Fracture identification method based on reservoir high-frequency pressure signals

Abstract

The application relates to the technical field of oil and gas exploitation, in particular to a fracture identification method based on reservoir high-frequency pressure signals, which comprises the following steps: collecting high-frequency pressure sensor data of each group of a fracturing section in a pump-off seepage stage; determining first-order difference characteristics of each frame of pressure fluctuation and damping attenuation fitting characteristics of a pressure envelope line, and constructing a first index reflecting the influence degree of a single group of data on water hammer effect; analyzing the correlation between different groups of high-frequency pressure data sequences, and constructing a second index reflecting the spatial consistency of pressure responses at different positions; determining a deviation index of each group of high-frequency pressure data, and performing weighted fusion on each group of high-frequency pressure data to obtain a comprehensive pressure data sequence; performing pressure drop fitting analysis based on the comprehensive pressure data sequence, determining the filtration type according to fitting residuals, and combining a fracture expansion model to obtain fracture morphology parameters. The application improves the accuracy of fracture filtration type identification and morphology parameter inversion.

Description

Technical Field

[0001] This application relates to the field of oil and gas extraction technology, specifically to a fracture identification method based on high-frequency pressure signals in reservoirs. Background Technology

[0002] Horizontal well staged hydraulic fracturing technology is a core method for developing unconventional oil and gas resources such as shale gas and tight oil. Currently, the fracturing effect is usually accurately evaluated by monitoring and analyzing the pressure response data during the fracturing process, and then using a fracture propagation model to invert the fracture morphology parameters (such as height, width, and SRV volume).

[0003] The complete fracturing process includes the initiation stage, the propagation stage, and the pump shutdown and seepage stage. Among these, the pressure data from the pump shutdown and seepage stage reflects the actual loss of fracturing fluid to the formation after fracture formation and is the most critical data source for modeling and inverting fracture morphology parameters. However, in actual operations, the moment the pump is shut down during fracturing, due to valve closure, sudden changes in fracturing fluid flow, or dynamic fracture closure, a strong water hammer effect can easily be triggered. This water hammer effect generates damped oscillating waves with high-frequency reflection and propagation characteristics in the wellbore and complex fracture network. This oscillating signal is directly superimposed on the actual seepage pressure response reflecting formation loss, resulting in severe distortion of the acquired high-frequency pressure drop signal. This signal distortion greatly interferes with subsequent pressure drop curve fitting and fracture modeling and inversion processes, leading to misjudgment of fracture loss type and ultimately severely affecting the accuracy of fracture morphology parameter calculations. Summary of the Invention

[0004] To address the aforementioned technical problems, this application provides a fracture identification method based on high-frequency reservoir pressure signals to solve the existing issues.

[0005] The fracture identification method based on high-frequency reservoir pressure signals in this application adopts the following technical solution:

[0006] One embodiment of this application provides a fracture identification method based on high-frequency pressure signals in reservoirs, the method comprising the following steps:

[0007] High-frequency pressure sensor data were collected from each group during the pump shutdown and seepage stage of the fracturing section to obtain a high-frequency pressure data sequence;

[0008] Each high-frequency pressure data sequence is processed by frame segmentation to determine the first-order difference feature of pressure fluctuation in each frame and the damping attenuation fitting feature of pressure envelope. Based on the first-order difference feature and the damping attenuation fitting feature, a first index reflecting the degree of influence of water hammer effect on a single set of data is constructed.

[0009] The correlation between different groups of high-frequency pressure data sequences was analyzed, and a second index reflecting the spatial consistency of pressure response at different locations was constructed based on the correlation.

[0010] The deviation index of each group of high-frequency pressure data is determined based on the first index and the second index, and the high-frequency pressure data of each group is weighted and fused based on the deviation index to obtain a comprehensive pressure data sequence.

[0011] Pressure drop fitting analysis was performed based on the comprehensive pressure data sequence. The filtering type was determined according to the fitting residual, and the crack morphology parameters were obtained by inversion using the crack propagation model.

[0012] In one embodiment, the first-order difference feature is specifically:

[0013] Perform a first-order difference operation on the same frame of high-frequency pressure data, calculate the mean of the absolute values ​​of all first-order difference values ​​corresponding to the same frame of high-frequency pressure data, and denote it as the first parameter;

[0014] The second parameter of each high-frequency pressure data sequence is obtained by analyzing the numerical distribution characteristics of the first parameter corresponding to all frames of high-frequency pressure data in each high-frequency pressure data sequence, and is used to evaluate the first-order difference feature.

[0015] In one embodiment, the second parameter is the range of the first parameter corresponding to all frames of high-frequency pressure data in each high-frequency pressure data sequence.

[0016] In one embodiment, the damping attenuation fitting characteristic of the pressure envelope is specifically as follows:

[0017] Extract the local maxima from each frame of high-frequency pressure data, and perform a logarithmic operation on each local maxima with the natural constant as the base; perform linear fitting on the logarithmic operation results of all local maxima in each frame of high-frequency pressure data, and calculate the degree of fit of the fitted line;

[0018] The fitting degree corresponding to all frames of high-frequency pressure data in each high-frequency pressure data sequence is positively fused to obtain a third parameter, which is used to measure the damping attenuation fitting characteristics of the pressure envelope.

[0019] In one embodiment, the process of determining the first index is as follows:

[0020] The second parameter is normalized, and the mean of the normalization result and the third parameter is used as the first index.

[0021] In one embodiment, the process of determining the second index is as follows:

[0022] Calculate the correlation between any set of high-frequency pressure data sequences and the remaining sets of high-frequency pressure data sequences, calculate the mean of the absolute values ​​of all the correlations corresponding to the any set of high-frequency pressure data sequences, and the second index is negatively correlated with the mean.

[0023] In one embodiment, the deviation index is the normalized result of the mean of the first index and the second index.

[0024] In one embodiment, the weights for weighted fusion of the high-frequency pressure data sets are:

[0025] Calculate the difference between the natural number 1 and the deviation index of each group of high-frequency pressure data, and use the proportion of the difference of each group of high-frequency pressure data in the sum of the differences of all groups of high-frequency pressure data as the weight.

[0026] In one embodiment, determining the filtering type based on the fitting residual includes:

[0027] Linear fitting is performed on the comprehensive pressure data sequence. If the mean of the fitting residuals of all data points is less than a preset threshold, the pressure drop curve corresponding to the comprehensive pressure data sequence is determined to be of the linear filtering type; otherwise, the pressure drop curve corresponding to the comprehensive pressure data sequence is determined to be of the nonlinear filtering type.

[0028] In one embodiment, when the pressure drop curve is of the linear filtration type, the filtration coefficient of the crack is calculated by combining the crack propagation model, and the crack morphology parameters are obtained by inversion; when the pressure drop curve is of the nonlinear filtration type, the pressure drop curve is segmented by the second derivative of each sampling point on the pressure drop curve, the filtration coefficient of each segment is calculated separately, and the comprehensive filtration coefficient is obtained after forward fusion, and the crack morphology parameters are obtained by inversion through the crack propagation model.

[0029] This application has at least the following beneficial effects:

[0030] This application addresses the problem of high-frequency pressure signal distortion caused by water hammer during the seepage stage of horizontal well fracturing and pump shutdown, which severely affects the accuracy of fracture modeling and inversion. By performing short-time frame processing on the high-frequency pressure data, the application accurately extracts the first-order difference feature and the damping attenuation fitting feature of the pressure envelope. This mechanism fully aligns with the physical essence of the water hammer effect—"short-time high-frequency transients" and "exponential damping attenuation"—and can accurately isolate abnormal fluctuations under a stable seepage background. Thus, a first index is used to precisely quantify the signal distortion degree of a single measuring point under local water hammer impact, achieving high-precision quantification of water hammer interference at the time domain. Considering the strong spatial consistency of actual fracturing fluid seepage within the wellbore, while water hammer reflection impact has local randomness and dissipative characteristics in complex fracture networks, this application constructs a second index by calculating the correlation between data from different measuring points. This effectively identifies and quantifies spatially isolated reflected wave interference, overcoming the limitations of single-point analysis and achieving effective identification of local abnormal distortions at the spatial level. Finally, the first and second indices are fused to construct a deviation index, which serves as the basis for weighted fusion. Data severely affected by water hammer and exhibiting high distortion are automatically assigned low weights, while stable and accurate seepage data are assigned high weights. This dynamic weighting mechanism eliminates signal distortion caused by the superposition of water hammer effects at local locations, obtaining a high-fidelity, pure comprehensive pressure data sequence that greatly restores the true seepage pressure drop signal. Based on the high-fidelity comprehensive pressure drop data that has removed water hammer noise interference, residual fitting analysis can extremely sensitively and accurately identify the true fracture filtration type, thus providing absolutely reliable data input for subsequent fracture propagation models. This significantly improves the inversion accuracy of morphological parameters such as fracture height, width, and stimulation volume, providing strong technical support for the scientific evaluation of fracturing effects in unconventional oil and gas resources, and significantly enhancing the accuracy of fracture filtration type identification and morphological parameter inversion. Attached Figure Description

[0031] To more clearly illustrate the technical solutions and advantages in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0032] Figure 1 A flowchart illustrating the steps of the fracture identification method based on high-frequency reservoir pressure signals provided in this application. Detailed Implementation

[0033] To further illustrate the technical means and effects adopted by this application to achieve the intended purpose of the invention, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of the fracture identification method based on high-frequency reservoir pressure signals proposed in this application. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0034] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0035] The following description, in conjunction with the accompanying drawings, details the specific scheme of the fracture identification method based on high-frequency reservoir pressure signals provided in this application.

[0036] This application provides a fracture identification method based on high-frequency reservoir pressure signals in one embodiment. Specifically, the following fracture identification method based on high-frequency reservoir pressure signals is provided. Please refer to [link to relevant documentation]. Figure 1 The method includes the following steps:

[0037] Step S001: Collect high-frequency pressure sensor data from each group during the pump shutdown and seepage stage of the fracturing section to obtain a high-frequency pressure data sequence.

[0038] A high-frequency pressure sensor probe is lowered into the horizontal well after perforation and positioned at the current fracturing stage to collect high-frequency pressure data of that fracturing stage during the hydraulic fracturing process.

[0039] N high-frequency pressure sensors are installed at equal intervals along the axial direction on the high-frequency pressure sensor probe. The distance between adjacent sensors is d, in meters. N is an odd number in the range [3, 11], and d is in the range [1, 5] meters. The specific values ​​of N and d can be set by the implementer according to the actual fracturing construction process requirements. In this embodiment, N is 7 and d is 2.

[0040] When placing the high-frequency pressure sensor probe, the first of the N high-frequency pressure sensors... Each sensor is aligned with the center of the perforation location in the current fracturing section, wherein... The up-rounding symbol is used to ensure that the sensor array symmetrically covers the fracturing area. The data sampling frequency of all high-frequency pressure sensors is set to 1kHz, but the implementer can adjust it according to the actual situation.

[0041] A complete fracturing operation consists of three stages: fracture initiation, propagation, and pump shutdown / seepage. After the fracture has propagated and fracturing fluid injection has ceased, the pump shutdown / seepage stage begins. The pressure drop response signal generated in this stage directly reflects the seepage pattern of the fracturing fluid in the reservoir and is the core basis for inverting fracture geometry and formation parameters.

[0042] However, at the moment the pump stops, the interaction between the inertia of the fracturing fluid and the complex geometry within the fracture triggers a significant water hammer effect. This high-frequency shock wave, with its reflection and propagation characteristics, superimposed on the normal seepage pressure drop signal causes the acquired pressure data to exhibit strong non-stationary oscillations. Because the fracture morphology is unknown at this time and the water hammer impact is highly random, this signal distortion masks the true seepage trend, making it difficult to accurately identify the true morphological parameters of the fracture during subsequent modeling and inversion.

[0043] Based on the above analysis, the high-frequency pressure data acquired by each high-frequency pressure sensor are extracted according to the starting time of the seepage stage after pump shutdown, resulting in a high-frequency pressure data sequence for the initial seepage stage after pump shutdown, denoted as follows: , , Where N represents the total number of high-frequency pressure sensors, and T represents the total number of sampling points corresponding to the initial stage of pump shutdown. The specific data interception duration is set by the implementer according to the actual observation needs of the pump shutdown seepage process. In this embodiment, the data for the first 3 minutes after pump shutdown is intercepted. This duration covers the complete fluctuation cycle of the water hammer effect from its occurrence to the main energy dissipation, and can effectively extract the complete pressure response reflecting the characteristics of water hammer impact.

[0044] Step S002: Perform frame-by-frame processing on each group of high-frequency pressure data sequences, determine the first-order difference features of pressure fluctuations in each frame and the damping attenuation fitting features of the pressure envelope, and construct a first index reflecting the degree of influence of water hammer effect on a single group of data based on the first-order difference features and the damping attenuation fitting features.

[0045] During the pump shutdown and seepage phase, as fracturing fluid is continuously lost to the formation, the downhole pressure response ideally exhibits a continuous and gradually decreasing trend. Even under the influence of minor fluid disturbances within the wellbore, the pressure curve shows only slight fluctuations during the descent, maintaining an overall stable and gently changing characteristic. Conversely, the water hammer effect is an impact phenomenon caused by the violent interaction between the kinetic inertia of high-speed fracturing fluid and the complex fracture geometry, and its pressure response exhibits typical high-frequency damped oscillation characteristics. Due to the large spatial scale and complex structure of the fracture network formed by fracturing, the manifestation of water hammer impact at different locations within the fracture exhibits extremely high uncertainty. When this localized damped oscillation impact is superimposed on normal high-frequency seepage pressure data, it inevitably leads to severe distortion and falsification of the local signal.

[0046] For the nth set of high-frequency pressure data sequences Considering the long duration of the pump shutdown seepage process, while the transient impact caused by water hammer typically only occurs within seconds, direct analysis of global data would easily mask the weak, high-frequency transient fluctuations as they are obscured by the long-term, gradual trend. Therefore, this embodiment focuses on... Short-time frame processing is performed to accurately analyze pressure fluctuation characteristics within a local time window. The duration of each frame is set to L seconds. To avoid the water hammer oscillation characteristics being diluted due to an excessively long time window, the value of L is set to the range of [0.1, 2]. The specific value can be set by the implementer according to the actual scenario and data sampling rate. In this embodiment, to balance the capture accuracy of transient features with the sample size, the value of L is set to 0.5.

[0047] right After frame segmentation, a total of [number] were obtained. Frame high-frequency pressure data, the first Frame high-frequency pressure data is denoted as First, calculate The first-order backward difference of the pressure amplitude at each sampling point is used. The first-order backward difference represents the magnitude of pressure attenuation between adjacent sampling times. Then, the following calculations are performed. The mean of the absolute values ​​of the first-order backward differences of the pressure amplitude at all sampling points is used as the first parameter. The first parameter represents the net change trend of pressure decay in this frame of data, that is, the average change of pressure over the duration of one frame.

[0048] Further calculations The first parameter of each frame of high-frequency pressure data, The first parameter set of all frames of high-frequency pressure data is represented as: ,So The second parameter The calculation method is as follows: ,in, , They represent respectively to The first parameter of the frame high-frequency pressure data takes the maximum and minimum values, i.e., the calculation... The range of the first parameter of the frame high-frequency stress data.

[0049] The second parameter characterizes The time-domain dispersion of pressure fluctuation amplitude at the corresponding measuring point during the observation period. When a local frame is subjected to a transient impact from the water hammer effect, the high-frequency oscillations within it will cause a sharp increase in the average pressure change amplitude (i.e., the first parameter) of that frame. Due to the strong non-stationarity and damping decay characteristics of water hammer impacts, the impact intensity varies greatly across different time frames. For example, the initial frame of the impact fluctuates violently, while the later frame tends to be flat, thus significantly increasing the range of the first parameter across frames. Therefore, a larger second parameter indicates a stronger non-stationar disturbance from the water hammer effect at the measuring point during the seepage stage; conversely, a smaller second parameter indicates a weaker second parameter. Unaffected by the water hammer effect, the pressure response only shows a stable seepage pressure drop trend, with minimal and consistent local fluctuations in each frame. At this time, the second parameter will be small or even close to 0, indicating that the high-frequency pressure data has not undergone serious distortion and can truly reflect the seepage characteristics of the crack.

[0050] Furthermore, the transient impact caused by the water hammer effect exhibits a typical damped oscillation mode, with its amplitude envelope conforming to an exponentially decaying function. ,in, The initial amplitude is denoted as 'e', ​​and 'e' is the natural constant. The attenuation coefficient is... This represents the independent variable. In contrast, the oscillations generated by fluid disturbances in fracturing fluid under seepage conditions typically exhibit relatively constant, stable fluctuations. Therefore, by analyzing whether the change pattern of the pressure curve envelope within a single frame of data deviates from a stationary state and conforms to exponential decay characteristics, the degree of interference from the water hammer effect in that frame of data can be accurately identified and quantified.

[0051] Specifically, firstly, the AMPD peak detection algorithm is used to identify... The local maxima in the data are arranged in chronological order. The upper envelope sequence. Secondly, for... The data in the upper envelope sequence are logarithmically calculated, with the natural constant e as the base and the data in the upper envelope sequence as the arguments. Then, the least squares method is used to perform linear fitting on the logarithmically calculated upper envelope sequence to obtain a fitted straight line. Analyze the fitted straight line The degree of fit is determined in this embodiment by calculating the fitted straight line. coefficient of determination The result serves as a measure of the degree of fit. Among them, the AMPD peak detection algorithm and the least squares method are existing technologies, and their specific processes will not be elaborated here.

[0052] Coefficient of determination This reflects the degree to which the fitted line interprets the actual observed data. The closer its value is to 1, the higher the goodness of fit between the actual data and the fitted line. In this embodiment, due to the damping attenuation function... After taking the logarithm, it exhibits a linear characteristic; therefore, the coefficient of determination... The larger the coefficient of determination, the higher the overlap between the upper envelope sequence and the fitted straight line in the logarithmic domain. This means that within the original pressure signal space, the pressure envelope of this frame closely matches the exponential damping decay mode, thus indicating that the data in this frame has been subjected to strong water hammer oscillation impact; conversely, the smaller the coefficient of determination, the lower the coefficient of determination. The smaller the value, the more it indicates that the distribution of pressure fluctuations deviates from the damping attenuation mode, meaning that the data frame mainly consists of stable seepage pressure drop and is not significantly affected by water hammer effects.

[0053] Further calculations The coefficient of determination of the fitted straight line corresponding to each frame of high-frequency pressure data is used to... The determination coefficients of the fitted straight lines corresponding to all frames of high-frequency pressure data are positively fused to obtain a third parameter, which is used to measure the damping attenuation fitting characteristics of the pressure envelope. Here, positive fusion means combining multiple variables, which can be calculated using methods such as addition, averaging, or weighted summation. In this embodiment, The third parameter The expression is: ; third parameter The larger the value, the more severe the impact of water hammer on the seepage pressure drop data; conversely, the smaller the value, the less the seepage pressure drop data is affected by the water hammer effect.

[0054] Based on the above analysis, the calculation The second parameter of the high-frequency pressure data sequence is normalized using the maximum-minimum normalization method to obtain the normalized result of each second parameter. The mean of the normalized result of the second parameter of each high-frequency pressure data sequence and the third parameter are used as the first index of each high-frequency pressure data sequence.

[0055] Step S003: Analyze the correlation between different groups of high-frequency pressure data sequences, and construct a second index based on the correlation to reflect the spatial consistency of pressure response at different locations.

[0056] During the pump shutdown seepage phase, the normal pressure response within the wellbore is primarily caused by the loss of fracturing fluid to the formation. This seepage process is an approximately static diffusion behavior, with a relatively uniform spatial evolution of the pressure field. Because the high-frequency pressure sensor array is distributed along the wellbore with small spacing, the pressure changes caused by seepage respond almost synchronously at each measuring point (propagation delay is negligible). Therefore, the seepage pressure curves obtained from different sensors should exhibit extremely high spatial consistency in terms of shape and trend.

[0057] In contrast, water hammer is a high-frequency shock caused by the obstruction of fracturing fluid movement within a complex fracture network. When this shock wave propagates in the coupling medium between the fracture and the formation, it experiences severe interface reflection and energy scattering due to the variable fracture aperture and complex path, while also being strongly absorbed by the fluid viscosity and the formation medium. This intense energy dissipation and complex wavefield interference result in a highly non-uniform spatial distribution of water hammer impacts. Therefore, the water hammer effect does not exert the same interference on all sensors, and the transient impact intensity experienced by pressure sensors at different locations will vary significantly.

[0058] Based on the above analysis, this embodiment calculates a second index by analyzing the correlation between different groups of high-frequency pressure data sequences, reflecting the degree of influence of water hammer effect on local data. Specifically, The second index The expression is: ,in, The function representing the calculation of the Pearson correlation coefficient. This represents the m-th high-frequency pressure data sequence. It should be noted that the correlation calculation in this embodiment uses the Pearson correlation coefficient; however, implementers may use other existing correlation calculation methods, such as cosine similarity.

[0059] The Pearson correlation coefficient can be used to quantify the waveform similarity between time series. In this embodiment, the spatial consistency of the pressure response at each measuring point is characterized by calculating the absolute value of the correlation coefficient between the nth high-frequency pressure data series and high-frequency pressure data measured by other sensors. The second exponent calculated based on the correlation coefficient... When it is smaller, it indicates The pressure curves at other measuring points show a high degree of consistency, reflecting that the signal in this area has not been disrupted by local transient shocks and is mainly dominated by uniformly diffused seepage pressure; conversely, if the second index... Larger, indicating The signal waveform deviates significantly from that of other measuring points, revealing that the sensor was subjected to a strong and isolated water hammer impact, resulting in severe waveform distortion and falsification of the local seepage pressure data.

[0060] Step S004: Determine the deviation index of each group of high-frequency pressure data based on the first index and the second index, and perform weighted fusion of each group of high-frequency pressure data based on the deviation index to obtain a comprehensive pressure data sequence.

[0061] The first index and the second index are positively fused to obtain the deviation index of each group of high-frequency pressure data sequences. In this embodiment, the mean of the first index and the second index of each group of high-frequency pressure data sequences is calculated, and the mean is normalized by the maximum-minimum normalization method based on the mean of all groups of high-frequency pressure data sequences. The normalization result is used as the deviation index of each group of high-frequency pressure data sequences.

[0062] The deviation index integrates the local high-frequency fluctuation characteristics represented by the time-domain difference amplitude and damping attenuation fitting degree of a single measuring point, as well as the spatial consistency represented by the correlation of high-frequency pressure data between measuring points. It comprehensively and multidimensionally quantifies the severity of water hammer impact on each group of high-frequency pressure data. A large deviation index indicates that the group of high-frequency pressure data has suffered strong water hammer impact interference, and the true seepage pressure drop signal has been severely masked by non-stationary distortion. Directly using such distorted data for crack inversion will introduce significant analytical errors. Therefore, when constructing a comprehensive pressure curve to assess the true seepage state, it is necessary to dynamically reduce the fusion weight of this group of high-frequency pressure data to eliminate the damage of local water hammer noise to the overall crack morphology identification accuracy.

[0063] For each set of high-frequency pressure data sequences, a contribution weight is constructed based on the deviation index, and the high-frequency pressure data sequences throughout the entire pump shutdown seepage stage are weighted to obtain a comprehensive pressure data sequence for evaluating crack morphology parameters. Specifically:

[0064] Calculate the difference between the natural number 1 and the deviation index of each group of high-frequency pressure data, and use the proportion of the difference of each group of high-frequency pressure data in the sum of the differences of all groups of high-frequency pressure data as the weight of each group of high-frequency pressure data sequence.

[0065] Calculate the weight of each group of high-frequency pressure data sequences and multiply it by the product of each data in the high-frequency pressure data sequence to obtain a weighted high-frequency pressure data sequence. Then, accumulate the data with the same position in all the weighted high-frequency pressure data sequences to obtain the final comprehensive pressure data sequence.

[0066] Step S005: Perform pressure drop fitting analysis based on the comprehensive pressure data sequence, determine the filtration type according to the fitting residual, and obtain the crack morphology parameters by combining the crack propagation model.

[0067] Based on the comprehensive pressure data sequence, analysis and inversion were performed to obtain crack morphology parameter data, specifically:

[0068] First, outlier removal is performed on the comprehensive pressure data sequence. In this embodiment, SG filtering is used to remove outliers. Implementers can choose other existing outlier removal algorithms. Second, the maximum-minimum normalization method is used to normalize the comprehensive pressure data sequence after outlier removal.

[0069] Then, a linear fitting analysis was performed on the normalized comprehensive pressure data sequence, and the least squares method was used for the fitting algorithm.

[0070] The filtering type of the pressure drop curve corresponding to the comprehensive pressure data sequence is determined based on the fitting results. When the fitted curve of the pressure drop curve is approximately a straight line, that is, when the mean of the fitting residuals of all sampling points in the normalized comprehensive pressure data sequence is less than a preset threshold... If the voltage drop curve is within a certain range, it is determined to be a linear filtration type; otherwise, it is determined to be a nonlinear filtration type. Its size can be set by the implementer according to the implementation scenario, without special restrictions. In this embodiment, Take 0.01.

[0071] When the pressure drop curve is determined to be of the linear filtration type, it indicates that the pressure drop characteristics conform to a constant filtration law. In this case, the pressure drop curve is directly substituted into a preset crack propagation model for global fitting to calculate the crack filtration coefficient. Based on the magnitude of the filtration coefficient, the crack width, height, and SRV volume, among other morphological parameters, are calculated using the crack propagation model. In this embodiment, the crack propagation model uses the PKN (Perkins-Kern-Nordgren) model; implementers can choose other crack propagation models, such as the KGD or Palmer model.

[0072] When the pressure drop curve is determined to be of the nonlinear filtration type, it indicates that the formation filtration characteristics undergo phased changes during the pressure drop process. First, the second derivative at each sampling point of the pressure drop curve is calculated. By identifying sampling points where the second derivative is not zero as segmented feature points, the entire pressure drop curve is divided into several locally smoothed sub-periods. Then, the pressure drop data for each segment are combined with a fracture propagation model for local calculations to obtain the segmented filtration coefficients corresponding to each sub-period. Finally, the average of all segmented filtration coefficients is calculated to obtain a comprehensive filtration coefficient characterizing the entire process. Based on this comprehensive filtration coefficient and the fracture propagation model, the fracture morphology parameters are inverted and identified.

[0073] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, specific embodiments of this specification have been described above. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0074] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0075] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them; modifications to the technical solutions described in the foregoing embodiments, or equivalent substitutions of some of the technical features, do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A fracture identification method based on high-frequency pressure signals in reservoirs, characterized in that, The method includes the following steps: High-frequency pressure sensor data were collected from each group during the pump shutdown and seepage stage of the fracturing section to obtain a high-frequency pressure data sequence; Each high-frequency pressure data sequence is processed by frame segmentation to determine the first-order difference feature of pressure fluctuation in each frame and the damping attenuation fitting feature of pressure envelope. Based on the first-order difference feature and the damping attenuation fitting feature, a first index reflecting the degree of influence of water hammer effect on a single set of data is constructed. The correlation between different sets of high-frequency pressure data sequences was analyzed, and a second index reflecting the spatial consistency of pressure response at different locations was constructed based on the correlation. The deviation index of each group of high-frequency pressure data is determined based on the first index and the second index, and the high-frequency pressure data of each group is weighted and fused based on the deviation index to obtain a comprehensive pressure data sequence. Pressure drop fitting analysis was performed based on the comprehensive pressure data sequence. The filtering type was determined according to the fitting residual, and the crack morphology parameters were obtained by inversion using the crack propagation model.

2. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 1, characterized in that, The first-order difference feature is specifically as follows: Perform a first-order difference operation on the same frame of high-frequency pressure data, calculate the mean of the absolute values ​​of all first-order difference values ​​corresponding to the same frame of high-frequency pressure data, and denote it as the first parameter; The second parameter of each high-frequency pressure data sequence is obtained by analyzing the numerical distribution characteristics of the first parameter corresponding to all frames of high-frequency pressure data in each high-frequency pressure data sequence, and is used to evaluate the first-order difference feature.

3. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 2, characterized in that, The second parameter is the range of the first parameter corresponding to all frames of high-frequency pressure data in each high-frequency pressure data sequence.

4. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 3, characterized in that, The damping attenuation fitting characteristics of the pressure envelope are specifically as follows: Extract the local maxima from each frame of high-frequency pressure data, and perform a logarithmic operation on each local maxima with the natural constant as the base; perform linear fitting on the logarithmic operation results of all local maxima in each frame of high-frequency pressure data, and calculate the degree of fit of the fitted line; The fitting degree corresponding to all frames of high-frequency pressure data in each high-frequency pressure data sequence is positively fused to obtain a third parameter, which is used to measure the damping attenuation fitting characteristics of the pressure envelope.

5. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 4, characterized in that, The process for determining the first index is as follows: The second parameter is normalized, and the mean of the normalization result and the third parameter is used as the first index.

6. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 1, characterized in that, The process for determining the second index is as follows: Calculate the correlation between any set of high-frequency pressure data sequences and the remaining sets of high-frequency pressure data sequences, calculate the mean of the absolute values ​​of all the correlations corresponding to the any set of high-frequency pressure data sequences, and the second index is negatively correlated with the mean.

7. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 1, characterized in that, The deviation index is the normalized result of the mean of the first index and the second index.

8. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 1, characterized in that, The weights for weighted fusion of each group of high-frequency pressure data are: Calculate the difference between the natural number 1 and the deviation index of each group of high-frequency pressure data, and use the proportion of the difference of each group of high-frequency pressure data in the sum of the differences of all groups of high-frequency pressure data as the weight.

9. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 1, characterized in that, The method of determining the filtering type based on the fitting residual includes: Linear fitting is performed on the comprehensive pressure data sequence. If the mean of the fitting residuals of all data points is less than a preset threshold, the pressure drop curve corresponding to the comprehensive pressure data sequence is determined to be of the linear filtering type; otherwise, the pressure drop curve corresponding to the comprehensive pressure data sequence is determined to be of the nonlinear filtering type.

10. The fracture identification method based on high-frequency reservoir pressure signals as described in claim 9, characterized in that, When the pressure drop curve is of the linear filtration type, the filtration coefficient of the crack is calculated by combining the crack propagation model, and the crack morphology parameters are obtained by inversion. When the pressure drop curve is of the nonlinear filtration type, the pressure drop curve is segmented by the second derivative of each sampling point on the pressure drop curve, and the filtration coefficient of each segment is calculated separately. After forward fusion, the comprehensive filtration coefficient is obtained, and the crack morphology parameters are obtained by inversion through the crack propagation model.

Images