Low frequency DAS well interference evaluation

Abstract

Methods and systems for assessing cross-well interference and/or optimizing hydrocarbon production from a reservoir by obtaining low frequency DAS and DTS data and pressure data from a monitor well, when both the monitor and production well are shut-in, and then variably opening the production well for production, and detecting the temperature and pressure fluctuations that indication cross-well interference, and localizing the interference along the well length based on the low frequency DAS data. This information can be used to optimize well placement, completion plans, fracturing plans, and ultimately optimize production from a given reservoir.

Claims

1. A system for evaluating cross-well interference in a hydrocarbon reservoir, comprising: a) a hydraulically fractured monitor well in a hydrocarbon reservoir; b) a hydraulically fractured production well fitted for production of hydrocarbons in said reservoir; c) one or more fiber optic cables along a length of said monitor well, said one or more fiber optic cables configured for low frequency distributed acoustic sensing (“DAS”) of <1 Hz and for distributed temperature sensing (“DTS”); d) one or more pressure sensor(s) in said monitor well and in said production well; e) a processor at a surface of said reservoir operably connected to said one or more fiber optic cables and said one or more pressure sensor(s) for analyzing recorded DAS data and DTS data and pressure data.

2. The system of claim 1, where said reservoir is an unconventional reservoir.

3. The system of claim 1, where said monitor well and said production well are horizontal wells.

4. The system of claim 1, where said monitor well and said production well are horizontal wells and said reservoir is an unconventional reservoir.

5. The system of claim 1, where said monitor well and said production well are perforated and cased wells.

6. The system of claim 1, where said monitor well and said production well are perforated, cased and cemented wells.

7. The system of claim 1, where said monitor well and said production well are perforated and cased horizontal wells.

8. The system of claim 1, wherein said one or more fiber optic cables are cemented in behind a casing in said monitor well.

9. The system of claim 1, wherein said one or more fiber optic cables are deployed into said monitor well via wireline, coil tubing, or carbon rod.

10. The system of claim 1, wherein said one or more pressure sensors is a borehole pressure gauge and one or more bottom hole pressure sensors.

11. A system for evaluating cross-well interference in a hydrocarbon reservoir, comprising: a) a hydraulically fractured, cased, cemented and perforated monitor well in a hydrocarbon reservoir; b) a hydraulically fractured, cased, cemented and perforated production well fitted for production of hydrocarbons in said reservoir; c) one or more fiber optic cables cemented behind a casing and along a length of said monitor well, said one or more fiber optic cables configured for low frequency distributed acoustic sensing (“DAS”) of <1 Hz and for distributed temperature sensing (“DTS”); d) one or more pressure sensor(s) in said monitor well and in said production well; and e) a processor at a surface of said reservoir operably connected to said one or more fiber optic cables and said one or more pressure sensor(s) for analyzing recorded DAS data and DTS data and pressure data.

12. The system of claim 11, where said reservoir is an unconventional reservoir.

13. The system of claim 11, where said monitor well and said production well are horizontal wells.

14. The system of claim 11, wherein said one or more pressure sensors is a borehole pressure gauge and one or more bottom hole pressure sensors.

15. The system of claim 11, wherein said production well comprises one or more fiber optic cables cemented behind a casing and along a length of said production well, said one or more fiber optic cables configured for low frequency distributed acoustic sensing (“DAS”) of <1 Hz and for distributed temperature sensing (“DTS”) and wherein said monitor well can function as a production well and said production well can function as a monitor well.

16. A system for evaluating cross-well interference in a hydrocarbon reservoir, comprising: a) a hydraulically fractured, cased, cemented and perforated monitor well in a hydrocarbon reservoir; b) a hydraulically fractured, cased, cemented and perforated production well fitted for production of hydrocarbons in said reservoir; c) one or more fiber optic cables cemented behind a casing and along a length of said monitor well and one or more fiber optic cables cemented behind a casing and along a length of said production well, said one or more fiber optic cables configured for low frequency distributed acoustic sensing (“DAS”) of <1 Hz and for distributed temperature sensing (“DTS”); d) one or more pressure sensor(s) in said monitor well and one or more pressure sensor(s) in said production well; and e) a processor at a surface of said reservoir operably connected to said one or more fiber optic cables and said one or more pressure sensor(s) for analyzing recorded DAS data and DTS data and pressure data.

Description

BRIEF DESCRIPTION OF DRAWINGS

(1) FIG. 1: Conceptual model of cross flows in the monitor well induced by the connection to the operation well.

(2) FIG. 2: DTS temperature measurement in the monitor well during shut-in period. Left: temperature profile. Right: temperature spatial gradient.

(3) FIG. 3: Borehole pressure gauge measurements. Top: Pressure measured in the operation well, with associated well operations. Bottom: Pressure measured in the monitor well.

(4) FIG. 4: Low-frequency DAS response in the monitor well compared with temperature spatial gradient and borehole pressure during the operation well choke changes. Colormap in the background is the DAS signal, vertical black curve is the temperature gradient profile measured by DTS (FIG. 2), and horizontal dashed line is the borehole pressure in the monitor well (FIG. 3).

(5) FIG. 5: Comparison between the raw DAS data and the approximation using EQ 4. Vertical solid line and horizontal dashed line in the middle panel show the eigenvectors u.sub.1(x) and v.sub.1(t).

(6) FIG. 6: Comparison between the spatial gradient of temperature dT/dx measured by DTS and the spatial eigenvector u.sub.1(x) estimated from DAS data.

(7) FIG. 7: Inversion results for the cross flow spatial variation. a) time-shifted u.sub.1 and the model prediction αRdT/dx. b) inverted αR(x) and the control points c. c) negative spatial gradient of R(x), with positive value indicates outflow when the operation well opens.

(8) FIG. 8: Same as FIG. 7, except the operations of the wells are switched.

(9) FIG. 9: Spatial distribution of the outflow in FIG. 7c and FIG. 8c.

(10) FIG. 10: The comparison between the data from a DAS channel and the co-located borehole temperature gauge. The gauge data is differentiated in time to obtain the temperature gradient.

(11) FIG. 11: Calculated cross-flow velocity in the monitor well.

DETAILED DESCRIPTION

(12) Herein, we use the data from two adjacent hydraulically fractured horizontal production wells. However, a similar procedure can be used for other kinds of wells.

(13) Because DAS is a strain rate sensor and the fiber is mechanically coupled with the formation, strain from the minute temperature variations caused by interference can be detected. The DAS data are recorded at an offset monitor well during production of an adjacent well. The fiber-optic cables are preferably installed outside the casing and cemented in place. The raw data are sampled at 10 kHz continuously at more than 6000 locations along the wellbore, with 1 m spatial sampling and 5 m gauge length. The recorded optical phase is differentiated in time, hence the DAS data are linearly correlated with the strain rate along the fiber.

(14) The raw DAS data are down-sampled to 1 s after a low-pass anti-aliasing filter (0-0.5 Hz) is applied. The data are then median filtered to remove any spiky noise. Another low-pass filter with a corner frequency of 0.05 Hz is then applied. A DC drift with an amplitude around 0.1 rad/s is removed from the data as well. The DC drift was channel invariant and does not vary significantly with time. The drift noise is most likely associated with interrogator noise. We estimated the DC drift by calculating the median value of the channels that were out of the zone of interest at each time interval. Compared to the industry standard waterfall visualizations, the low-frequency processing not only increased the signal-to-noise ratio of the signal, but also preserved the strain rate polarity, which is important for our interpretations. The strain change recorded by DAS at this frequency band can be caused by thermal variation and/or mechanic strain perturbation.

(15) The DTS data are recorded at the same monitor well as the DAS data. The DTS data can be recorded during or before the DAS data recording. The raw data are sample at 5 minutes continuously with 1 ft spatial resolution. The data are averaged for several hours to obtain a reliable borehole temperature profile during shut-in. The recorded DTS data are calibrated to remove the attenuation induced measurement error.

(16) Because the wells are hydraulically fractured, the uneven completion at each perforation induces a thermal spatial gradient during the shut-in period. FIG. 2 shows an example of the temperature profile in a monitor well after a 24-hour shut-in, measured by DTS. The heel-most perforation in this well is located around 13000 ft, where the temperature drops dramatically. Spatial temperature gradients around 10.sup.−3−10.sup.−2 F/ft can be observed in the stimulated section (13000-16500 ft), which is important to create the signals required for this method.

(17) After the operation well is opened, the borehole pressure drops due to the production. This pressure perturbation propagates away from the operation well through the conductive fracture network. If the monitor well and the operation well are interconnected by fractures, the pressure in the monitor well will also be perturbed, whereas the pressure would otherwise not change. These pressure changes will cause flow from the monitor well towards the lower pressure zone near the production well, and that can be detected by temperature changes causes by the flow.

(18) FIG. 3 shows the pressure response in both wells due to a series of choke-size changes in the operation well. In this example, the well spacing was around 700 ft. Pressure was measured using bottom hole sensors. The pressure perturbation near the monitor well was not uniform because the conductivity of the fractures was spatially heterogeneous. The highly connected fractures had lower pressure than the less connected fractures. The spatial gradient of pressure along the monitor well induced cross flows in the monitor well borehole, with the fluid flowing from the weakly connected fractures towards the highly connected fractures.

(19) Due to the spatial gradient of temperature in the monitor well (FIG. 2), the cross flows produce small temperature perturbations, which can be approximated as:

(20) $\begin{array}{cc}\frac{\mathrm{dT}}{\mathrm{dt}}=-v\frac{\mathrm{dT}}{\mathrm{dx}},& \left(1\right)\end{array}$

(21) where v is the cross-flow velocity, T is the monitor well borehole temperature, t=time, and x is distance or position. We only consider the convection induced temperature perturbation, while ignoring the temperature mixing due to the reservoir fluid entering the borehole through perforations. We also ignore thermal conduction from surrounding formations. This assumption significantly simplifies the data analysis, and captures the majority of the signal amplitude.

(22) Because the DAS signal at the ultra low-frequency band (<0.1 Hz) is sensitive to temperature variations as small as 10.sup.−5° F., it can be used to measure the cross-flow induced temperature perturbations. FIG. 4 shows the DAS response at the monitor well during a series of choke changes in the operation well, compared with the spatial gradient of temperature measured by DTS (FIG. 2) and borehole pressure measured by a pressure gauge (FIG. 3). The DAS response is highly correlated with the pressure changes in the temporal domain, and with spatial gradient of temperature in the spatial domain. The DAS response is interpreted as small thermal perturbations due to the cross flows between the monitor well perforations. The cross flows are caused by the spatial heterogeneity of connectivity between the operation well and the monitor well.

(23) The thermal perturbation measured by DAS is mainly controlled by EQ 1, which can be rewritten as:

(24) $\begin{array}{cc}D\left(x,t\right)=-\lambda v\left(x,t\right)\frac{\mathrm{dT}}{\mathrm{dx}}\left(x\right),& \left(2\right)\end{array}$
where D is the low-frequency DAS signal, and λ is a constant that converts optical phase measured by DAS into temporal gradient of temperature. If we assume the connectivity does not change during the period of data acquisition, we can further simplify the signal as:

(25) $\begin{array}{cc}D\left(x,t\right)=-\lambda V\left(t\right)R\left(x\right)\frac{\mathrm{dT}}{\mathrm{dx}}\left(x\right)=A\left(t\right)B\left(x\right),& \left(3\right)\end{array}$
where V(t) and R(x) describe how the magnitude of the cross-flow velocity changes with time and space, respectively. From this equation, we can see that the DAS signal can be approximated by the product of two one-dimensional, separable functions (A and B) that describe the variation in time and space respectively.

(26) The A(t) and B(x) can be obtained by applying singular-value decomposition (SVD) on the DAS data. The SVD operation decompose the DAS data D(x; t) into the summation of a series production of eigenvectors and eigenvalues:

(27) $\begin{array}{cc}D\left(x,t\right)=\underset{i}{.Math.}{u}_i\left(x\right){\sigma }_i{v}_i\left(t\right)\approx u_1\left(x\right){\sigma }_1{v}_1\left(t\right),& \left(4\right)\end{array}$
where u.sub.i and v.sub.i are the left and right eigenvectors, and σi is the eigenvalue. The eigenvalues are sorted in descending order. It is worth mentioning that u.sub.i is a column vector while v.sub.i is a row vector, and the outer product of the two is a 2D matrix. Based on EQ 3, we can use the first (largest) eigenvalue and its corresponding eigenvectors to approximate the signal.

(28) Extra processing steps may be considered to acquire better u.sub.1 and v.sub.1 estimation. For example, u.sub.1 and v.sub.1 can be calculated independently using different section of the data. For the DAS data in FIG. 4, u.sub.1(x) is evaluated using only the data from 1.5-4.5 hours, where the signal is strongest and crossflow has subsided.

(29) On the other hand, only the data from measured depth (MD) 13500 ft and beyond is used to evaluate v.sub.1(t) in order to avoid the effect of the large un-related signal around 13200 ft. u.sub.1 and v.sub.1 are then low-pass filtered to reduce the noise.

(30) A comparison between the original DAS data and the approximation using the first (largest) eigenvalue σ1 and corresponding eigenvectors u.sub.1 and v.sub.1 (EQ 4) is shown in FIG. 5. This operation preserves the majority of the signal amplitude, while dramatically reduces the noise.

(31) More importantly, it decomposes the DAS signal into two separate 1-D functions that describe the temporal and spatial variations separately. In this method we assume the communication does not change within the measurement period, which is usually only a few hours.

(32) Substituting EQ 4 into 3 results in:

(33) $\begin{array}{cc}\frac{1}{\alpha }{u}_1\left(x\right)=-R\left(x\right)\frac{\mathrm{dT}}{\mathrm{dx}}\left(x\right){\mathrm{\alpha \sigma }}_1{v}_1\left(t\right)=\lambda V\left(t\right),& \left(5\right)\end{array}$
where α is a scaling constant.

(34) FIG. 6 shows the comparison between the spatial gradient of temperature dt/dx measured by DTS and the eigenvector u.sub.1(x) estimated from DAS data. It is clear that parts of these two curves are correlated, while the other parts are anti-correlated. This is due to the different sign of αR(x), which indicates the direction of the cross flows changes along the wellbore, like the one shown in FIG. 1. It is also noticeable that there is a small shift between these two curves, especially around 14000-15000 ft. This spatial shift is due to the small moveout in the signal due to the convection, which can be easily removed by dynamic warping or other time-shift corrections.

(35) αR(x) can be inverted by minimizing the misfit between u.sub.1(x) and αR(x) dt/dx, which can be achieved by a least-square inversion minimizing the penalty function:

(36) $\begin{array}{cc}{ϵ}^2=\int {\left(u_1\left(x\right)+\alpha R\left(x\right)\frac{\mathrm{dT}}{\mathrm{dx}}\right)}^2.& \left(6\right)\end{array}$

(37) This inversion can be further stabilized by reducing degrees of freedom for αR(x). Herein we use piecewise cubic interpolation with ten evenly spaced control points, which can be performed by matrix operations:
c=(G.sup.TG).sup.−1G.sup.Tu.sub.1,  (7)
where coefficient matrix G=T.sub.xM. T.sub.x is a diagonal matrix with the diagonal elements equal to dT/dx, and M is the interpolation matrix. c is the value at the control points.

(38) FIG. 7 shows the results of the least-square inversion. Positive value of R(x) indicates toe-ward cross flows. The spatial gradient of R(x) indicates inflow/outflow at each section. As demonstrated in FIG. 1, the well sections with stronger connections are associated with outflows (fluid flows from wellbore into formation) in the monitor well when the operation well opens. In this case, there are three zones that indicate stronger connections, by ignoring the outflow at the heel (12500 ft) which is probably due to the edge effect of the cubical interpolation.

(39) The connection between the wells should be bidirectional, which means that similar outflow locations should be observed if the operations of the wells are switched. FIG. 8 shows the result of the same inversion, except monitor and operation wells are interchanged. Three similar outflow zones can be clearly observed in FIG. 8c, although the DAS response in FIG. 8a is very different from that in FIG. 7a.

(40) FIG. 9 shows the spatial distribution of these outflow zones in both wells by plotting them along the well paths. These outflow zones indicate the locations of stronger connections between the two wells, which is consistent with the regional maximum stress direction, as well as the cross-well fracture hits detected during completion, using the method described in (Jin & Roy, 2017).

(41) The connectivity between the wells can be further quantified by acquiring cross-flow velocity. This estimation requires knowing the scaling factor λ between the DAS optical phase measurement and the temporal gradient of temperature. λ can be estimated by comparing the DAS response with the co-located temperature gauge or DTS data. FIG. 10 shows the data comparison between a DAS channel in the monitor well and the co-located borehole temperature gauge. The two signals are linearly correlated, and λ can be easily estimated by a linear regression. In the case of evaluating λ using DTS, the workflow described in Jin et al. (2017b) can be referred.

(42) After λ is known, 1/α V (t) can be easily obtained by:

(43) $\begin{array}{cc}\frac{1}{\alpha }V\left(t\right)=\frac{{\sigma }_1{v}_1}{\lambda },& \left(8\right)\end{array}$
which can then by multiplied by previously calculated R(x) to get the cross-flow velocity v(x; t)=R(x)V (t). FIG. 11 shows the calculated cross-flow velocity. Velocities as slow as 1 ft/hour can be detected using this method. If the radius of the borehole casing is known, the volume rate of the outflow can also be calculated.

(44) The analysis described herein provides a means to measure the spatial variation of inter-well connectivity during the production stage. The demonstrated example is from two nearby hydraulically fractured production wells. However, the method can be applied to any wells that have a spatial gradient of temperature during a shut-in period.

(45) It is worth emphasizing that the outflow zone locations shown in FIG. 9 are not the only connected locations between the wells, but the locations with stronger connectivity. Therefore, this method measures well connectivity in relative terms. By combining the borehole pressure measurement (FIG. 3), it is possible to constrain the fracture conductivity using reservoir models. This method provides the spatial information of well interference that no other method provides, which is valuable for completion and well spacing optimization.

(46) This method can be applied on either temporarily deployed or permanently installed fiber cables. However, if the fiber used for DAS measurement is installed behind production casing and cemented in place, the heat conduction effect should be corrected when calculating the V(t), since the temporal variation of temperature can be delayed and attenuated as it propagates from the borehole, through the casing and cement, and into the fiber. The correction can be applied by solving a 1D radial diffusion equation, and is described in Kreuger (2017). A similar correction should be applied if the fiber is temporarily deployed through coil tubing or wirelines with large radius, where the thermal conductivity effect is not negligible.

(47) The following references are each incorporated by reference in its entirety for all purposes:

(48) Awada, A., et al. (2016). Is that interference? A work flow for identifying and analyzing communication through hydraulic fractures in a multiwell pad. SPE Journal, 21 (05), 1-554.

(49) Jin, G., & Roy, B. (2017). Hydraulic-fracture geometry characterization using low-frequency DAS signal. The Leading Edge, 36 (12), 975-980.

(50) Le Calvez, J. H., et al. (2007). Real-time microseismic monitoring of hydraulic fracture treatment: a tool to improve completion and reservoir management. AAPG Search and Discovery Article #90171 CSPG/CSEG/CWLS GeoConvention 2009, Calgary, Alberta, Canada, May 4-8, 2009

(51) SPE-140561-MS (2011) Molenaar M., et al., First Downhole Application of Distributed Acoustic Sensing (DAS) for Hydraulic Fracturing Monitoring and Diagnostics.

(52) SPE-149602 (2012) Johannessen K., et al., Distributed Acoustic Sensing—a new way of listening to your well/reservoir

(53) SPE-173640-MS—Grayson, S., et al. (2015). Monitoring acid stimulation treatments in naturally fractured reservoirs with slickline distributed temperature sensing.

(54) SPE-179149-MS—Wheaton, B., et al. (2016). A case study of completion effectiveness in the eagle ford shale using DAS/DTS observations and hydraulic fracture modeling.

(55) SPE-186091-PA—Wu, K., et al. (2017). Mechanism analysis of well interference in unconventional reservoirs: Insights from fracture-geometry simulation between two horizontal wells.

(56) SPE-90541-MS—Ouyang, L.-B., et al. (2006). Flow profiling via distributed temperature sensor (DTS) system-expectation and reality.

(57) URTEC-1581750-MS—Portis, D. H., et al. (2013). Searching for the optimal well spacing in the eagle ford shale: A practical tool-kit.

(58) US20150146759 Temperature sensing using distributed acoustic sensing.

(59) US20170260839 Hydraulic fracture monitoring by low-frequency DAS

(60) US20170260842 Low frequency distributed acoustic sensing

(61) US20170260846 Measuring downhole temperature by combining DAS/DTS data

(62) US20170260849 DAS method of estimating fluid distribution

(63) US20170260854 Hydraulic fracture monitoring by low-frequency DAS.

(64) US20170342814 LOW-FREQUENCY DAS SNR IMPROVEMENT

(65) US2018045040 Production Logs From Distributed Acoustic Sensors (42437).

(66) US8505625 Controlling well operations based on monitored parameters of cement health.