Numéro
Sci. Tech. Energ. Transition
Volume 78, 2023
Characterization and Modeling of the Subsurface in the Context of Ecological Transition
Numéro d'article 25
Nombre de pages 12
DOI https://doi.org/10.2516/stet/2023018
Publié en ligne 14 septembre 2023

© The Author(s), published by EDP Sciences, 2023

Licence Creative CommonsThis is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

1 Introduction

Full Waveform Acoustic Logging (FWAL) and Vertical Seismic Profile (VSP) were conducted over two boreholes from an experimental site in the Cher region (France). The site was also investigated (from the surface) by hybrid seismic imaging. The acoustic tool was modified to allow for passive monitoring in addition to more classical active measurements. To our knowledge, performing both passive and active acoustic loggings over the same well is a novelty that reveals how both types of signals can be used for characterizing a geological formation, mainly regarding lithology and flow detection.

Active acoustic logging is the record of acoustic waves transmitted in a borehole between one or more sources and one or more receivers housed by the same logging tool. Either monopole or dipole tools with multidirectional transmitters and receivers are used, the monopole being the most employed. In a borehole filled with fluid or mud, the transmitters generate a compression wave, which generates the formation of compression wave (P-wave) at the refraction limit angle. The measurement of arrival times of the refracted P-waves recorded by at least two adjacent receivers of the acoustic tool gives the P-wave Velocity (VP) of the geological formation. Relying upon so-called sonic logging to determine the velocity VP is a common and relatively well-established practice [1, 2].

A geological formation can simply be defined by its acoustic parameters, namely: P and S velocities (VP and VS), density, and QP and QS the factors of attenuation for P and S waves, respectively. A geological formation is considered “fast” if the shear velocity (VS) in the formation is higher than the P-wave velocity in the fluid (VP fluid). If this is not the case, the geological formation is referred to as a “slow” formation. In fast formations, a simple monopole tool is able to record a converted refracted S-wave, which then gives the shear velocity of the formation. In slow formations, a dipole tool should be used. Dipole acoustic tools are equipped with polarized transmitters and receivers generating and recording compression waves perpendicular to the borehole direction. These compression waves generate flexure modes at the borehole wall that induce pseudo-shear waves (flexural waves) in the formation propagated parallel to the borehole direction. Dipole tools are also used to evaluate the anisotropy of the formation by studying the phenomenon of shear wave splitting [3].

FWAL is based on the analysis and processing of the various wave trains (refracted waves, guided waves, reflected waves) recorded by an acoustic tool. Acoustic logging has a vertical resolution of a few centimeters, and a lateral one of centimeters for the interface-guided modes (Stoneley waves), of decimeters up to one meter for refracted modes, and up to ten meters for reflected modes. FWAL provides detailed information on a borehole as a function of depth, in terms of acoustic wave velocities and rock petrophysical characteristics [4]. In deviated wells, the exploitation of reflected modes may render an image comparable to a time microseismic section which, in favorable cases, allows for tracking layer boundaries and the estimation of their dip [5].

FWAL and VSP are currently used to obtain a time-to-depth conversion law, allowing for the transformation of a seismic block into an impedance or velocity block, the latter being then convenient to calculate pseudo porosity distributions for reservoir model building [6], detection of geological features such as fractured zones, karstic bodies [7] or preferential flow zones [8].

Flows can also be detected using ambient noise measurement, FWAL and VSP data recorded with a hydrophone sensor [9]. As an example, an experiment was carried out at the Hydrogeological Experimental Site (HES) of the University of Poitiers (France). Several VSP’s were conducted accompanied by ambient noise measurements recorded before each VSP shot. The VSP are highly corrupted, mainly because of the conversion of down going P-wave in down and up going Stoneley waves, especially at the level of karstic bodies (also detected by FWAL) crosscut by the well. Because flow circulation may introduce some changes in acoustic ambient noise properties, the analysis of ambient noise was also carried out. For that purpose, the average and the standard deviation of the amplitude spectrum of each noise trace were computed. The ambient noise factor defined as the ratio of the amplitude mean value to standard deviation significantly increases at the level of karstic bodies. The analysis of the ambient noise shows that the variations of the ambient noise factor are correlated with the level of P-wave conversion into Stoneley waves [9].

More recently, a source-free Distributed-Acoustic-Sensing (DAS) logging method based on borehole DAS ambient noise monitoring has been introduced [10]. The receivers of the source-free DAS logging tool are fiber optic cables cemented between the casing and the wellbore. The ambient noise Root-Mean-Squared (RMS) amplitudes correlate well with anomalies revealed by conventional logging data. The feasibility of a source-free DAS logging was evaluated by analysing a borehole DAS ambient noise dataset collected at the Frontier Observatory for Research in Geothermal Energy site in Utah. The borehole DAS ambient noise was collected within the granitic basement rock using a 1-km optic fiber cable cemented in the narrow interspace between casing and wellbore of a vertical observation well. The RMS noise amplitudes were computed to construct a profile showing RMS still varying over depth, with a few peaks at several distinct depths. By comparing the noise RMS profile with available logging data, a clear correlation appeared between major RMS peaks and Low-Velocity Layers (LVL) bounded by sharp structural interfaces. It was also found that the noise RMS amplitude peak zones only correlated with LVLs associated with both low Poisson’s ratio and high porosity, indicating highly fractured zones [10].

2 Active and Passive acoustic logging: the APEC field case

The experimental site investigated in this study is located in the Cher region (France) at the transition from Triassic to Jurassic geological formations, partly overlaid by thin superficial formations. The sedimentary formation is mainly composed of limestones up to 120 m depth and sandstones with some argillite and dolomite intercalations between 120 m and 200 m. The site was investigated from the surface via hybrid seismic imaging and from two boreholes (B1 and B2, Fig. 1) via FWAL and VSP [11]. Borehole B1 was drilled by Geocentre in 2006 up to a depth of 90 m but is now fully steel cased (internal diameter: 154 mm, external diameter: 178 mm) and cemented. Borehole B2 was drilled by Geocentre-Forsol [11*] in two phases between September 2019 and September 2020. The first drilling phase from the surface up to 78 m depth resulted in a steel cased (internal diameter: 154 mm, external diameter: 178 mm) but not cemented borehole. Re-handling the borehole within the second drilling phase allowed for reaching the depth of 200 m in open hole, with a borehole completely cored between 78 and 200 m, then equipped with a slotted PVC casing (internal diameter: 80 mm, external diameter: 90 mm). A lithological column was obtained in B2 by core description. A VSP was recorded in borehole B1 along the 25 −90 m depth interval (Fig. 1a), and FWAL were conducted both in boreholes B1 and B2.

thumbnail Fig. 1

Seismic imaging: (a) 2D seismic spread – 2D refraction tomography, borehole locations (B1 and B2), view of the seismic source, VSP recorded in B1, (b) 2D hybrid section over depth. After [11].

A seismic line was also recorded at the site with a seismic spread composed of 48 fixed geophones (2 m lag distance between neighbors, Fig. 1a), while the source, as a weight dropper (Fig. 1a), was moved and fired in the middle of all pairs of adjacent geophones. Hybrid seismic imaging combining refraction (tomography, Fig. 1a) and reflection seismic results produced an extended depth reflectivity section starting from the surface up to a depth of 240 m (Fig. 1b). Time to depth conversion was calculated using the time-depth law given by the VSP recorded in borehole B1 [11].

3 Active acoustic logging

The acoustic tool used for the field experiments is a monopole-type flexible tool with a small diameter of 50 mm. It holds a magnetostrictive transmitter (transmission frequencies: 17–22 kHz) and can be equipped with two pairs of piezoelectric receivers offering an acquisition in near offset configuration (receivers at 1 and 1.25 m beneath the source), and in far offset configuration (receivers 3 and 3.25 m beneath the source). A far offset configuration must be favored for the measurement of the wave parameters such as P-wave propagation velocity (VP), especially in the case of poorly cemented boreholes (see Appendix).

In the acquisition of acoustic data in boreholes B1 and B2, with the aim of detecting preferential flow horizons and calibrating seismic data, we finally relied upon the far offset configuration and a logging speed of 5 m/min. The acquisition parameters were set as follow, with a depth sampling interval: 5 cm, a time sampling interval: 5 μs, and a recording length: 5 ms. Borehole B2 being steel cased but not cemented in the upper part, strong resonances are observed on the acoustic sections. Consequently, acoustic data recorded in borehole B1 (steel cased hole) in the 30–78 m depth interval have been merged with acoustic data recorded in borehole B2 (slotted PVC cased hole) in the 78–192 m depth interval to obtain a composite acoustic section (Fig. 2). Acoustic data were processed to obtain a very high-resolution velocity log (Fig. 2) which was used as a constraint to transform the seismic section into pseudo-velocities. After calibration, and under the assumption that the slower the wave velocity, the higher the permeability – porosity, the seismic line informs on preferential areas where flow should occur. At the scale of an open wellbore, acoustic loggings reporting on wave velocities over a short distance within the well, also inform on open features crosscut by the borehole. The velocity distribution has been converted into porosity using the Raymer equation adapted to carbonate formation. The results are shown within the 30–190 m depth interval in Figure 2, with high-porosity layers appearing in red.

thumbnail Fig. 2

Seismic imaging and active acoustic logging: from left to right: lithological column, acoustic velocity, seismic section converted in porosity, acoustic porosity, Full wave acoustic composite section (0–80 m: borehole B1, 80–190 m: borehole B2 slotted PVC cased hole) and flowmeter. After [11].

A flow meter was recorded in borehole B2 (Fig. 2). Flow-log measurements were performed with a downward moving probe at constant speed corresponding to a constant rotation of the propeller of the flowmeter, the rotation speed being on the order of 3.3 rotations/second. The flowmeter measurements evidence that water flows from the formation into the well at 83–85 m depth, then flows downward into the well and goes back into the formation at 143 m. The same geometry of flow occurs between 152 and 156 m, and then from 159 m to 181–183 m. The water “loops” are indicated by blue arrows on the flow log, the arrows also pointing out that the Stoneley waves (between 2.5 and 3.5 ms on the FWAL, in Fig. 2) are strongly attenuated at the depths of water loops. Flows is also witnessed by low velocity levels corresponding to high porosity horizons.

4 Passive acoustic logging in borehole B2

The acoustic tool was modified to perform passive full waveform acoustic logging in which the magnetostrictive source is inactivated. The logging speed is 2 m/min. The acquisition parameters are depth sampling interval: 1 cm, time sampling interval: 10 μs, recording length: limited to 10 ms, due to the acquisition device. Four runs were recorded in the passive mode to obtain a set of acoustic noise sections, with two runs in the downward direction and two in the upward direction. The acoustic noise sections were resampled by Fourier transform at a depth sampling interval of 5 cm, Figure 3a giving the passive acoustic data between 80 and 180 m depth in well B2 for 10 ms of recording duration at each depth.

thumbnail Fig. 3

Passive acoustic logging in borehole B2. Noise section (a), Noise analysis in different frequency bandwidths (0–5 kHz (black curve), 5–10 kHz (red curve), 10–15 kHz (green curve), 15–20 kHz (blue curve)). Noise spectral amplitude logs (b), Noise frequency logs (c).

Acoustic noise sections have been processed to perform both a spectral noise analysis and an interferometry analysis [12].

4.1 Spectral Noise Analysis

At each depth, the RMS amplitude and the amplitude spectrum of the acoustic noise trace were computed in separating the raw signal into different frequency bandwidths (0–5 kHz, 5–10 kHz, 10–15 kHz, 15–20 kHz). In each bandwidth, the frequency associated with the maximum of the amplitude spectrum was selected, leading to the assemblage of a noise frequency log (Fig. 3c) and an associated noise spectral amplitude log (Fig. 3b) per frequency bandwidth.

Figures 3b and 3c show that the major part (80%) of the energy (highest amplitudes, in short) is concentrated in the 0–5 kHz frequency bandwidth. A strong decrease of the frequency content and of spectral amplitude is noticeable in the 142–150 and 155–157 m depth intervals, with, in these intervals, a dominant frequency smaller than 0.5 kHz. We can also notice the presence of peaks both on the RMS noise amplitude log and on the spectral amplitude in the 15–20 kHz frequency band at depths of 152 and 156 m (Fig. 4).

thumbnail Fig. 4

Spectral noise analysis. From left to right: Noise RMS amplitude log, spectral amplitude log and frequency log in 0–5 kHz frequency band, spectral amplitude log in 15–20 kHz frequency band.

4.2 Interferometry

As the noise is simultaneously recorded by two receivers of the tool, an interference noise section was constructed by correlating and summing pairs of acoustic traces at each depth, the correlation being done between the noise recorded by the first receiver and the noise recorded by the second receiver, 0.25 m away from the first receiver. This procedure can be seen as an Interferometry analysis [12]. The correlation is calculated at each depth by Fourier transform. The cross-correlation sections computed for each run (4 runs being recorded) were stacked to obtain the interference noise section (Fig. 5a).

thumbnail Fig. 5

Passive acoustic logging processed by interferometry. a: interference noise section. The section is normalized and displayed with a color scale ranging from 0 (blue) to 1 (red). Arrows indicate the presence of acoustic waves. b: from top to bottom: zoom of the cross-correlation noise section in 0–0.5 ms time interval, picked times DTmax of the maximum of the cross-correlation noise section, flattening of the cross-correlation noise section with DTmax.

The causal part of the interference is associated with correlation over positive times and the acausal part with correlation over negative times, those times corresponding to up-going and down-going wave fields, respectively. In this analysis, we consider that the noise recorded by the first receiver acts as a source for the noise recorded by the second receiver. The presence of events, coherent between two receivers but delayed in time (not located at the same relative time zero) and with significant amplitude indicates that acoustic waves are propagating in the borehole. Such events, indicated by red arrows in Figure 5a, can be seen in the interference noise section. The phenomenon is better evidenced in the causal part of the interference noise section. As the events exist both in the causal and acausal parts of the interference noise sections, the wave propagation occurs in both directions, downward and upward, in the borehole.

Figure 5b shows a zoom of the causal part of the interference noise section in the 0–0.5 ms time interval, where the acoustic wave is clearly visible. The picking of the maximum of the interference section gives the (statistical) transit times DTmax of the acoustic wave between the two receivers (0.25 m lag distance) of the acoustic tool. The picked times are used to flatten the interference noise section. A correct flattening of the interference section with the picked DTmax guarantees an accurate selection of those transit times. The picked times DTmax were used to obtain a log of propagation velocity V-DTmax of the acoustic wave as well as a frequency log and a normalized energy log by extracting the instantaneous frequency and amplitude of the interference section at the time DTmax. The results are shown in Figure 6. The normalized energy log shows peaks of energy between 175 and 185 m. The variations of velocity, frequency and normalized energy logs point out different noise units: 80–118 m depth, 118– 58 m, 158–175 m, 175–185 m, and 185–194 m (Fig. 6b). The acoustic wave identified by interferometry is a Low Frequency Acoustic wave (LFA wave) visible in the interference noise section at time DTmax (Fig. 6a) for all depths. The LFA wave is identified as a Stoneley wave by comparison with acoustic data obtained in the active mode.

thumbnail Fig. 6

Passive acoustic logging: Identification of a Low Frequency Acoustic (LFA) wave. a: causal part of the interference noise section. The section is normalized and displayed with a color scale ranging from 0 (blue) to 1 (red). b: acoustic logs computed in the 0–0.5 ms interval, c: acoustic logs associated with the LFA wave.

For the purpose of inter-comparison between active and passive acoustic logging, Figure 7a shows the 3 m constant offset acoustic section obtained with the acoustic tool working in active mode. Stoneley waves are clearly visible in the 2.5–5 ms time interval. The 3 m and 3.25 m constant – offset acoustic sections have been filtered in the 0.5–7.5 kHz frequency bandwidth corresponding to the band of the LFA wave identified via passive acoustics (Fig. 8c).

thumbnail Fig. 7

Active vs Passive acoustic logging. a:3 m constant offset acoustic section, b: constant offset section filtered in the 0.5–7.5 kHz interval, displayed in the 2.5–5 ms interval showing Stoneley waves, c: active acoustic logging: velocity log of the Stoneley wave (V-St), passive acoustic logging: velocity of the LFA wave (V-DTmax), Relative uncertainty DV/V between the 2 velocity logs. The acoustic sections are normalized and displayed with a color scale ranging from 0 (blue) to 1 (red).

thumbnail Fig. 8

Passive acoustic logging: processing of the causal part of the interference noise section. a: Raw interference noise section, b: events associated with the Stoneley wave, c: residual interference noise, d: frequency and normalized energy logs associated with the residual section. The acoustic sections are normalized and displayed with a color scale ranging from 0 (blue) to 1 (red).

One can notice, on the active acoustic log that the Stoneley waves are strongly attenuated in the 120–133 m and 137–153 m depth intervals. The propagation velocity (V-St) of the Stoneley waves was obtained by cross-correlating the signals in the time window 2.5–5 ms of the two receivers at 3 and 3.25 m offset. The Stoneley wave velocity log (V-St) is shown in Figure 7c (top). In the 120–133 m and 137–153 m depth intervals, the velocity is close to the fluid velocity, with an average value of 1560 m/s and a standard deviation of 105 m/s. In the 80–120 m, 133–137 m and 153–190 m depth intervals, the velocity varies between 1000 m/s and 1400 m/s. In these depth intervals, the low frequency events correspond to Stoneley waves. When comparing the velocity log (V-St) with the LFA wave velocity log (V-DTmax, Fig. 7c middle), the 80–120 m and 153–190 m depth intervals, show very similar value indicating that the LFA wave is a Stoneley wave in these intervals. In the 120–153 m depth interval, the LFA-wave velocity is close to the fluid velocity (1527 m/s). The relative uncertainty ΔV/V (Fig. 7c, bottom) between the two velocity logs is approximately of 10–20%.

This field example shows that ambient noise recorded over short time windows (10 ms) can be used to detect wave propagation and to estimate the velocity of the detected wave with acceptable uncertainties. Consequently, acoustic logs such as RMS amplitude, frequency and spectral amplitude computed from ambient noise records can be obtained with the same level of uncertainty as that from active acoustic logs.

A 11-terms moving average filter was applied to the causal part of the interference noise section to enhance the characterization of the Stoneley wave detected by interferometry. The results are shown in Figure 8. The raw interference noise section (Fig. 8a) is separated into two parts: the events associated with the Stoneley wave (Fig. 8b) and the residual interference noise section (Fig. 8c). A spectral analysis of the residuals recovers the noise units identified by the Stoneley wave (Fig. 6) detected by interferometry. This result would indicate that the coda of passive recordings is dominated by multiples of the previously identified Stoneley wave, which is consistent with records of Stoneley waves up to 4.0 ms in the active acoustic logs (Figs. 7a and 7b).

Conventionally in seismic, ambient noise data processing is divided into three main phases: (1) single station data preparation, (2) cross-correlation and temporal stacking, and (3) measurement of dispersion curves (performed with frequency – time analysis for both group and phases velocities) for the reconstructed surface waves (pseudo-Rayleigh waves) [12]. The dispersion curves of surface waves can fruitfully be used to recover the shear velocity of the geological formations investigated via the surface waves [13].

In the present case dealing with acoustic noise, the velocity profile of the Stoneley wave evidenced by interferometry can be used to estimate a shear velocity log. The results are shown in Figure 9.

thumbnail Fig. 9

Passive acoustic logging: Shear velocity estimation. From left to right: lithology, P-wave velocity, Stoneley wave velocity, S-wave velocity, Poisson’s ratio (red arrows indicate the depth intervals where Poisson’s ratio is fixed at 0.3).

According to White [14], VS can be derived from a simplified version of the dispersion equation relating the shear-wave velocity (VS), the low-frequency Stoneley velocity (VSt), the fluid velocity (Vf), and formation and fluid densities (ρ and ρf, respectively):

As no density log is available, an estimation of ρ can be given by a Gardner equation: ρ = αVPβ, with α and β scalar coefficients, here prescribed at 0.3 and 0.25, respectively. The shear velocity (VS) estimation from Stoneley-wave velocity assumes both that VS < 0.65 × VP and the Poisson’s ratio ranges between 0.25 and 0.45.

The simplified dispersion equation is valid if VSt is smaller than Vf. Otherwise, VS must be assessed in another way, for example relying upon VP to VS relationships established by comparing in-situ and laboratory data [15]. In our case, the Poisson’s ratio for which the dispersion equation can be applied is fixed at 0.3 which is a mean value for a Poisson’s ratio. The depth intervals with this value of 0.3 are marked in red in Figure 9. The shear velocity log is noted VS-St in Figure 9 to remind that the shear velocity is estimated from Stoneley wave velocity. We can notice that the same phenomenon is observable in the same depth intervals using VSt measured from acoustic data recorded with an activated acoustic source (Fig. 7).

In seismology, the great variety of noise sources imposes some forms of correction applied to the raw signal to avoid capturing the characteristics of diverse sources in the correlation process. It is well known that changes in the frequency content of the sources may affect the apparent velocity of the reconstructed phases.

Hence, the interference acoustic noise section was constructed by deconvolution and summing pairs of acoustic traces at each depth, the deconvolution being done between the noise recorded by the first receiver and the noise recorded by the second receiver, 0.25 m away from the first receiver. The comparison with Figure 5 shows that both processes provide homogenized interference acoustic sections in which there is no strong variations over depth. In both cases, the resulting acoustic wave field is smoother, and mainly restricted to the direct arrival phase at 0.15 ms (the most energetic phase). Interestingly after deconvolution and sum, two maxima of energy are located at depths of 140–151 m and 153–157 m (Fig. 10b). Interestingly, the interferometry analysis, shows increased energy of the “synthetic” wave (a covariance between two signals monitored at short distance between them) in the interval 140–151 m where no flow occurs in the well (see flow log in Fig. 2). For its part, the raw noise signal (Fig. 4) shows, in the same non-flowing interval, a strong decrease in the band 0–5 kHz of both the frequency and the amplitude of the signal. It must be raised that interferometry, i.e., correlation, can either “correlate” a weak signal with a weak signal, or a strong signal with its strong counterpart. This would also mean that interferometry could either see flow or conversely no-flow in a well. At this stage of our pioneering investigations regarding passive acoustics, we would advise to be confident in the raw noise signal to infer flow along the well generating a low frequency Stoneley wave (evidenced via raw noise signal spectral analysis), and interferometry to qualify the propagated wave and return by-products such as wave transit times and velocities, or formation velocities (see above).

thumbnail Fig. 10

Interference noise section obtained by deconvolution and summing. a: interference section, b: causal part and associated acoustic logs computed in the 0–0.5 ms interval: frequency log and normalized energy log.

Notably, the efficiency of the interferometry process could be enhanced by increasing the number of experiments (here runs). For that purpose, the acquisition of ambient noise could be done simultaneously on several receivers of an acoustic tool and also by increasing the record duration.

5 Discussion

When a fluid is moving through or along a medium it produces an acoustic noise stemming from both the fluid itself and vibrating elements streamlined by flow. The fluid noise is the result of internal friction, usually audible in highly turbulent flows, with a noise spectrum strongly depending on the fluid type, pressure, temperature, and flow rate. Different types of rocks generate acoustic noise in different frequency bandwidths [16].

The acoustic logs obtained from passive acoustic logging were compared with P-wave acoustic velocity, core data and flowmetry.

The interpretation of passive acoustic data processed by spectral analysis render three main sets of results.

  • The distribution of noise frequencies and of noise RMS amplitudes are correlated with the lithology (core description) and the (seismic) P-wave velocity log (Fig. 11a). A blocking algorithm (separating data in blocks of coherent information) was applied to the P-wave velocity log to identify the geological units characterized by linear velocity laws (linear behavior overprinted on the velocity log). The limits of geological units are consistent with both the variations of the passive acoustic logs and the lithological column.

  • The distribution of noise frequencies in the 0–5 kHz bandwidth are strongly correlated with the variations of the flowmeter (Fig. 11b, the levels of in-flows or out-flows are indicated by blue arrows on the flowmeter). The flows observed at different depth intervals (83–143 m, 152–156 m, and 159–181 m) are characterized by both a sharp decrease in the rotation of the propeller of the flowmeter, and a complete independence over depth of the flow loops, those being well-separated by segments where no flow occurs in the well. The phenomenon is clearly visible on the spectral analysis of passive acoustic logs in the 0–5 kHz bandwidth, with an increase of noise frequency and noise amplitude where flow loops occur in the well.

  • Processing the passive acoustic data by interferometry (seeking correlation between two signals collected at a short lag distance) allows for reconstructing an acoustic wave identified as a Stoneley wave, the velocity of which can be used to estimate the shear velocity of geological formations crosscut by drilling.

thumbnail Fig. 11

Passive acoustic logging vs lithology and flowmetry. a: lithology. From left to right: lithological column, P-wave velocity log, noise RMS amplitude log. b: flowmetry. From left to right: flowmeter, frequency log and spectral amplitude log in 0–5 kHz from noise analysis.

6 Conclusion

APEC (Association Pédagogique et Expérimentale du Cher) has developed a site for the training in applied geophysics of both students and professionals. Two boreholes are available at the site. The site is currently used for experimental studies in near surface geophysics and logging.

The paper has been focused on acoustic logging with a tool modified to conduct both passive and active acoustic monitoring. The paper has shown how active and passive acoustic logging can be used for identifying preferential flow locations in the first hundred-meter depth of an aquifer.

Active acoustic logging is currently run to estimate the P-wave velocity of the formation and to calibrate data from seismic sections with the aim to convert monitored seismic wave amplitudes into velocity and porosity (Fig. 2).

The acoustic data from passive acoustic logging have been compared with P-wave acoustic velocity from active acoustic logging, with core data, and flowmetry. The distribution of noise frequencies and noise RMS amplitudes are correlated with both the lithology (core description) and the P-wave velocity log. The interferometry process has revealed the presence of a Low Frequency Acoustic wave (LFA wave) in the noise data, which, by comparison with active acoustic logging, has been identified as a Stoneley wave. The velocity variations of the LFA-wave and the distribution of noise frequencies in the 0–10 kHz are strongly correlated with the lithology. The distribution of noise frequencies in the 0–5 kHz bandwidth, obtained by spectral analysis, are strongly correlated with the variations of the flowmeter with an increase of noise frequency and noise amplitude where flow loops occur in the well.

The field example has shown that recording ambient noise over short time windows (limited to 10 ms due to the acquisition device) can be used to detect wave propagation and estimate velocities of detected waves, to compute acoustic logs such as RMS amplitude, frequency, and spectral amplitude from ambient noise records with acceptable uncertainties.

The acquisition of ambient noise simultaneously on several receivers of an acoustic tool provides data redundancy without significantly increasing the acoustic logging time. As an example, passive acoustic data could be recorded when the acoustic tool is run downward and active acoustic data when the tool is run upward.

In this context, active and passive full waveform acoustic logging can be seen as measurements of very high resolution for reservoir description that complement hybrid seismic imaging. This type of approach, which would condition the building of static reservoir models by detecting locations of preferential subsurface flows, could also be fruitful to dynamic reservoir monitoring or CO2 storage.

Acknowledgments

We thank APEC (Association Pédagogique et Expérimentale du Cher) for granting us the permission to use full waveform acoustic data and logging data. The time picking of acoustic sections was done with Earth-Quick software. The processing of seismic data (2D seismic imaging and refraction tomography) was done with the SPW software (Parallel Geoscience). We thank Thomas Kremer from Geolinks for useful discussions, especially regarding the interferometric aspects presented in the paper.

Funding

The authors acknowledge financial support from APEC.

References

Appendix

Acoustic logging – Near offset versus Far offset

In a vertical well, monopole tools can enable the recording of five propagation modes as: refracted compression waves (P), refracted shear waves (S, only in fast formations VS > VP fluid), fluid waves (F), and two dispersive guided modes as pseudo-Rayleigh waves (in fast formations), and Stoneley waves (ST).

In addition to these modes, acoustic profiles of constant offset section may show coherent slanted events and resonances (R). The slanted events, conventionally named criss-cross events, are refracted events reflected on the edges of geological discontinuities (acoustic impedance discontinuities), such as fractures. For their part, the resonances are related to poor cementation between the casing and the formation. A high level of resonances can result in unusable acoustic data.

Figure A1 shows a comparison between acoustic data recorded using the tool in the near (or short) offset configuration (receivers at 1 and 1.25 m from the source) and in the far (or large) offset configuration (receivers at 3 and 3.25 m). In borehole B1 which is poorly cemented resulting in strong resonances, the short offset configuration (1 m) only shows refracted P-waves in the 60–70 m depth interval. In opposition, the large offset is less sensitives to resonances, letting clearly appear refracted P-waves along the profile. For both offsets, the differentiation between the refracted P-waves (P) and the Stoneley waves (ST) can be done easily. In the open hole part of borehole B2, the presence of a piece of casing generates resonances in the depth interval 88–91 m on the 1 m offset section. The influence of the piece of casing is local on the 3 m offset section, indicating that the length of the piece of casing is a slightly larger than 3 m. On the 1m offset section, it is possible to identify the refracted P-waves, locally the converted refracted S waves, the Stoneley modes and the fluid modes. With a short offset configuration, the different wave trains can interfere. However, we can notice that a large offset (3 m) better separates the different wave trains over time due to the difference in their propagation velocities. On the 3 m offset section, criss-cross events are visible. The short offset configuration must be favored to evaluate the borehole cementation. For the measurement of the wave parameters such as amplitude, frequency content, propagation velocity, a large offset configuration must be favored.

thumbnail Fig. A1

Active acoustic logging: comparison of short offset (1 m) and large offset (3 m) acoustic sections recorded in borehole B1 (a, poorly cemented borehole) and in borehole B2 (b, open hole part). The different wave trains are identified by letters: C casing resonances, P refracted P-wave, S refracted S-wave, ST Stoneley wave, F fluid waves, Criss-cross. The acoustic sections are normalized and displayed with a color scale ranging from 0 to 1.

All Figures

thumbnail Fig. 1

Seismic imaging: (a) 2D seismic spread – 2D refraction tomography, borehole locations (B1 and B2), view of the seismic source, VSP recorded in B1, (b) 2D hybrid section over depth. After [11].

In the text
thumbnail Fig. 2

Seismic imaging and active acoustic logging: from left to right: lithological column, acoustic velocity, seismic section converted in porosity, acoustic porosity, Full wave acoustic composite section (0–80 m: borehole B1, 80–190 m: borehole B2 slotted PVC cased hole) and flowmeter. After [11].

In the text
thumbnail Fig. 3

Passive acoustic logging in borehole B2. Noise section (a), Noise analysis in different frequency bandwidths (0–5 kHz (black curve), 5–10 kHz (red curve), 10–15 kHz (green curve), 15–20 kHz (blue curve)). Noise spectral amplitude logs (b), Noise frequency logs (c).

In the text
thumbnail Fig. 4

Spectral noise analysis. From left to right: Noise RMS amplitude log, spectral amplitude log and frequency log in 0–5 kHz frequency band, spectral amplitude log in 15–20 kHz frequency band.

In the text
thumbnail Fig. 5

Passive acoustic logging processed by interferometry. a: interference noise section. The section is normalized and displayed with a color scale ranging from 0 (blue) to 1 (red). Arrows indicate the presence of acoustic waves. b: from top to bottom: zoom of the cross-correlation noise section in 0–0.5 ms time interval, picked times DTmax of the maximum of the cross-correlation noise section, flattening of the cross-correlation noise section with DTmax.

In the text
thumbnail Fig. 6

Passive acoustic logging: Identification of a Low Frequency Acoustic (LFA) wave. a: causal part of the interference noise section. The section is normalized and displayed with a color scale ranging from 0 (blue) to 1 (red). b: acoustic logs computed in the 0–0.5 ms interval, c: acoustic logs associated with the LFA wave.

In the text
thumbnail Fig. 7

Active vs Passive acoustic logging. a:3 m constant offset acoustic section, b: constant offset section filtered in the 0.5–7.5 kHz interval, displayed in the 2.5–5 ms interval showing Stoneley waves, c: active acoustic logging: velocity log of the Stoneley wave (V-St), passive acoustic logging: velocity of the LFA wave (V-DTmax), Relative uncertainty DV/V between the 2 velocity logs. The acoustic sections are normalized and displayed with a color scale ranging from 0 (blue) to 1 (red).

In the text
thumbnail Fig. 8

Passive acoustic logging: processing of the causal part of the interference noise section. a: Raw interference noise section, b: events associated with the Stoneley wave, c: residual interference noise, d: frequency and normalized energy logs associated with the residual section. The acoustic sections are normalized and displayed with a color scale ranging from 0 (blue) to 1 (red).

In the text
thumbnail Fig. 9

Passive acoustic logging: Shear velocity estimation. From left to right: lithology, P-wave velocity, Stoneley wave velocity, S-wave velocity, Poisson’s ratio (red arrows indicate the depth intervals where Poisson’s ratio is fixed at 0.3).

In the text
thumbnail Fig. 10

Interference noise section obtained by deconvolution and summing. a: interference section, b: causal part and associated acoustic logs computed in the 0–0.5 ms interval: frequency log and normalized energy log.

In the text
thumbnail Fig. 11

Passive acoustic logging vs lithology and flowmetry. a: lithology. From left to right: lithological column, P-wave velocity log, noise RMS amplitude log. b: flowmetry. From left to right: flowmeter, frequency log and spectral amplitude log in 0–5 kHz from noise analysis.

In the text
thumbnail Fig. A1

Active acoustic logging: comparison of short offset (1 m) and large offset (3 m) acoustic sections recorded in borehole B1 (a, poorly cemented borehole) and in borehole B2 (b, open hole part). The different wave trains are identified by letters: C casing resonances, P refracted P-wave, S refracted S-wave, ST Stoneley wave, F fluid waves, Criss-cross. The acoustic sections are normalized and displayed with a color scale ranging from 0 to 1.

In the text

Les statistiques affichées correspondent au cumul d'une part des vues des résumés de l'article et d'autre part des vues et téléchargements de l'article plein-texte (PDF, Full-HTML, ePub... selon les formats disponibles) sur la platefome Vision4Press.

Les statistiques sont disponibles avec un délai de 48 à 96 heures et sont mises à jour quotidiennement en semaine.

Le chargement des statistiques peut être long.