A New Nuclear Logging Method to Locate Proppant Placement in Induced Fractures

Transcription

1 A New Nuclear Logging Method to Locate Proppant Placement in Induced Fractures Harry D. Smith Jr. Consultant Robert Duenckel, Xiaogang Han CARBO Ceramics, Inc Copyright 2013, held jointly by the Society of Petrophysicists and Well Log Analysts (SPWLA) and the submitting authors. This paper was prepared for presentation at the SPWLA 54th Annual Logging Symposium held in New Orleans, Louisiana June 22-26, ABSTRACT Traditional proppant placement evaluation in hydraulically induced fractures utilizes detection of radioactive (R/A) tracers such as iridium 192, scandium 46 and antimony 124, which are manufactured in nuclear reactors, and then shipped to the wellsite and pumped downhole with the frac slurry. Although this technique has proven useful, it involves environmental, safety, and regulatory concerns/issues. Recently a new technology has become available that offers a viable alternative to radioactive tracers. The new technology utilizes a nonradioactive ceramic proppant that contains a high thermal neutron capture compound (HTNCC). This high thermal neutron capture compound is inseparably incorporated into each ceramic proppant grain during manufacturing in sufficiently low concentration so as not to affect proppant properties. The non-radioactive tracer proppant (NRT) is detected using standard pulsed neutron capture tools (PNC) or compensated neutron tools (CNT), with detection based on the high thermal neutron absorptive properties of the tagged proppant relative to other downhole constituents. Monte Carlo modeling data and several field examples presented in this paper demonstrate the viability of both the PNC and CNT proppant detection technologies. INTRODUCTION Hydraulic fracturing is a commonly used technique to economically produce hydrocarbons from subsurface formations, especially formations with low permeability. In hydraulic fracturing, proppant laden fluid is pumped downhole under high pressure, causing the formations to fracture. When pumping ceases, the fractures close on the proppant leaving high permeability conduits that promote the flow of hydrocarbons into the wellbore. The fracture closure stress often exceeds 5,000 to 10,000 psi, which must be borne by the proppant. This closure stress, coupled with the reservoir conditions and expected flow rates, will dictate the optimal choice of proppant (uncoated sand, resin coated sand or ceramic). The optimal proppant will be one that maximizes the economic return of the well under realistic conditions [Palisch 2007]. There is often a desire to confirm which intervals have been hydraulically fractured. It is possible that there are zones within the desired fracture interval(s) that were ineffectively fractured, either due to an unknown stress regime along the wellbore, anomalies within the formation or problems within the borehole, such as ineffective or blocked perforations. It is also desirable to know whether the fractures extend vertically across the entire desired fracture interval(s), and also to know whether or not any fracture(s) may have extended vertically outside of the desired interval. In the latter case, if the fracture has extended into an adjacent waterbearing zone, the resulting water production would be highly undesirable. In all of these situations, knowledge of the location of both the fractured and unfractured zones is very useful for planning remedial operations in the subject well and/or in utilizing the information gained to improve treatment designs on future wells. Fracture height is typically used by fracturing engineers to calibrate fracture design models. Having an accurate frac height measurement reduces the uncertainty and nonuniqueness of fracture pressure matching, thereby better determining placed frac length and width, stress profile across the target zone and its boundaries, and fracture containment. Knowing the actual frac height at downhole conditions is useful in developing models that can be calibrated with more certainty, so that future designs can be improved to provide the optimal drainage and recovery from the reservoir. Conventional technologies for induced fracture detection in the near wellbore region generally involve either acoustic or nuclear logging technologies, or temperature logs. Technologies for estimation of the induced fracture location and dimensions in the region extending some distance from the wellbore utilize tiltmeter or microseismic measurements. By far, the most common of the nuclear methods employs the use of radioactive (R/A) materials with half

2 lives on the order of days to months. These radioactive materials, generally gamma ray emitters, are added to the proppant laden slurry as it is being pumped downhole. It is typical that only small volumes (less than one pound) of highly radioactive particles are added to the slurry which may contain as much as 100,000 pounds of standard undetectable proppant. A gamma ray detector, usually with spectroscopic capabilities, is then used to log across the interval above and below, as well as within, the desired fracture interval after the frac job. The gamma rays from the multiple radioisotopes placed downhole are detected, recorded, and analyzed either in real time during a wireline log run [Gadeken 1986], or recorded in memory and processed later when the sensor is retrieved to the surface [Bandy 1989]. These techniques are often effective and have the ability to spectroscopically resolve signals from multiple tracers, and also distinguish the radioactive tracer originating near the logging tool from the tracer that is placed into the fracture. There are two primary advantages of the new method compared to this prior method employing radioactive tracers. First, the HTNCC tag material is manufactured into every proppant grain. This ensures that any place proppant is present, the tag material will also be present, and no tag material will be present where proppant is not present. Traditional R/A taggants are added to the proppant slurry flowstream in very low concentrations during pumping, thereby providing the possibility for potential non-uniform R/A distribution in the slurry. The HTNCC taggant is incorporated in very small concentrations throughout each proppant pellet, and the presence of a few such pellets will not provide a measureable log response. This avoids the false positives that can be observed with traditional R/A technology when the presence of merely a few highly radioactive particles near the wellbore can incorrectly suggest the presence of a significant fracture. Second, and more important, this new NRT methodology utilizes only inert materials in the frac slurry, thereby eliminating the hazards and regulatory issues associated with handling, transporting, pumping and flowing back radioactive materials, hence providing a much more environmentally and logistically friendly option relative to traditional R/A tracer methods. General Discussion of the New Methods This paper presents both modeling results and illustrative field examples demonstrating the viability and effectiveness of the new proppant detection methodology. In the new technology, a high thermal neutron capture compound (HTNCC) is incorporated throughout all proppant grains in the manufacturing process at low concentrations. This low concentration produces no measurable deleterious effects on any proppant physical property, including compressive strength and conductivity. Three different methods for detecting the tagged proppant have been developed and are discussed briefly below. In most cases, not all of these interpretation methods will be needed to determine proppant location, depending on the available logs and whether or not there have been changes in borehole conditions and/or formation hydrogen index values between the before-frac and after-frac logs. However, in the case studies contained in this paper, multiple techniques have been utilized to demonstrate the consistency and robustness of the interpretation methods. More detailed discussions, including Monte-Carlo modeling results, can be found in prior publications [Smith 2009, Duenckel 2011, and Grae 2012]. Overview of the New Methods using Pulsed Neutron Capture Logging Tools In the first two methods, pulsed neutron capture (PNC) systems are utilized to detect the proppant in the fractures. In the first PNC method, the tool is logged both before and after the frac job across intervals of the wellbore, including the zones to be fractured and also zones outside the anticipated fracture interval(s). PNC decay curve count rate data detected in capture gamma ray (or possibly thermal neutron) sensors are recorded both before and after the fracturing operation. Observed PNC after-frac parameters are then compared to corresponding parameters recorded before the well was fractured. The formation and borehole thermal neutron absorption/capture cross sections are calculated from the PNC decay curves [Schultz 1983]. Increases in the formation and/or borehole capture cross sections computed from the after-frac log data relative to the before-frac log data, as well as decreases between the logs in the observed count rates in time gates dominated by the formation and/or borehole decay components, are used to identify the presence of the tagged proppant. Proppant is detected both in the induced fractures and/or in the borehole region adjacent to the fractured zones, with qualitative discrimination indicated from the different radial regions and depths of investigation sensed by the different PNC measurements. To date we have modeled Gd 2 O 3, Sm 2 O 3, and B 4 C as the HTNCC material. In all the modeling described in this paper, Gd 2 O 3 was used as the HTNCC. In the second PNC logging method, which can be utilized in combination with the first PNC method, the observed capture gamma ray energy spectra as a function of well depth are deconvolved to determine the computed yields of the downhole elements present in the formation and - 2 -

3 borehole region, including the yield of the nonradioactive taggant in the proppant. The taggant yields determined from the after-frac spectra are then compared to the corresponding yields from the before frac spectra, with increases in the taggant yields on the after frac log indicating the presence of tagged proppant. This second method also provides for the possibility of spectrally resolving the signals from two (or more) different HTNCC materials, for use in applications where it is desirable to employ more than one tag material. If the HTNCC absorber used in the proppant is not present, or is only minimally present, in the other downhole materials, it may also be possible to employ this PNC spectral method to locate the tagged proppant without requiring a before frac log. Overview of the New Method using Neutron or Compensated Neutron Logging Tools In the third method, a conventional neutron tool, preferably a compensated neutron tool (CNT), with a continuous neutron source and one or more thermal neutron detectors (or with capture gamma ray sensing detectors) is utilized. As with the PNC methods, the CNT is logged across an interval of the wellbore, including the zones to be fractured, before the frac job. The resulting near and far detector count rates and the N/F count rate ratios are recorded. These observed count rates are then compared to corresponding values recorded in a corresponding log run after the well has been fractured, preferably (although not required) with the same or a similar logging tool under the same borehole conditions as the before-frac log. Proppant is indicated in zones where the after-frac count rate is lower than the beforefrac count rate due to the HTNCC material in the proppant. A variant of the CNT method has been developed which does not require comparison of before-frac and after-frac logs. In this variant, compensated neutron logs are required only after the frac job, and the N/F ratio and the individual detector count rates are utilized in the log interpretation process. This method also has particular utility when the formation gas saturation changes (or there are other changes in the formation hydrogen index, HI) in zones between the before-frac and after-frac logs, since gas saturation changes will result in CNT count rate changes unrelated to the presence of the proppant. This can be a significant issue in some wells, since frac processes are frequently conducted in gas bearing intervals. Further discussion of this variant follows in the section below where the CNT modeling results are presented. Comparison of before-frac and after-frac logs in both the PNC and CNT methods above is straightforward and clearly indicates the presence of HTNCC tagged proppant when other parameters are relatively constant. If the borehole environment (such as the fluid inside the well casing) changes between the log runs, or if different tools are utilized for the two log runs, or if the neutron outputs of the sources used in the before-frac and afterfrac logs are different, log responses may need to be normalized utilizing logged intervals or zones known to be outside of the fractured interval(s). It is also possible in some situations to eliminate the before-frac log entirely if a PNC or CNT log had previously been run in the well [Duenckel 2011]. Monte Carlo Modeling Results for PNC Logging Tools Our initial work in the development and testing of the concepts described above was to conduct extensive Monte Carlo N-Particle Transfer Code Version 5 (MCNP-5) modeling of both PNC and CNT tool responses in realistic downhole formations / borehole environments containing induced fractures. The tools modeled were 1-11/16 PNC tools with gamma ray detectors (thermal neutron detector tools were also modeled). Also both 3-3/8 and 1-11/16 compensated neutron tools with He3 thermal neutron detectors were modeled. These tools were modeled with water saturated limestone and sandstone formations with varied porosities (ø = 3-42%), which contained vertical/axial biwing fractures of varying widths, and with various borehole conditions. A typical borehole/formation environment was modeled and used in obtaining most of the MCNP data in this paper: a 28.2% water saturated sandstone formation with an 8 diameter wellbore containing a 5.5 OD casing centered in the borehole, and neat-cement filling the annular space between the casing and borehole wall. For all tools, data were collected with the fracture plane bisecting the borehole and aligned with an eccentered tool, as seen in Figures 1a and 1b. Data were also collected with the 1-11/16 PNC and CNT tools with the tools repositioned 90 o around the borehole from the fracture plane, and/or with other changes in borehole/formation parameters (such as borehole fluid salinity, formation water salinity, formation gas saturation, etc)

4 fm = cu bh = cu fm = cu bh = cu Figure 1a. An X-Z view of MCNP modeling of downhole geometry of PNC tool Figure 1b. An X-Y view of MCNP modeling of downhole geometry of PNC tool Numerous combinations of HTNCC proppant concentrations and fracture widths were modeled to quantify the percentage changes observed in formation and borehole capture cross-sections (Ʃ fm and Ʃ bh ), and detector count rates (in various time gates relative to the neutron bursts) in near and far spaced PNC gamma ray detectors in a 1-11/16 tool. Data were also collected for PNC thermal neutron detectors, but are not discussed in this paper [see Smith 2009]. The near and far spacings modeled are typical of those in commercial multidetector PNC tools. Typical modeled before frac (blue) and after frac (red) decay curves and the computed formation and borehole decay components and sigma values are illustrated in Figure 2. The borehole and formation time gates ( μsec and μsec after the neutron burst, respectively) illustrated in the figure are representative of typical gates used to obtain the corresponding borehole and formation count rates from the MCNP decay curves. The fracture width modeled was 1.0 cm. It can be seen that the presence of HTNCC in the fracture results in increases in Ʃ fm of ~2-3cu; however since no taggant was placed in the borehole region in this illustration, there was no significant change in Ʃ bh. This type of PNC data was computed for numerous formation/borehole scenarios, and a small sample of the resulting data are presented in Figures 3, 4a, 4b, and Table 1 below. For formation/borehole conditions typical of those described above, Figure 3 illustrates the effect of changing the concentration of the HTNCC tag material in the frac slurry in a 1.0 cm fracture on the percentage change in the observed PNC near detector formation capture cross-section (Ʃfm ) relative to that with no fracture present. In this data the hydrogen index (HI) and Ʃ fm of the untagged frac slurry and the formation were modeled to be approximately the same, so all Ʃ fm changes will be due to the HTNCC tag material in the proppant, and not due any non-tag related changes with the fracture Figure 2. MCNP simulated gamma ray decay curve. Two exponential fitting is applied to calculate the formation cross section and borehole cross section. The blue data relate to the gamma ray decay curve with no fracture, the red data relate to the gamma ray decay curve with a 1.0 cm bi-wing fracture in the formation. present. From Figure 3 it can be seen that Ʃ fm increases with increasing HTNCC concentration, asymptotically approaching a limit of about 11-12% at HTNCC concentrations greater than ~ %. Above 0.5%, virtually all thermal neutrons passing through the fracture will be absorbed, and hence further increasing the HTNCC concentration is of limited value. Comparable changes in Ʃ fm were observed in long spaced PNC detector data. Figure 4a presents corresponding percentage increases in PNC near detector Ʃ fm values as a function of changing widths of the fracture in the formation. In Figure 4a (and also in Figure 4b, discussed below) the HTNCC concentration was held constant at 0.42%. The asymptotic nature of the effect of changing fracture width can be seen. Little additional increase in Ʃ fm (above ~11-12%) is observed when fractures width is ~1.0 cm (for this HTNCC concentration). Coupled with the HTNCC concentration changes observed in Figure 3, these data indicate that as long as the HTNCC material is limited to the fracture itself, one should not expect changes in Ʃ fm to exceed about 13%, regardless of fracture width and/or HTNCC tag concentration. However, since the statistical uncertainty of major service company PNC tool Ʃ fm log values is almost an order of magnitude less than 13%, it is clear that fractures tagged with HTNCC should be easily observable

5 Percentage increase of PNC formation sigma _fm (%) SPWLA 54 th Annual Logging Symposium, June 22-26, % 10% 8% 6% 4% 2% 0% HTNCC Concentration (%) Figure 3. MCNP simulated PNC tool formation sigma in near detector as a function of the concentration of the HTNCC (frac width = 1.0 cm) With HTNCC tagged proppant present in the fracture, even larger percentage changes (decreases) will be observed in the capture gamma ray count rate in a PNC near detector formation time gate (positioned μsec after the neutron source pulses), as seen in Figure 4b. In this figure, with HTNCC concentration in the proppant held at 0.42%, the width of the fracture was varied as in Figure 4a, and the computed capture gamma ray count rate was observed as a function of fracture width. Up to about 35% decreases in count rate with increasing fracture width are observed in Figure 4b, and, as seen above, the changes reach an asymptotic limit. Note that for the same fracture width conditions, the percentage changes in formation gate count rates are roughly three times as large as the corresponding changes in Ʃ fm in Figure 4a. The effects of HTNCC material in proppant were studied for a large number of other changes in downhole parameters for both PNC and CNT tools, and more detail can be found in prior publications [Smith 2009]. Data representative of the additional PNC modeling of a formation containing a 0.5 cm wide fracture can be seen in Table 1. In this table Ʃ fm, Ʃ bh, near detector formation gate count rate and near detector borehole gate count rate are shown: (1) for a different tool position inside the casing relative to the fracture plane, (2) for a different borehole fluid salinity, (3) with tagged proppant slurry replacing cement in 20% of the borehole annular (cement) region outside the casing, and (4) with tagged proppant in both the borehole region and in the formation fracture. From Table 1, several observations can be made. As expected when using the PNC two-exponential decay curve processing method, Ʃ fm is essentially unaffected by a change in the borehole fluid. Ʃ fm has roughly the same sensitivity to the tagged proppant in the formation versus the cement annulus outside the casing. Also as expected, Ʃ bh is affected very little by tagged proppant in the formation, but is sensitive to tagged proppant in the borehole region (and to borehole fluid salinity). The formation gate count rate is highly sensitive to tagged proppant in both regions, with a greater sensitivity to proppant in the borehole than in the formation fracture. The borehole gate count rate is also affected by tagged proppant in both regions, but has a very high percentage signal coming from the borehole relative to the formation. The combinations of these various signals can be very useful in locating the radial location of the tagged proppant. For example, if there are increases in both Ʃ fm and Ʃ bh, and only fracture-related changes have taken place between the before and after frac logs, it is likely that the tagged proppant is present in both the fracture and the near borehole region adjacent to the fracture. The relative magnitudes of the borehole and formation gate count rates can also be highly useful in locating HTNCC tagged proppant and estimating radial proppant distribution. However, if there have been changes in borehole or formation parameters between the logs unrelated to the presence of the tagged proppant, in some cases it may be necessary to normalize log responses or utilize multiple signals in the process of comparing the log data. Perhaps the most significant non-proppant related formation change that can take place between the before and after frac logs is a change in formation gas saturation. In several of the 40+ wells logged to date, zones with increased gas saturation on the after frac log were observed. PNC modeling data related to this situation is presented in Table 2. The same formation, borehole, and fracture conditions were modeled in obtaining the data in this table as in developing Table 1. In Table 2, modeled data is shown when the formation gas saturation is changed from 0% to 20% and 40% between the before and after frac logs. As expected, increased gas saturation in the formation pore space on the after frac log decreases Ʃ fm, but has no effect on Ʃ bh. This decrease in Ʃ fm will tend to offset any increase due to the tagged proppant. As can be seen in the Table 2 data with ΔS g = 40%, the tagged proppant Ʃ fm signal is completely masked by the presence of the gas on the after frac log. The same, however, is not true of the formation gate count rates, where the presence of gas has only a small effect on the count rate relative to tagged proppant. This indicates that changes in PNC formation gate count rates can be used to identify tagged proppant even in situations where the formation gas saturation changes

6 Percentage decrease of gamma ray counts (%) Percentage increase od formation SIGMA S_fm (%) SPWLA 54 th Annual Logging Symposium, June 22-26, % 10% 8% 6% 4% 2% 0% Figure 4a. Percentage increase of PNC tool computed formation thermal neutron capture cross section vs. fracture width (HTNCC concentration = 0.42%) 0% -5% -10% -15% -20% -25% -30% -35% -40% Fracture width (cm) Fracture width (cm) Figure 4b. Percentage decrease in MCNP computed PNC tool near detector capture gamma ray count rate vs. fracture width (HTNCC concentration = 0.42%) Limited studies have also been conducted [Smith 2012] to determine whether it might be possible to spectrally identify the capture gamma ray energy signals from the thermal neutron absorber in the HTNCC material in the presence of all the capture gamma rays from the other downhole elements present in the formation and borehole. Gadolinium and samarium (as well as several other thermal neutron absorbing rare earth elements) have distinctive capture gamma ray energy spectra, and one of the service companies has included gadolinium in their capture gamma ray library for spectrally deconvolving pulsed neutron capture gamma ray spectral field log data. Although HTNCC materials in the fractures and/or borehole region occupy only a very small fraction of the overall downhole formation and borehole volume, due to their extremely high capture cross sections, their presence results in a significant percentage (up to 30-50%) of the overall detected capture gamma ray counts coming from the HTNCC. The spectral presence of Gd 2 O 3 in the proppant can clearly be seen in the last field example in the paper. Monte Carlo Modeling Results for CNT Logging Tools Our initial Monte Carlo work in the development and testing of the NRT concept was conducted using 3-3/8 and 1-11/16 CNT logging tools with He3 detectors, and has been extensively described in prior publications [Smith 2009, Duenckel 2011]. Therefore, only very limited CNT modeling results will be presented in this paper. In the MCNP data shown in Figure 5, a 1.0 wide bi-wing fracture was utilized, and the same formation and borehole parameters have been modeled as in the PNC tool discussion above. The modeled CNT tool diameter was 3-3/8 (as opposed to 1 11/16 for the PNC modeling), and the modeled near and far detector spacings are typical of those in commercial dual-detector compensated neutron tools. Figure 5. MCNP simulated suppression in CNT near and far detector count rates as a function of the concentration of the HTNCC (frac width = 1.0 cm) In Figure 5, the percentage suppression in the CNT near and far detector count rates are plotted as a function of HTNCC (Gd 2 O 3 ) concentration in the proppant. As seen in Figure 5, count rate decreases in both the near and far detectors of up to 7-8% are observed. Since the statistical repeatabilities of near and far spaced CNT count rates at 30 ft/min logging speeds are usually significantly better than 1% in typical formations, it is clear that differences between before-frac and after-frac count rates due to tagged proppant should be easily detectable. If there is additional tagged proppant in the near-borehole region adjacent to these fractures, even larger decreases in count rate will be observed. Also, since the same percentage count rate decreases are observed in each detector, the direct corollary is that the CNT N/F ratio is unaffected by the presence of the tagged proppant. This has resulted in the development of a method, described below, to detect tagged proppant without the requirement for a before-frac log. When both before and after frac CNT logs are available, the insensitivity of N/F to the HTNCC has also provided a method to locate tagged proppant even when - 6 -

7 the formation gas saturation changes between the before and after frac logs. The N/F ratio, like the detector count rates, is sensitive to non-proppant related variations in formation porosity and gas saturation. If wellbore conditions are fairly uniform, logged intervals on the after-frac log known not to contain the HTNCC proppant can be used to develop a unique relationship between near detector count rate and N/F (and similarly between far detector count rate and N/F). Using these relationships, N/F from the after frac log can be used to compute synthetic near and far detector count rates across the entire logged interval, effectively creating a computed synthetic log which can be used in place of a before-frac log, but using only after frac log data. These computed/synthetic count rates, like the N/F ratio, are independent of the presence of HTNCC in the proppant. The synthetic log count rates can be compared to the actual observed after-frac count rates, which are sensitive to the presence of the proppant. The computed/synthetic count rates will be greater than the observed count rates in intervals containing HTNCC proppant, and log interpretation can be made without using a before-frac log. If before-frac and after-frac CNT logs are both available, changes in N/F between the two logs, caused by changes in formation hydrogen index (ΔHI), can be used to compute (for each detector) changes in count rates (ΔCR) between the logs resulting from the HI differences. ΔCR at each well depth can then be combined with the beforefrac count rate to produce a HI-corrected before-frac count rate log. This log will have the same HI response as the observed after-frac log, hence removing any count rate overlay sensitivity to gas saturation changes between the two logs. The HI-corrected before-frac count rate log (which is independent of tagged proppant) can then be compared to the observed after-frac count rate log, with tagged proppant indicated where the after-frac count rate is lower, thus providing a way to determine tagged proppant location from CNT log data even when the formation gas saturation changes between the logs. It should be noted that the data in Figure 5 were modeled with the fracture plane in the orientation shown in Figure 1b relative to the tool. When a small CNT tool (e.g. 1 11/16 ) is used in a large casing, however, additional modeling indicates that there can be up to a ~50% reduction in signal when the tool and fracture are misaligned. Therefore it is recommended that CNT tools having diameters as close as possible to the ID of the well casing be used for optimum results. A large tool also generally has reduced statistical uncertainties, and if there is a change in borehole fluid between the before-frac and after-frac logs, any required count rate curve normalizations will be much less significant. Other Monte Carlo Modeling Results In addition to the extensive modeling of the effects of HTNCC tagged proppant in vertical fractures in vertical wells, we have also done significant MCNP fracture modeling in horizontal and deviated wellbore geometries. Although the orientation of the tool relative to the fracture in a horizontal borehole geometry is radically different from the situation in vertical wells, the modeling indicates that both PNC and CNT tools can be used to locate HTNCC tagged proppant in horizontal and deviated wells. This is also confirmed from the many successful field frac identification logs run in horizontal wells using conventional R/A tracer logging systems, which are generally subject to the same geometrical constraints as our HTNCC based method. We have also modeled PNC and CNT logging with HTNCC material placed in gravel packs and cement. Our results indicate that PNC logs, and especially Ʃ bh and the observed capture count rates, can be effective in locating tagged proppant in gravel packs, and that both PNC and CNT tools can locate HTNCC material placed in cement. We have even had a successful field test with HTNCC tagged material added to cement. These modeling results will be presented in a future paper after adequate field testing has been completed. Log Examples In the following log examples, small diameter PNC tools and larger diameter CNT tools were utilized. All the PNC and CNT logs were run at reasonably normal logging speeds and utilized commercially available logging systems. We generally provided the log and log overlay processing, however similar results were obtained when done by the operating and/or service company personnel. In none of these examples were there any significant changes in gas saturations between the before and after frac logs, however the effects of gas saturation changes have been well documented previously [Smith 2009, Duenckel 2011]. Log Example 1. A major operator constructed an instrumented midcontinent well in which they wished to monitor/evaluate the production following a multi-stage frac job. State-ofthe-art distributed fiber-optic temperature (DTS) sensors, capable of making continuous measurements across all the zones of interest, and wireline temperature and spinner production logs were utilized to determine the intervals which had been effectively propped/stimulated

8 The NRT proppant was detected using both PNC and CNT logging tools. The operator used the NRT data to determine propped intervals, and also to evaluate the effectiveness of the NRT method by comparing the results with the DTS and production log data. To further test the viability of the NRT method, some stages were fractured using HTNCC tagged proppant and the remaining stages with non-tagged proppant, and the results compared. The initial plan was to also validate the NRT method by comparing it with data obtained using conventional radioactive tracers placed in the slurry during the stimulation operation. Unfortunately environmental, safety, and regulatory concerns caused the operator to cancel the use of radioactive tracers. Figures 6a and 6b present selected NRT logs, diagnostic DTS temperature logs, and production logs (temperature, differential temperature, and spinner). Data from two different frac stages in the well are illustrated. In both figures, the before-frac CNT and PNC logs used to detect NRT are indicated in blue and the after-frac logs in red. The gamma log is on the left (track 1); perforations (track 2); CNT ratio and detector count rates (tracks 3, 4, 5); PNC sigma and near detector formation count rates (tracks 6, 7); NRT proppant flag (track 8); DTS temperature (track 9); and production logs (temperature, differential temperature, and spinner) in track 10. Figure 6a presents data from stage 14, where NRT proppant was used. Before frac vs. after frac Ʃ fm overlays and PNC and CNT count rate overlays are shown in the figure. Increases in Ʃ fm and decreases in PNC and CNT count rates on the after frac logs were observed in and around all three perforated intervals. Since no changes were observed in the CNT N/F ratio between the logs, there were no gas saturation changes between the logs which might have otherwise affected the Ʃ fm overlay interpretation. It should be noted that the three intervals indicating the presence of NRT proppant are the identical intervals indicating subsequent production from the diagnostic temperature and spinner logs. In Figure 6b (stage 15), where untagged proppant was used, there were no NRT signals observed in any zones (as expected), including the two perforated zones where the diagnostic logs indicated production. Figure 7a shows the before vs. after frac PNC time decay crossplot from perforated interval X in the log in Figure 6a, where tagged proppant was utilized and indicated. Figure 7b presents the before frac borehole and formation decay components in the same interval calculated from the decay curve using a two-exponential model. Figure 7c shows an expansion of the CNT near and far detector count rate overlays and near and far Ʃ fm overlays in the same zone. The presence of the tagged proppant is clearly indicated. Figures 8a, 8b, and 8c are corresponding illustrations from a zone in the same well in a stage where standard untagged proppant was utilized. Note that no indication of tag material was observed in this zone on the crossplot or on any of the logs. Based on the excellent agreement between the NRT logs and the diagnostic logs in this well, the operator is continuing to use NRT technology in other wells in the region as a primary indicator of propped fractures. Log Example 2. Log example 2 is from a well in China in which NRT tagged proppant was placed in the deepest frac stage, and standard proppant was placed in the upper frac stages. The well was completed with 5.5 casing, and 35,000 lbs of NRT tagged proppant was pumped in the NRT stage to evaluate the frac job. The NRT stage was only 10 m above an aquifer, and hence the NRT tagged proppant was also useful in determining if the frac had extended into the water zone. A local Chinese logging service company was selected to run both CNT and PNC tools in the well. Figure 9 presents CNT and PNC before frac vs. after frac log overlay data in the stage where NRT tagged proppant was utilized. The CNT N/F ratio and near and far detector count rates are shown in the figure after the before frac logs had been normalized for a change in borehole fluid that took place between the times the logs were run. Also shown are PNC overlays of Ʃ fm from the far detector and both the near and far detector capture gamma ray count rates. The lack of N/F ratio separation confirmed that there had been no change in gas saturation between the logs. All the CNT and PNC logs gave similar indications of the presence of tagged proppant, and confirmed that the frac was contained and did not extend downward into the water zone. The insensitivity of N/F to the presence of the tagged proppant is also apparent in Figure 9. Log Example 3. The logs in example 3 are from a US mid-continent well which was fractured in multiple stages, and 0.4% Gd2O3 tagged proppant was used in some stages and untagged proppant in other stages. Tagged proppant was utilized in all the stages shown in Figure 10. PNC logs were run in the well at ft/min logging speeds by two major logging service companies, and one of the service companies also logged a separate run at a slower (<5 ft/min) logging speed to obtain capture gamma ray energy spectra in order to locate the gadolinium in the proppant using their standard spectral deconvolution software. Figure 10 gives before-frac log vs. after-frac PNC log overlays of Ʃ fm and the near and far detector capture gamma ray count rates, and also an overlay of the - 8 -

9 gadolinium yields computed from the spectra collected during the separate before and after frac logging passes at the slower logging speed. The location of the tagged proppant is indicated on all of the logs, and an excellent gadolinium tag signal (separation between the before and after frac yield curves) is seen from the spectral data. In this log, as in the other logs in this paper, relative differences seen in the HTNCC signals on the different logging curves is likely due to the differences in the radial depth of investigation of the different HTNCC indicating measurements. Figure 11 shows PNC overlay presentations/comparisons of data in this well using logs run by both service companies. All the stages in the illustrated logged interval contained the tagged proppant. The second service company also collected capture gamma ray spectral data during their PNC log runs, but as yet does not include gadolinium spectra in their spectral processing software library. As can be seen, very similar tagged proppant locations were obtained from the PNC logs of both service companies. This data, taken with the data shown in log example 2, indicate that our NRT method to detect non-radioactive tagged proppant is not restricted to using one particular logging company. Successful results have been obtained using CNT and/or PNC logs from all major US service companies and a number of different service providers in several international locations, including China, Russia, and Latin America. Summary/Conclusions A new non-radioactive traceable proppant and logging methods for near wellbore fracture height measurement have been developed and field tested. The tagged proppant contains a high thermal neutron capture compound (HTNCC) inseparably incorporated into each ceramic proppant grain during manufacturing in sufficiently low concentration that it does not affect mechanical strength, conductivity, durability, or density of the particles. The presence of the HTNCC is detected using commercial pulsed neutron capture (PNC) and/or compensated neutron (CNT) logging tools. One tagged proppant identification technique utilizes the comparison of before-frac vs. after-frac thermal neutron capture crosssections (Ʃ fm and Ʃ bh ) and detector capture gamma ray count rates using a PNC logging tool. The Ʃ fm measurement is sensitive to tagged proppant in the fracture, and is less affected by the presence of NRT material in the borehole region than the other measurements. The formation gated count rates indicate tagged proppant virtually independent of changes in formation gas saturation between the logs. A second tagged proppant identification technique utilizes PNC capture gamma ray energy spectroscopy to spectrally deconvolve and identify the HTNCC material (Gd 2 O 3 ) in the tagged proppant in the fracture. If no significant gadolinium is otherwise present downhole, it may be possible to employ this method, at least qualitatively, without requiring a before frac log. A third tagged proppant identification technique compares before-frac compensated neutron near and far detector count rates and count rate ratios (N/F) with corresponding after-frac measurements. The CNT technique also provides a method utilizing relationships between N/F and the observed detector count rates to eliminate the afterfrac log or to correct for changes in formation gas saturation between before-frac and after-frac logs. Extensive Monte Carlo modeling and field testing, including the modeling data and log examples in this paper, have validated the viability of all three tagged proppant identification techniques mentioned above. The HTNCC does not deteriorate over time, and after-frac logging operations may be conducted whenever desired to determine near-wellbore proppant concentrations or changes in concentrations. The new HTNCC based non-radioactive proppant tagging technology for locating propped fractures and determining frac height eliminates virtually all of the safety, environmental, and logistical/regulatory concerns associated with today s most common induced fracture identification method, which employs radioactive tracers placed in the frac slurry. ACKNOWLEDGEMENTS The authors wish to thank CARBO Ceramics for permission to publish this paper.

10 REFERENCES Bandy, T.R. Tracer Technology Finds Expanding Applications, Petroleum Engineer International, June 1989 Duenckel, Robert, et al, Field Application of a New Proppant Detection Technology, paper SPE , presented at the SPE Annual Technical Conference and Exhibition, October 2011 Gadeken L. L. and Smith, Jr., H. D., Tracerscan - A Spectroscopy Technique for Determining the Distribution of Multiple Radioactive Tracers in Downhole Operations, paper ZZ, Trans. SPWLA 27th Annual Logging Symposium, June 1986 Palisch, T, et al., Determining Realistic Fracture Conductivity and Understanding Its Impact on Well Performance Theory and Field Examples, paper SPE , presented at the SPE Hydraulic Fracturing Technology Conference, College Station, Texas, January 2007 Harry has had numerous SPWLA and SPE publications and has 89 issued US Patents. Robert Duenckel is Director Technical Development for CARBO Ceramics. Before joining CARBO Ceramics he held numerous engineering, supervisory and management positions with Marathon Oil Company. He holds a BS in Petroleum Engineering from Missouri University of Science and Technology. He has published a number of technical papers and has 7 issued US patents, and is a member of SPE. Xiaogang Han is a Petrophysicist at CARBO Ceramics. Before he joined Carbo Ceramics, he worked as research scientist at Baker Hughes for 5 years. He holds a Ph.D. in Nuclear Engineering from North Carolina State University, and has published over 30 academic and conference papers and has 6 US patents. Dr. Han is a recipient of the prestigious Mark Mills Award, granted by the American Nuclear Society. Before coming to the U.S., Xiaogang spent 6 years as a Research Assistant at the China Institute of Atomic Energy. He is a member of the SPWLA and SPE. Smith Jr., Harry D. and Duenckel, Robert, Method of Logging a Well using a Thermal Neutron Absorbing Material, US Patent Application 2009/ , August 2009 (subsequently issued as US 8,100,177) Smith Jr., Harry D. and Duenckel, Robert, Spectral Identification of Proppant in Subterranean Fracture Zones, US Patent Application 2012/ , April 2012 Schultz, W.E. et al, Experimental basis for a New Borehole Corrected Pulsed Neutron Capture Logging System (TMD*), paper CC, Trans. SPWLA 24th Annual Symposium, June 1983 ABOUT THE AUTHORS Harry D. Smith Jr. is sole proprietor of Harry D Smith Consulting. Harry was employed in Halliburton R&D for 28 years, during the last seven of which he was the Director of HES Research. He worked in Texaco Logging Research for 10 years prior to joining Halliburton. Harry was selected as an SPE Distinguished Lecturer twice, and has twice been an SPWLA Distinguished Speaker. He has been president of the Houston SPWLA chapter and a member of the SPWLA Board. Harry is the first person in SPWLA to have received both the Distinguished Technical Achievement Award and the Gold Medal for Technical Achievement

11 Table 1. MCNP simulated PNC results of different cases with 8 borehole and 5.5 casing. (borehole gate is sec, formation gate is sec) fm (cu) bh (cu) Near detector counts in formation time gate Near detector counts in borehole time gate No fracture HTNCC tagged proppant is present in 0.5 cm fracture HTNCC tagged proppant is present in 0.5 cm fracture; PNC logging tool is in Perpendicular position HTNCC tagged proppant is present in 0.5 cm fracture; kppm KCl in borehole No fracture, 20% of the cement space is filled with frac slurry HTNCC tagged proppant is present in 0.5 cm fracture and 20% of the cement space is filled with frac slurry HTNCC tagged proppant is present in 1.0 cm fracture Temperature Traces Production Logs X X X Figure 6a. Log example 1. Evaluated NRT log presentations from both CNT and PNC tools in stage 14 where tagged proppant was used. The NRT indicated propped fracture height (in yellow) compares favorably with subsequent temperature and spinner production logs and distributed fiber-optic temperature logs. 11

12 Table 2. MCNP simulated PNC results of different scenarios with 8 borehole and 5.5 casing. The borehole is filled with water. Formation is 28.2pu sandstone containing a vertical bi-wing 0.5 cm wide fracture (borehole gate is sec, formation gate is sec) fm (cu) bh (cu) Near detector counts in formation time gate Near detector counts in borehole time gate No fracture HTNCC tagged proppant is present in formation fracture, formation gas saturation is 0% HTNCC tagged proppant is present in formation fracture, formation gas saturation is 20% HTNCC tagged proppant is present in formation fracture, formation gas saturation is 40% Temperature Traces Production Logs X X X Figure 6b. Log example 1. Evaluated NRT log presentations from both CNT and PNC tools in stage 15 where untagged proppant was used. Temperature and spinner production logs and distributed fiber-optic temperature logs indicated production from several perforated zones. As expected, no NRT tagged proppant was indicated from any of the logs in this stage. 12

13 Figure 7a. PNC tool recorded gamma ray decay curves in well in Log Example 1 in the perforated interval x585 x594, where HTNCC tagged proppant was utilized Figure 7b. Computed PNC near detector before-frac formation and borehole decay components in interval x585-x594 Figure 7c. CNT N/F ratio and near and far detector count rate log overlays in the perforated interval x585 x594, where HTNCC tagged proppant was utilized. Also PNC near and far detector Ʃ fm log overlays in the same interval Figure 8a. PNC tool recorded gamma ray decay curves in well in Log Example 1 from a typical interval in a stage where untagged proppant was utilized Figure 8b. Computed PNC near detector before-frac formation and borehole decay components from a typical interval in a stage where untagged proppant was utilized 13

14 Figure 8c. CNT N/F ratio and near and far detector count rate log overlays from a typical interval in a stage where untagged proppant was utilized. Also PNC near and far detector Ʃ fm log overlays in the same interval. x x Figure 9. Log example 2. Before and after frac CNT and PNC log overlays. All logs were run by a local Chinese logging service company. Both CNT and PNC results indicated similar tagged proppant placement. 14

15 x x x Figure 10. Log example 3. A PNC tool was run using sigma mode and also a spectral data acquisition mode (at a slower logging speed). Gadolinium yield changes due to tagged proppant (derived from differences between before and after frac spectral yields) were computed and compared to the proppant related changes in formation sigma and near and far detector capture gamma ray count rates. 15

16 x y Figure 11. Log example 3. Excellent comparison of NRT evaluated PNC logs from two major service companies in an interval in the well in log example 3 where tagged proppant was utilized. 16