIGF) APS computed standoff (STOF) The primary porosity measurement is APSC. SIGF usually correlates with GR as a shaliness indicator but has added value in terms of thinner bed resolution and lithology evaluation. The figure shows two interesting shale sections from 730 to 685 ft and 668 to 655 ft. Both sections have a relatively constant bulk density and PEF but show a trend of decreasing apparent neutron porosity (HI) with decreasing depth. This trend is mirrored in the GR log which also decreases with decreasing depth. The measurements are consistent with a gradual decrease in the clay content across the shale section with decreasing depth. The FPSC curve for this well reads 4 to 8 p.u. higher than APSC in the shales, as expected due to the increased dehydrated formation density. Comparison of the two curves across the above two shale sections shows a gradual reduction in separation consistent with a reduction in clay content with decreasing depth. In the sand from 600 to 550 ft FPSC exceeds APSC by 3 p.u., indicating that the sand has some clay present. This is confirmed by the value of PEF, which reads 2.2 units. The CNL apparent thermal neutron porosity, also presented on a sandstone scale, is significantly higher than the epithermal array porosity (APSC) across the whole section for two reasons: increased dehydrated formation density effect and thermal neutron absorbers. Increased dehydrated formation density effect is similar to that for FPSC. The thermal neutron absorber effect can be deduced from the APS formation sigma measurement, SIGF. Values of SIGF in the shale -10-

11 sections range from 26 to 40 c.u. Published correction charts for sandstones (Ellis et al., 1987) indicate that in this sigma range the necessary correction is about -6 p.u., consistent with the separation between CNL neutron and FPSC. The SIGF sigma curve is very similar in overall structure to the uranium-free GR curve in this well. The improved bed resolution of this curve over GR is visible at bed boundaries. The computed tool standoff (STOF) reads close to zero over the section indicating good eccentralization of the tool string in a goodquality borehole. Other Examples The above example illustrates the use of the APS for improved formation evaluation in shaly sand reservoirs. Several other examples, including improved evaluation for gas, are available and the subject of another publication (Olesen et al., 1994). CONCLUSIONS The development of a new neutron porosity tool using a pulsed source of 14-MeV neutrons and multiple neutron detector spacings provides a significant improvement to formation HI measurements. By optimizing the sourceto-detector spacing, the unwanted effects of formation atom density have been substantially reduced, particularly for shaly formations. The use of epithermal neutron detection removes the influence of thermal neutron absorbers commonly encountered in shaly formations and in formations with salty brines. The dynamic standoff measurement and porosity correction obtained from the slowing-down time data improve the neutron porosity behavior in moderately rugose boreholes. The improved vertical resolution of the APS measurement plus the lower sensitivity to clay makes it easier to identify and evaluate thin beds. The inclusion of a thermal neutron detector to measure formation capture cross section (sigma) has provided benefits of improved bed resolution and an independent indicator of gas in the region close to the borehole. Laboratory measurements and Monte Carlo modeling show that the APS carbonate response is only slightly influenced by the degree of dolomitization in the formation. In carbonates with unknown dolomite content the APS provides a more accurate porosity determination. Finally, the use of an electronically controlled pulsed neutron source eliminates the need for the conventional radioactive AmBe neutron source for this type of measurement, improving radiation safety. ACKNOWLEDGMENTS We thank Amoco Production Company for permission to publish the example from the Catoosa field. Thanks are due to John Locke and Robert Jorro, AEA Technology UK, for assistance in gathering data at the EUROPA facility. We also acknowledge many helpful discussions with our colleagues P. Albats, B. Couet, D. Ellis, C. Flaum, T. Loomis, R. Namjoshi and K. Stephenson. REFERENCES Alger, R.P., Locke, S., Nagel, W.A., and Sherman, H., 1972, The dual spacing neutron log-cnl, Journal of Petroleum Technology, v. 24, p Allen, L.S., Tittle, C.W., Mills, W.R., and Caldwell, R.L., 1967, Dual-spaced neutron logging for porosity, Geophysics, v. 32, p Arnold, D.M., and Smith Jr, H.D., 1981, Experimental determination of environmental corrections for a dual-spaced neutron porosity log, paper VV, in 22nd Annual Logging Symposium Transactions of the Society of Professional Well Log Analysts. API RP 33 Third Edition, 1974, Recommended practice for standard calibration and format for nuclear logs, American Petroleum Institute, Dallas, Texas. Belknap, W.B., Dewan, J.T., Kirkpatrick, C.V., Mott, W.E., Pearson, A.J., and Rabson, W.R., 1959, API calibration facility for nuclear logs, Drilling and Production Practices, API. Briesmeister, J.F. (ed.), MCNP - A General Monte Carlo Code for Neutron and Photon Transport,Version 3A, Los Alamos National Laboratory report LA-7396-M Rev. 2, Couet, B., and Watson, C., 1993, Applications of Monte Carlo differential neutron sensitivity calculations to geophysical measurements, Nuclear Geophysics, v. 7, no. 2, Edmundson, H., and Raymer, L.L., 1979, Radioactive logging parameters for common minerals, paper O, in 20th Annual Logging Symposium Transactions of the Society of Professional Well Log Analysts. -11-