Surface-to-Borehole Electromagnetics Hold Promise for 3D Waterflood Monitoring
You have access to this full article to experience the outstanding content available to SPE members and JPT subscribers.
Monitoring the waterflooding oil-recovery process is a difficult task for seismic-based methods in hard carbonate reservoirs. The changes in velocity and density caused by water/oil substitution are too small when compared with the errors involved in repeating the measurements. The authors detail the development of a technique based on surface-to-borehole controlled-source electromagnetics (CSEM), which exploits the large contrast in resistivity between injected water and oil to derive 3D resistivity distributions, proportional to saturations, in the reservoir.
CSEM techniques for reservoir-fluid characterization and monitoring are applied on a commercial basis for cross-well configurations. The method is based on electromagnetic (EM) induction and, as such, uses magnetic sources and magnetic receivers. While it has been implemented successfully in many field trials, the method is limited to 2D geometries and its sensitivity is biased toward detecting conductors. Modeling has shown that the setup of magnetic sources and magnetic receivers in a 3D surface-to-borehole configuration does not provide a useful signal above the estimated noise floor. For such configurations, an electric source needs to be adopted. Realistic modeling of 3D surface-to-borehole CSEM suggested that the vertical electric field (Ez) is the only component of EM radiation showing a signal above the estimated noise floor for waterfront variation over a period of 2 years. The results of the study led to modifications of the cross-well EM acquisition system where electric field sensors were added to the existing magnetic field sensors. A powerful custom electric-current source was also developed.
Three-dimensional surface-to-borehole CSEM data were acquired in a deep (reservoir-level) research well drilled near the known waterfront position. Surface current dipoles (transmitting antennae) were prepared by drilling shallow boreholes and completing them with a steel pipe coupled to the rock formations using a slurry of carbon backfill (coke breeze) to achieve sufficient electrical contact. The electrode setup was completed by cementing the top. Electrodes forming the dipole antennas were installed at a 200-m separation forming a setup of 48 inline (radial-direction) and 48 cross-line (tangential-direction) dipoles generating Ez and a vertical magnetic field (Hz) at the observation well.
The borehole receiver comprised two vertical magnetic field sensors (H1 and H2) and two vertical electric field sensors (E1 and E2) coupled to the borehole walls by means of well centralizers. The receiver array was operated as a wireline tool, with which special care had to be taken to insulate the electrodes from the wireline cable and from the body of the tool (Fig. 1). The receiver (Rx) setup was operated by means of a crane, and synchronization of the Rx and the transmitting (Tx) position was provided by Global Positioning System timing. Low-frequency magnetotelluric and surface/surface CSEM data were also acquired to build a 3D model of the overburden. The additional data provide primary sensitivity to the overburden and allow control and discrimination of the response of the overburden from the response of the reservoir during interpretation.
Measured data are of good quality and recorded the fundamental transmitted frequency of 8 Hz even from the most-distant Tx positions at 3.5 km (horizontal) from the observation well. The 8-Hz transmitted signal is not affected by other signals. Considering that this was the first time that such surface-to-borehole EM measurements were taken, this represented a milestone of the experiment, suggesting that the acquisition setup was effective and working as expected.
The amplitude measurements slowly decay with depth as the distance from the Tx source increases. The relative offset between the curves is related to the increasing distance from the Tx position from the well. An azimuthal analysis around the well also shows small variations of Hz, suggesting little sensitivity to reservoir-saturation variations. The Ez measurements provide a large sensitivity to the resistivity structure at the well when compared with the resistivity log. Vertical changes in resistivity cause changes in Ez at the borehole because of the requirement that the vertical current density (Jz) be continuous. The measured Ez field is influenced primarily by the reservoir resistivity structure at the well position. Using a reference resistivity of an Ez logged section beneath the reservoir, the relative Ez variations can be calibrated and the reservoir resistivity profile reconstructed directly from the Ez measurements. The reconstructed resistivity distribution matches quite well with the log resistivity, except in the upper section of the reservoir, where the effect of the casing begins to influence the Ez measurements. The Ez measurements need to be corrected for the secondary field produced by the steel casing, which decays rapidly with increasing distance from the casing shoe.
Azimuthal and offset variations in the data are compared using circular plots of the interpolated Ez data and the corresponding resistivity variations derived from the reservoir simulator. A finite-difference representation of the steel casing is also incorporated in the model during the forward calculation. The residual Ez phase data are therefore the primary source of information for the saturation-related resistivity distributions in the reservoir. A marked anisotropy (azimuthal variation) exists from the west to the east directions relative to the well. This pattern is in very good agreement with the saturation estimates provided by the reservoir simulator. Similar spatial variations are observed in Ez amplitudes, while variations in Hz (amplitude or phase) are much smoother and lower-amplitude, though they still correlate well with estimated reservoir-saturation variations. These observations are in agreement with the results of the modeling study on the same well.
Measurement Repeatability. The authors, to this point, have evaluated the sensitivity of the surface-to-borehole CSEM data to the static distribution of reservoir resistivity around the observation well. The next step is to analyze the sensitivity of the Ez measurements to the variations of the waterfront in time (time-lapse analysis). The first objective of the project, in fact, is to detect and map the variations of the waterfront position to enable the use of this information for reservoir-management decisions. Time-lapse analysis is useful for this goal because, given a reliable baseline resistivity distribution, the time-related changes are occurring only in the reservoir and are unbiased by previous assumptions on saturation distributions. The effectiveness of time-lapse observations is related inherently to the amount of signal and noise in the measurements that are also related to the rate of movement of the waterfront. It is therefore critical to analyze the amount of signal change produced in a certain time frame or the time frame necessary to achieve a sufficient signal change above an estimated noise threshold.
Acquisition of a depth profile was repeated with an interval of 4 days from a Tx position 2.3 km from the well. The Ez profiles, in terms of amplitude and phase, indicate overall good repeatability of the measurements, where the upper section of the well performed better (0.3% error in log amplitude and 1.2% error in phase) than the lower interval (0.4% error in log amplitude and 4.7% error in phase), where caliper information indicates the presence of a washout. During the survey, the well fluids were depleting slowly and required repeated pumping operations. Tilting, depth errors, and fluid-composition variations can be modeled to obtain estimates of their effect on the measurements.
The repetition errors in the Ez field are then compared with the signal changes estimated from reservoir simulator modeling. The calculated 2-year and 5-year time-lapse Ez signal is then compared with the amplitude differences from the repeated measurements. The repeatability analysis suggests that the time-lapse measurements should be able to provide the requested information about the waterfront evolution.
With the signal-to-noise and repeatability of the surface-to-borehole measurements assessed, the key steps to derive a robust estimate of 3D resistivity distribution can be identified. Some of the most important aspects in this work flow are modeling the steel casing effect and adopting the strategies to account for it in the data. The work flow is described in the complete paper.
The 3D inversion for the reservoir resistivity distribution is performed using the Ez field data at 8 Hz. A representation of the well casing is incorporated into the model using a finite-difference scheme and material-property-averaging techniques. The inversion was run for a single depth of observation located 70 m below the casing shoe.
Inversion results are consistent with those already observed from spatial variations in the Ez field. Repeating the inversion using a smoothed version of the reservoir simulator resistivity distribution as a starting model further extends the sensitivity of the inversion to approximately 1.8 km from the monitoring well. The consistency of the two inversion results indicates the overall robustness of the inversion procedure, with little dependency on the starting model.
A surface-to-borehole CSEM technology was tested for the first time over a relatively deep reservoir in a large onshore oil field. The pilot survey, with the technological solutions identified and the theoretical framework built around the interpretation of the data, provided very positive indications for the technology to be upscaled to a potential new oilfield service.
For a limited time, the complete paper SPE 191544 is free to SPE members.
Surface-to-Borehole Electromagnetics Hold Promise for 3D Waterflood Monitoring
01 March 2019
Machine Learning Improves Accuracy of Virtual Flowmetering and Back-Allocation
In this study, the authors investigated a fully data-driven approach using artificial neural networks (ANNs) for real-time virtual flowmetering and back-allocation in production wells.
Permanent Fiber-Optic System Monitors Oil-Rim Movement
In this paper, the authors describe a project to design, field trial, and qualify an alternative solution for real-time monitoring of the oil rim in carbonate reservoirs that overcomes these disadvantages.
This industry is one often considered reactive and overly tradition-bound. These new technologies, however—and, more importantly, the drive of these researchers to harness their capabilities—prove that petroleum engineers remain at the forefront of innovation and discovery.
Don't miss out on the latest technology delivered to your email weekly. Sign up for the JPT newsletter. If you are not logged in, you will receive a confirmation email that you will need to click on to confirm you want to receive the newsletter.
11 March 2019
14 March 2019