Carbonate rock thermal storage acidizing fracturing permeation enhancement inversion method

Abstract

The invention belongs to the technical field of data processing and analysis, and particularly relates to a carbonate rock thermal storage acidizing fracturing permeability increase rate inversion method. Comprising the following steps: step 1, installing an optical fiber distributed sound wave monitoring and optical fiber distributed temperature monitoring composite cable at a target well section to collect monitoring data; 2, processing the collected monitoring data, and constructing a unified observation sequence; 3, a combined response template library corresponding to the established well periphery modeling unit is constructed, and orthogonal matching pursuit type strip selection and residual error updating are executed based on the compressed sensing thought by taking the constructed unified observation sequence as a target so as to generate a sparse permeability increasing rate field; and 4, performing numerical checking on the generated sparse permeability increasing rate field, and outputting the corresponding sparse permeability increasing rate field. According to the technical scheme, the calculation efficiency and adaptability are remarkably improved while the result reliability is guaranteed.

Description

Technical Field

[0001] This invention belongs to the field of data processing and analysis technology, and particularly relates to the field of intelligent sensing systems, specifically a method for inverting the permeability enhancement rate of carbonate rock thermal reservoirs through acid fracturing. Background Technology

[0002] Carbonate formations, characterized by well-developed natural fractures, diverse dissolution pores, and significant differences in seepage characteristics between the matrix and fractures, have long been considered one of the most complex reservoir types in geothermal development and acid fracturing engineering. Existing studies on carbonate acid fracturing generally employ methods based on wellbore pressure curve analysis, chemical reaction kinetic modeling, and finite element or finite difference numerical simulations to estimate the spread and conductivity of acid within the formation. These methods typically establish fluid-rock interaction models and, combined with the relationship between acid etching rate and porosity evolution, calculate permeability changes after acidification. However, in practical applications, because the acid etching reaction is influenced by flow rate, temperature, lithology, and fracture structure, acidification channels often exhibit non-uniform, sparsely distributed high-permeability zones, making it difficult for traditional models to accurately infer their spatial location and intensity.

[0003] In terms of on-site monitoring methods, existing technologies mainly include wellhead pressure monitoring, microseismic location, acoustic imaging, temperature profile measurement, and fiber-optic distributed temperature sensing (DTS) and distributed acoustic sensing (DAS). Fiber-optic distributed sensing technology can provide continuous temperature and acoustic change information along the wellbore, enabling real-time observation of dynamic changes during acid ingress, fracture opening, and flowback. However, existing studies mostly use DTS or DAS data only for qualitative identification of acidized sections or fracture activity areas, failing to effectively establish a quantitative relationship between monitoring data and reservoir permeability increases, especially in the identification of sparse, strip-shaped high-permeability channels, where existing algorithms lack accuracy and stability. At the data processing and inversion level, the conventional approach is to fit numerical simulation results or empirical models with measured signals, optimizing parameters to obtain the permeability distribution that best matches the monitoring data. These methods typically require significant computational resources and heavily rely on initial assumptions and boundary conditions. For example, three-dimensional acid etching simulations based on finite element meshes often require long iterations to achieve stable convergence, and the uncertainty of model parameters leads to multiple solutions in the final results. In addition, DAS and DTS data suffer from problems such as non-uniform sampling in time and space, signal-to-noise ratio variations, and nonlinear response, making it impossible to obtain sparse and accurate imaging of high-permeability regions by directly using traditional least squares fitting or filtering methods. Summary of the Invention

[0004] The main objective of this invention is to provide a method for inverting the permeability enhancement rate during acid fracturing in carbonate reservoirs. By jointly inverting acoustic and temperature responses, the ability to identify acidized bands is significantly enhanced. Utilizing the sparse reconstruction characteristics of compressed sensing algorithms, efficient inversion is achieved under undersampling conditions. Through band-level stepwise matching and residual iteration, computational complexity is reduced and real-time performance is improved. Numerical verification ensures the dynamic consistency between the inversion results and monitoring data. This invention enables real-time identification of high-permeability channels and quantitative assessment of permeability enhancement rates during acid fracturing in carbonate reservoirs, providing a reliable basis for the optimized design and construction control of conductivity near the wellbore, and significantly improving the efficiency and accuracy of reservoir development.

[0005] To address the aforementioned technical problems, this invention provides a method for inverting the permeability enhancement rate of carbonate rock thermal reservoirs via acid fracturing, comprising the following steps: Step 1: Install fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring composite cables in the target well section to collect monitoring data, and establish a fixed spatial discretization of the well perimeter modeling unit formed by the combination of well depth, radial ring and equiangular sector Cartesian combination. Step 2: Process the collected monitoring data to construct a unified observation sequence; Step 3: Construct and establish a joint response template library corresponding to the well perimeter modeling unit, and based on the compressed sensing concept, perform orthogonal matching tracking strip selection and residual update with the constructed unified observation sequence as the target to generate a sparse permeability field; Step 4: Perform numerical verification on the generated sparse permeability field and output the corresponding sparse permeability field.

[0006] Furthermore, in step 1, the fiber-optic distributed acoustic monitoring uses a measurement distance of 10m, a spatial sampling interval of 1m, and a time sampling frequency of 1000Hz; the fiber-optic distributed temperature monitoring uses a spatial sampling interval of 1m and a time sampling period of 1s; a well depth range with a total length of 240m is formed by extending 120m upwards and 120m downwards from the center of the perforation cluster; a near-well perimeter region with a radius of 6m is constructed starting from the outer wall of the wellbore; the aforementioned well depth range is divided into 240 well depth layers with a thickness of 1m, the aforementioned near-well perimeter region is divided into 12 radial rings with a radial thickness of 0.5m, and the circumferential region is divided into 36 equiangular sectors with an angle of 10 degrees; the well perimeter modeling unit is formed by a Cartesian combination of the 240 well depth layers, the 12 radial rings, and the 36 equiangular sectors.

[0007] Further, step 2 includes: performing bandpass filtering on the raw gathers of the acquired fiber-optic distributed acoustic monitoring at each well depth, setting a lower passband limit of 40Hz and an upper passband limit of 160Hz; performing Hilbert envelope calculation and moving average on the filtered gathers, setting a moving average window of 3s; performing baseline correction on the acquired fiber-optic distributed temperature monitoring sequence at each well depth, taking the average value of the baseline 30s before zero time, and then smoothing it according to a 3s moving average window; at each well depth, concatenating the acoustic envelope sequence and the temperature sequence point by point according to the time index to form a joint response sequence; stacking the joint response sequences sequentially from top to bottom in the well depth direction to form a unified observation sequence.

[0008] Furthermore, step 3 includes the following steps: Strip candidate generation step: Within each well depth and each isoangular sector, 12 radial rings within the current well depth are connected radially to form a high-permeability conductive strip candidate, resulting in 240×36 high-permeability conductive strip candidates; Joint response template library construction step: For each generated high-permeability conductive strip candidate, corresponding acoustic and temperature templates are generated and cascaded to form a strip joint template, thereby constructing a joint response template library; Orthogonal matching tracking strip selection and residual update step: Using the constructed unified observation sequence as the target sequence and the constructed joint response template library as the candidate set, strip selection and residual update are performed; Sparse permeability field generation: Based on all selected strips, a sparse permeability field is generated.

[0009] Furthermore, the joint response template library construction steps include: setting the time evolution of the acoustic template, including: dividing 3600s into three continuous acoustic phases; the first acoustic phase is from 0s to 120s, with the amplitude linearly increasing to 100% of the peak value; the second acoustic phase is from 120s to 720s, with the amplitude remaining at 100% of the peak value; the third acoustic phase is from 720s to 3600s, with the amplitude linearly decreasing to 15% of the peak value; and setting the time evolution of the temperature template, including: dividing 3600s into... The temperature is divided into three continuous temperature stages. The first temperature stage 1 is from 0s to 300s, during which the temperature linearly decreases to 2 degrees Celsius below the baseline. The second temperature stage is from 300s to 1200s, during which the temperature remains 2 degrees Celsius below the baseline. The third stage is from 1200s to 3600s, during which the temperature linearly rises back to 0.2 degrees Celsius below the baseline. The generated acoustic template and the generated temperature template are cascaded to form a strip joint template. The formed strip joint template is then normalized, and its 95th percentile value is scaled to 1.

[0010] Furthermore, the joint response template library construction steps also include: setting the radial amplitude factor of the acoustic template: for the 12 radial rings within the candidate high-permeability conductive strip, a relative amplitude factor array [1.00, 0.70, 0.50, 0.35, 0.25, 0.18, 0.13, 0.09, 0.06, 0.04, 0.03, 0.02] is sequentially assigned; at the corresponding well depth, the set acoustic template time evolution is multiplied by the relative amplitude factor ring by ring along the radial direction, and then summed radially, and a longitudinal smoothing effect caused by a 10m measurement distance is applied, specifically including: using a moving average with a length of 10 well depths in the well depth direction to obtain... Acoustic templates for cascading; setting radial influence factors for temperature templates: For the 12 radial rings within the candidate high-permeability conductive strip, a relative influence factor array [1.00, 0.85, 0.72, 0.60, 0.50, 0.42, 0.35, 0.29, 0.24, 0.20, 0.17, 0.15] is sequentially assigned; at the corresponding well depth, the time evolution of the set temperature template is multiplied by the aforementioned relative influence factors ring by ring along the radial direction and then summed along the radial direction, and a longitudinal smoothing effect caused by 1m spatial sampling is applied, specifically including: using a moving average with a length of 3 well depths in the well depth direction to obtain the temperature templates for cascading.

[0011] Furthermore, the orthogonal matching tracking strip selection and residual update steps include: normalizing the constructed unified observation sequence strip by strip according to well depth to obtain an initial residual sequence; in each round, calculating the correlation coefficient between the aforementioned initial residual sequence and each joint template in the joint response template library at the corresponding well depth, and using the average value of the calculated correlation coefficient in the time dimension as the similarity; selecting the strip with the highest similarity from the candidate set as the selected strip for this round, and removing the template corresponding to the selected strip from the candidate set; in the well depth where the selected strip is located, linearly scaling the strip joint template by amplitude so that the 95th percentile value of the strip joint template is consistent with the 95th percentile value of the initial residual in the corresponding well depth; subtracting the initial residual from the aligned strip joint template at the corresponding well depth point by point to obtain a new residual sequence; and performing a moving average of 5 well depths on the two well depths above and below the well depth where the selected strip is located in the new residual sequence to achieve orthogonalization.

[0012] Furthermore, the termination criteria for the orthogonal matching tracking strip selection and residual update steps include: the root mean square value of the calculated new residual sequence over all well depths and all times; termination when the calculated root mean square value is less than 10% of the root mean square value of the initial residual sequence; or termination when the number of rounds reaches 50.

[0013] Furthermore, the sparse permeability field generation steps include: for all selected strips, within the 12 radial rings and corresponding isoangular sectors covered by the selected strips, marking the corresponding well perimeter modeling units as high permeability units and assigning them an equivalent permeability multiple of 20; assigning an equivalent permeability multiple of 1 to unmarked well perimeter modeling units; and performing connectivity merging on adjacent well depth layers where the selected strips appear consecutively in the well depth direction, with the rule that consecutive occurrences of 3 well depth layers are treated as continuous strips, and those with a length of 1 to 2 well depth layers are treated as isolated strips.

[0014] Further, step 4 includes: establishing a near-wellbore two-dimensional annular fan grid on the generated sparse permeability field, setting the inner boundary of the wellbore to a constant pressure of 30 MPa, the outer boundary to a constant pressure of 20 MPa, the fluid viscosity to 1 cP, and the simulation duration to 3600 s; using explicit time-progression to perform flow and energy transfer calculations, extracting the acoustic energy proxy and temperature change curves at the fiber optic location, and generating a simulated joint response sequence; comparing the generated simulated joint response sequence with the constructed unified observation sequence on three indicators: duration of the time rise phase, peak occurrence time, and peak band thickness in the well depth direction, with an allowable deviation of ±10% for each of the three; when the aforementioned allowable deviation conditions are met, outputting the sparse permeability field generated in step 3.

[0015] The method for inverting the permeability enhancement rate of carbonate rock thermal reservoirs through acid fracturing in this invention has the following beneficial effects: By fully utilizing the spatiotemporal continuous observation capabilities of fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring, and constructing a joint response template library corresponding one-to-one with well perimeter modeling units, and using compressed sensing principles for progressive selection of sparse strips and residual updates, rapid identification of high-permeability conduction channels around wells and quantitative inversion of the permeability field are achieved. The beneficial effects of this invention are: Firstly, by constructing a joint response sequence of acoustic waves and temperature, the complementary characteristics of acoustic disturbances and temperature anomalies during acid fracturing are fully preserved, enabling the dynamic evolution of high-permeability conductive strips to be simultaneously characterized under a unified observation framework, thereby significantly improving the sensitivity to acid-etched strips at the information fusion level.

[0016] Secondly, this invention introduces compressed sensing algorithm into the field of acid fracturing data inversion, replacing the high-dimensional solution of traditional continuous medium simulation with sparse strip combination, so that the inversion process is transformed from global grid solution to stepwise matching of finite strips, which greatly reduces the computational complexity and can complete the identification of high-permeability areas under near real-time conditions.

[0017] Third, through the orthogonal matching tracking strip selection and residual update mechanism, this invention prioritizes the interpretation of the main energy structure in the observation data in each round of selection, and suppresses repeated fitting by smoothing the deep layers of adjacent wells, so as to keep the number of selected strips to a minimum while ensuring the integrity of the interpretation, thereby obtaining a sparse permeability field with clear physical meaning.

[0018] Fourth, by combining a unified observation sequence and numerical verification process, this invention enables the inversion results to not only correspond to the monitoring data in terms of spatial distribution, but also to be consistent with the field observations in terms of dynamic characteristics. It can be directly used for evaluating the acidizing effect of the well section and optimizing subsequent fracturing parameters.

[0019] Compared with traditional inversion methods that rely on full-field solutions, the technical solution of this invention significantly improves computational efficiency and adaptability while ensuring the reliability of the results. It can quickly provide real-time assessment results of conductivity distribution at the acid fracturing site, which is of great significance for improving the development efficiency and construction control accuracy of carbonate reservoirs. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the three-dimensional spatial discretization of the well perimeter modeling unit provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the time evolution curve of the strip acoustic wave template provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the residual descent curve of the orthogonal matching pursuit iteration provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the two-dimensional annular fan grid distribution of the sparse permeability field provided in an embodiment of the present invention. Detailed Implementation

[0021] The method of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0022] The method for inverting the permeability enhancement rate of carbonate rock thermal reservoirs through acid fracturing includes the following steps: Step 1: Install fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring composite cables in the target well section to collect monitoring data, and establish a fixed spatial discretization of the well perimeter modeling unit formed by the combination of well depth, radial ring and equiangular sector Cartesian combination.

[0023] In one embodiment, a composite cable for fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring is laid along the outer wall of the completion pipe in the target well section. Monitoring data is collected using this composite cable. Simultaneously, a fixed spatial discretization of the well perimeter modeling unit, formed by a Cartesian combination of well depth, radial ring, and equiangular sector, is established. The specific process is as follows: First, the target well section and well perimeter range are determined. A well depth range of 240m is formed by extending 120m upwards and 120m downwards from the center of the perforation cluster. This length is chosen because the effective influence of acid fracturing on the well perimeter seepage field is usually concentrated within a range of several tens of meters above and below the perforation cluster. 240m can completely cover the response along the well depth direction during the multi-cluster intervention and flowback stages. A near-well perimeter region with a radius of 6m was constructed starting from the outer wall of the well. The 6m radius was chosen because the observability of changes in the well perimeter conductivity is most significant within a scale of several meters for fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring. Beyond this radius, the attenuation and diffusion of the signal along the well depth and radial direction will make the strip structure tend to be smoothed in observation, which is not conducive to subsequent strip-level inversion.

[0024] refer to Figure 1 The wellbore modeling unit established in this invention adopts a fixed spatial discrete structure formed by a Cartesian combination of well depth layers, radial rings, and equiangular sectors. The central circular area represents the wellbore. Starting from the outer wall of the wellbore, it extends radially outward to construct a near-wellbore region with a radius of 6m. This near-wellbore region is divided into 12 concentric radial rings with a thickness of 0.5m in the radial direction, arranged sequentially from the innermost 1st radial ring to the outermost 12th radial ring. On a section perpendicular to the wellbore axis, using a fixed zero-degree direction as a reference, the circumference is uniformly divided into 36 equiangular sectors, each with an angle of 10 degrees. The sector numbers are sequentially labeled from 1 to 36 starting from the 0-degree direction. The figure marks four main orientations: 0 degrees, 90 degrees, 180 degrees, and 270 degrees, ensuring a unified reference for the circumferential index across the entire well depth range. Along the well depth direction, the target well section is divided into 240 well depth layers with a thickness of 1m, consistent with the spatial sampling interval of fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring. By combining well depth, radial rings, and equiangular sectors in a Cartesian fashion, 240×12×36 well perimeter modeling units are formed. Each modeling unit has a unique spatial index, enabling point-to-point correspondence with monitoring data. Arrows in the figure indicate the extension range along the well depth direction, and radial labels show the 6m radius distance from the wellbore center to the outer boundary. This fixed spatial discretization provides a clear spatial basis for subsequent strip candidate generation, joint response template library construction, and sparse permeability field inversion.

[0025] Subsequently, the composite cable was laid and coupled. A fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring composite cable with an outer sheath pressure and temperature resistance rating no lower than the upper limit of the construction temperature was selected. The two optical fibers in the composite cable were laid coaxially and parallel to reduce spatial errors along the well depth direction. The composite cable was laid close to the outer wall of the completion casing and fixed with stainless steel clamps at 2m intervals. A coupling gasket with high thermal conductivity and micro-strain transfer capability, approximately 2mm thick, was placed between the clamp and the casing. During installation, an epoxy-based coupling agent was evenly applied to the clamp contact surface to eliminate minor gaps. The clamp tightening torque was controlled within the manufacturer's recommended range to ensure consistency in acoustic and thermal coupling. A centralizer was installed every 10m to ensure uniform force on the composite cable along the wellbore outer wall, avoiding local suspension that could lead to signal attenuation. Three additional clamps were installed before and after the perforation cluster location, shortening the interval to 1m, to ensure stable coupling in high-gradient flow and temperature-changing sections. The upper end of the composite cable is led to the ground via a pressure isolation crossing device, connecting to fiber-optic distributed acoustic monitoring and demodulation equipment and fiber-optic distributed temperature monitoring and demodulation equipment. Fiber continuity and echo attenuation checks are performed to ensure there are no additional losses caused by micro-bends or macro-bends along the route. After mechanical installation is completed, baseline data is acquired for 300 seconds before pump startup to record the static background noise and temperature field of the wellbore, providing a reference for subsequently distinguishing between the dynamic response and static background caused by construction.

[0026] Fiber optic distributed acoustic monitoring employs a measurement distance of 10m, a spatial sampling interval of 1m, and a temporal sampling frequency of 1000Hz. The 10m measurement distance is chosen because this length ensures sufficient resolution along the well depth while integrating minute phase differences along the well depth, achieving coherent enhancement of fracturing-induced acoustics and fluid disturbances. If the measurement distance is too short, the coherent gain along the well depth is insufficient, increasing the proportion of construction background noise; if the measurement distance is too long, although the noise is further averaged, the details along the well depth are overly smoothed, reducing the sharpness of subsequent strip positioning. The 1m spatial sampling interval allows for the formation of an oversampled grid without changing the measurement distance, facilitating the one-to-one correspondence between fiber positions and well depth layers. The 1000Hz temporal sampling frequency is chosen to cover the broadband acoustic components generated by the construction fluid and the near-field of the fracture, while also preserving sufficient sampling margin for subsequent time-domain feature extraction. Fiber optic distributed temperature monitoring employs a spatial sampling interval of 1m and a temporal sampling period of 1s. The spatial sampling interval is consistent with that of fiber optic distributed acoustic monitoring, facilitating point-to-point alignment of the two types of data. The 1-second temporal sampling period can capture sudden and gradual temperature changes during acid injection and backflow, while limiting the data volume, facilitating continuous acquisition during construction. Taking the pump start-up moment as time zero, continuous acquisition is performed for 3600 seconds, covering the acid injection, pressure maintenance, and backflow stages, making the entire process of strip formation and flow capacity establishment traceable.

[0027] To ensure consistency of well depth coordinates, depth registration was performed after the composite cable was laid. Identification marks with known spacing on the composite cable were recorded synchronously with ground length measurements. The first sampling point at the wellhead for both fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring was defined as the upper boundary of the well depth, and aligned with well completion tracer data or electronic depth sounding data. Accumulated errors were corrected to create a one-to-one mapping table between well depth and composite cable sampling points. This ensures that the two types of monitoring are completely consistent in location along the well depth, facilitating subsequent discretization by well depth layer.

[0028] When establishing a fixed spatial discretization, a three-dimensional reference frame is first defined. The well depth axis is defined along the wellbore direction, with the wellbore axis as the central axis; the radial direction is defined extending from the wellbore outer wall towards the surrounding rock; and the circumferential angle is defined on a section perpendicular to the wellbore axis. The zero-degree direction of the isotropic sector requires a fixed reference. Therefore, a reference mark with a known azimuth is set on the wellbore outer wall during well completion. The azimuth is taken from the well trajectory drilling azimuth data or the result of independent gyroscope azimuth measurement. This azimuth is projected onto the horizontal plane as the zero-degree direction. In this way, the zero-degree direction can be used to divide the circumference into 36 isotropic sectors with an angle of 10 degrees on any well depth section, ensuring that the circumferential index is traceable throughout the entire well depth range.

[0029] Along the well depth direction, the aforementioned 240m well depth range is divided into 240 well depth layers with a thickness of 1m. A thickness of 1m is chosen because the spatial sampling interval between fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring is 1m; using the same thickness ensures that each well depth layer is naturally aligned with a set of monitoring sampling points, avoiding positional deviations introduced by interpolation. Along the radial direction, from the outer wall of the wellbore outwards, 12 radial rings are divided with a radial thickness of 0.5m, forming twelve annular bands from near the well to 6m. A radial thickness of 0.5m is chosen because the enhancement of conductivity by the strips after acidizing has a significant gradient in the near-well meter range; using 0.5m allows for the stratified expression of strong gradients near the well and weak gradients far from the well, providing a clear radial hierarchy for subsequent strip priority path identification. Along the circumferential direction, the cross-section is divided into 36 equiangular sectors, each with an angle of 10 degrees. Combined with the aforementioned zero-degree directional reference, this gives the circumferential index a fixed starting point and uniform angular stepping, facilitating the traversal of strip candidates circumferentially.

[0030] The well depth layers, radial rings, and equiangular sectors are combined using Cartesian methods to form 240×12×36 well perimeter modeling units, which serve as fixed spatial discretizations. To align the monitoring sampling points with the well perimeter modeling units, each 1m well depth sampling location is assigned to its corresponding well depth layer number according to a mapping table. Since the optical fiber is located on the outer wall of the wellbore, its physical location is in the innermost radial ring, and the corresponding radial ring number is fixed as the innermost layer. Circumferentially, the optical fiber is linearly distributed, but the interpretation of strips needs to cover the entire circumference. Therefore, when constructing the well perimeter modeling units, the monitoring values ​​are considered as linear observations of the full circumferential response. Subsequently, in strip candidate generation and template construction, the circumferential decomposition is undertaken by the equiangular sectors. The core consideration of this process is that the high-density observation of the optical fiber along the wellbore direction provides a strong constraint on the longitudinal continuity of the strips, while the circumferential decomposition establishes comparability by fixing the zero-degree direction and equiangular discretization, making the circumferential indices between different well depth sections have consistent semantics, which facilitates sequential scanning when generating strip candidates.

[0031] After installation and discrete setup are completed, continuous data acquisition begins. Fiber optic distributed acoustic monitoring records data in real time at a measurement distance of 10m, a spatial sampling interval of 1m, and a time sampling frequency of 1000Hz; fiber optic distributed temperature monitoring records data in real time at a spatial sampling interval of 1m and a time sampling period of 1s. The pump start-up time is taken as time zero. Before time zero, the demodulation equipment's gain self-check and clock calibration are completed, unifying the timestamps of both types of monitoring to the same starting point to ensure that secondary timing correction is not required during subsequent alignment.

[0032] Considering site variations, several optional implementation methods are available. In optional implementation method one, when the well diameter is large or the thermal coupling between the wellbore outer wall and the surrounding rock is weak, the radius of the near-wellbore area can be taken as 8m, the radial ring number remains at 12, and the radial thickness is adjusted accordingly to approximately 0.67m to maintain a consistent number of layers from near the wellbore to the outside, thereby maintaining comparability of the layers at the strip scale. In optional implementation method two, when the perforation cluster density is high and varies more drastically along the well depth, the well depth range is still taken as 240m, but the thickness of the deep layer can be taken as 0.5m, and the number of deep layers is correspondingly 480. At the same time, the spatial sampling interval of the fiber optic distributed acoustic monitoring is set to 0.5m to ensure that the monitoring sampling corresponds one-to-one with the deep layer. In optional implementation method three, to enhance the stability of the circumferential reference, a positioning component with a single azimuth opening can be set on the completion tool, so that the zero-degree direction remains consistent with the surface azimuth after being lowered and consolidated, thereby reducing the azimuth uncertainty caused by wellbore deviation or tool rotation.

[0033] The above implementation method constructs a stable and landable fixed spatial discretization by highly coupling the fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring composite cable, discretizing the layer thickness at a depth consistent with the monitoring sampling interval, representing the near-well gradient with meter-level radial loops, and dividing the area into equiangular sectors with a fixed zero-degree direction. This fixed spatial discretization, on the one hand, ensures that the monitoring data corresponds point-by-point with the spatial unit, reducing positional errors caused by interpolation; on the other hand, it provides a clear spatial index and traversal order for the subsequent construction of a joint response template library corresponding to the well perimeter modeling unit and orthogonal matching tracing strip selection, thereby reducing interpretation ambiguity and improving the stability of strip-level identification from the source.

[0034] Step 2: Process the collected monitoring data to construct a unified observation sequence.

[0035] In one embodiment, the collected monitoring data is processed to construct a unified observation sequence, including time alignment, feature extraction, and joint encoding of fiber-optic distributed acoustic monitoring data and fiber-optic distributed temperature monitoring data. This enables any subsequent well perimeter modeling unit to perform strip identification and sparse permeability field inversion using the unified observation sequence. The specific implementation process is as follows: During the construction of the target well section, the fiber-optic distributed acoustic monitoring demodulation equipment continuously outputs a three-dimensional dataset at a time sampling frequency of 1000Hz and a spatial sampling interval of 1m. Each well depth corresponds to a time series, and the gather includes the baseline stage before pump start-up and the entire process for 3600s after pump start-up. The fiber-optic distributed temperature monitoring demodulation equipment outputs a dataset showing temperature variations with time and well depth at a time sampling period of 1s and a spatial sampling interval of 1m, covering the same well depth range and time range. By synchronizing the two demodulation equipment with a unified clock before construction and setting the pump start-up time as the zero point, it is ensured that the acoustic monitoring data and temperature monitoring data have a unified time reference, so that the same time index corresponds to the same physical process in the two datasets. After the data collection was completed, a subset with a time range of 0s to 3600s and a well depth range of 240m was extracted from the two types of raw data for the construction of a unified observation sequence.

[0036] For fiber-optic distributed acoustic monitoring data, the original time series was first bandpass filtered at each well depth, with a lower passband limit of 40Hz and an upper passband limit of 160Hz. This frequency band covers the mid-to-high frequency effective energy generated by fluid disturbances in perforations, fracture propagation, and high-permeability flow channels around the well, while suppressing large-scale slowly changing background noise and high-frequency electronic noise. By limiting the frequency band, the retained signal mainly reflects energy changes related to flow channel activity, which is beneficial for amplifying the difference between the flow channel response and the background region in subsequent processing. In implementation, a finite impulse response filter or an equivalent digital filtering method can be used. Linear phase design maintains the reliability of event time locations, avoids introducing spurious shifts along the time direction, and ensures that subsequent peak times match actual construction events.

[0037] After bandpass filtering, Hilbert envelope calculations are performed on the filtered time series at each well depth. By constructing an analytic signal and calculating the instantaneous amplitude, the originally oscillating acoustic signal is transformed into a single-valued energy curve, so that the value at each time point represents the strength of the acoustic disturbance in the vicinity of that moment. The reason for using the envelope is that acid fracturing causes changes in energy level and duration, rather than the phase characteristics of a single-period signal. The envelope form is beneficial for comparisons between different depths and time periods, and also facilitates a unified description with slow variables such as temperature changes.

[0038] To maintain consistency in temporal resolution between fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring, while suppressing the instability caused by transient spikes, the envelope curves at each well depth were downsampled and smoothed. Specifically, with a time step of 1 second, the values ​​of every 1000 consecutive envelope sampling points were averaged to obtain an acoustic envelope sequence with a time interval of 1 second. This method reduces the amount of data while preserving short-to-medium cycle energy evolution information, ensuring that the acoustic and temperature signals have unified sampling points on the time axis. Subsequently, a 3-second moving average window was applied to the 1-second acoustic envelope sequence, using the envelope average of several points before and after the current time point as the update value, so that short-term abnormal spikes are balanced by the energy levels of surrounding periods. This processing helps to highlight continuous energy enhancement and attenuation, and continuous energy changes are precisely the manifestation of acidizing fluid selectively entering high-permeability conductive strips and maintaining conductivity, which can more accurately reflect the location of the strips and their relative activity.

[0039] For fiber-optic distributed temperature monitoring data, baseline correction was performed on the time series at each well depth. Specifically, the average value of the temperature sample taken 30 seconds before pump start-up was calculated and used as the static baseline temperature for that well depth. This baseline temperature was then used as a reference to calculate the temperature change at each subsequent time point. This process removes static differences caused by formation temperature gradients, well completion structures, and long-term stable environments, allowing subsequent analysis to focus on construction-driven temperature disturbances. This directly correlates temperature changes with dynamic behaviors such as acid breakthrough and hot / cold fluid exchange. Subsequently, a 3-second moving average was applied to the temperature change time series at each well depth. This balanced single-point measurement noise and instantaneous errors with values ​​from adjacent time periods, highlighting persistent cooling and recovery zones. This persistent temperature anomaly is associated with preferential acid entry channels and the formation of stable flow paths in high-permeability regions. The smoothed temperature change data clearly identifies the duration and recovery rate of the anomaly on the time axis, providing a basis for constructing strip response characteristics along with acoustic energy.

[0040] After the above processing, fiber-optic distributed acoustic monitoring generates a data sequence with a time step of 1 second and a length of 3600 units at each well depth, while fiber-optic distributed temperature monitoring generates a temperature change data sequence with a time step of 1 second and a length of 3600 units at each well depth. Since their time indices are perfectly aligned, the acoustic envelope value and temperature change can be combined point-by-point at each well depth and time point to form a joint response sequence for that well depth. In the joint response sequence, each time step corresponds to a set of observations composed of acoustic energy and temperature change indices, used to simultaneously characterize the flow disturbance intensity and thermal breakthrough behavior at that location at that moment. For subsequent strip identification, the combination of enhanced acoustic envelope and continuously decreasing temperature often corresponds to the rapid passage of acid in high-permeability flow strips; the combination of stable acoustic energy and small temperature changes corresponds to the matrix zone or low-permeability region. This construction of the joint response allows the algorithm to identify differences between different spatial units under a unified data structure.

[0041] To construct a unified observation sequence, all well depth locations from the upper to the lower well depth boundary are arranged sequentially according to well depth layers. For each well depth layer, its corresponding joint response sequence is used as the observation vector. By stacking the joint response sequences of all well depth layers along the well depth direction, a unified observation sequence is formed. Structurally, this unified observation sequence corresponds to the well depth direction, the time direction, and the joint response components at each time point, allowing for direct comparison with a joint response template library on a well-depth and time-by-time basis. Since the well depth layer division, sampling interval, and time step are all kept consistent during the construction process, the mapping relationship between the unified observation sequence and the well perimeter modeling units is clear. Each well depth location corresponding to any well perimeter modeling unit has a unique corresponding joint response trajectory in the unified observation sequence, providing direct input for subsequent strip candidate response matching and sparse permeability field construction.

[0042] In an alternative implementation, after completing the Hilbert envelope and 1-second downsampling, a 5-second moving average window can be used to process the acoustic envelope sequence for well sections with high noise levels requiring stronger smoothing. In this implementation, the time step of the unified observation sequence remains at 1 second, and a wider neighborhood is introduced into the averaging process only in the calculation of a single time step, thereby enhancing sensitivity to continuous events. This approach is suitable for conditions where acidizing duration is long and the impact of a single pressure fluctuation on the overall trend is relatively small.

[0043] In another optional implementation, when the on-site temperature changes relatively drastically, the temperature baseline interval can be extended to 60 seconds before pump start-up to obtain a more stable baseline reference. After baseline correction, the amplitude of temperature changes can be truncated, and outliers exceeding a preset extreme threshold can be replaced with the average value of nearby time points to avoid individual outlier readings affecting strip feature judgment. This processing method makes the temperature components in the unified observation sequence more focused on the effective response related to acidification flow, which helps to improve the accuracy of strip matching.

[0044] In another alternative implementation, the acoustic envelope sequence and the temperature change sequence can be normalized before generating the joint response sequence. For example, the percentile value of each time series at each well depth can be used as a scaling reference to ensure that the numerical ranges at each well depth are within a uniform scale. This normalization process makes the unified observation sequences comparable across different well depths, facilitating the subsequent algorithm to identify high-permeability conduction strips using a unified threshold and matching criteria.

[0045] Through the above processing steps, the collected fiber-optic distributed acoustic monitoring data and fiber-optic distributed temperature monitoring data are transformed from their original form into a unified observation sequence with clear structure, unified time and well depth coordinates, and integrated feature expressions. This not only preserves the key dynamic information induced by acid fracturing, but also provides a clear data foundation and computational entry point for subsequent joint response template library matching and orthogonal matching tracking strip selection.

[0046] Step 3: Construct and establish a joint response template library corresponding to the well perimeter modeling unit, and based on the compressed sensing concept, perform orthogonal matching tracking strip selection and residual update with the constructed unified observation sequence as the target to generate a sparse permeability field.

[0047] In one implementation, a joint response template library corresponding to the well perimeter modeling units is constructed. Based on the compressed sensing concept, orthogonal matching pursuit-style strip selection and residual update are performed with a unified observation sequence as the target to generate a sparse permeability field. The specific implementation process is as follows: After completing the fixed spatial discretization of the well perimeter modeling units and the construction of the unified observation sequence, each high-permeability conductivity strip candidate is regarded as a continuous combination of the well perimeter modeling unit set in the well depth direction and radial direction, used to describe the dominant conductivity paths formed in local directions after acid fracturing. In order to deduce the location and number of these dominant paths from the unified observation sequence without directly solving complex flow equations, a strategy of interpreting most of the observation responses with the superposition of a small number of strips is adopted. That is, it is assumed that among all 240×36 possible directions, only a small number of strips are fully penetrated during acidizing, and their responses exhibit stable and identifiable characteristics in fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring. By mapping each high-permeability diversion strip candidate to a preset strip joint template, and then using the orthogonal matching pursuit method to gradually select the strip closest to the unified observation sequence from these strip joint templates, sparse selection based on the idea of ​​compressed sensing is completed, thereby generating a sparse permeability field.

[0048] First, high-permeability conductivity strip candidates are generated based on the fixed spatial discretization of the well perimeter modeling unit. For each well depth and each isoangular sector, 12 radial rings starting from the outer wall of the wellbore within that well depth are sequentially connected in the radial direction to form a high-permeability conductivity strip candidate pointing from the wellbore towards the surrounding rock. By performing the above connection on all combinations of 240 well depths and 36 isoangular sectors, 240×36 high-permeability conductivity strip candidates can be obtained. Each high-permeability conductivity strip candidate has a unique spatial location, covering an entire row of radial rings under the corresponding well depth and the corresponding isoangular sector. This construction method restricts potential conductivity channels to strips with clear directionality and continuity, so that each strip represents both a potential acid-preferred flow path and can participate in matching as an independent interpretation unit in subsequent processing, which is beneficial for reproducing the overall observation response using combinations of finite strips.

[0049] Then, a joint template for each candidate high-permeability conduction strip is constructed, and a joint response template library is formed by all the joint templates. The construction process of the joint response template library includes setting the time evolution morphology, setting the radial influence distribution, smoothing processing consistent with the actual monitoring resolution, and amplitude normalization, to ensure that the response of each strip in the template library has a clear physical meaning and comparability.

[0050] For the acoustic component, the temporal evolution of the strip acoustic template was first defined. Based on the characteristics of acid fracturing operations, the 3600 seconds were divided into three continuous stages: From 0 to 120 seconds, the acoustic energy linearly increased from the baseline level to a high disturbance level within the strip, reflecting the accumulation of early disturbances caused by pump start-up and the initial acid entering the strip; from 120 to 720 seconds, the acoustic energy remained at a high level, indicating continuous acid flow within the high-permeability conductive strip, generating stable flow noise and fracture surface friction signals; from 720 to 3600 seconds, the acoustic energy linearly decreased to approximately 15% of the high level, corresponding to a reduction in operational intensity and a gradual weakening of flow and backflow. This three-stage setting resulted in each template exhibiting a "rapid establishment, sustained maintenance, and slow decay" pattern on the time axis, consistent with the energy evolution trend in actual acid fracturing processes, and facilitating matching with the continuous high-energy intervals in a unified observation sequence.

[0051] In the radial direction, a radial amplitude factor decreasing from near the well to far from the well is set for the strip acoustic template to characterize the spatial distribution of flow disturbance within the high-permeability conductive strip, which is strongest near the wellbore and gradually decays outward. For the 12 radial rings within the strip, relative amplitude factors of [1.00, 0.70, 0.50, 0.35, 0.25, 0.18, 0.13, 0.09, 0.06, 0.04, 0.03, 0.02] are assigned sequentially. This setting reflects the consideration that the flow cross-sectional area is small, the flow velocity is high, and the shearing effect is significant near the wellbore, thus the proportion of acoustic energy is larger; as the radial distance increases, the fluid diffuses in the fracture network or porous medium, the local flow velocity decreases, and the randomness increases, resulting in a weakening contribution of the equivalent disturbance to the optical fiber deployed along the well depth. By multiplying the time evolution curve with the radial amplitude factor ring by ring within the strip and summing them radially, the acoustic time curve of the strip at the corresponding well depth is obtained. Then, a moving average with a length of 10 well depths is applied in the well depth direction to simulate the integral effect at a distance of 10 meters, transforming the local disturbance into a smooth response along the well depth direction, so that the template response form is consistent with the measurement characteristics of actual fiber-optic distributed acoustic monitoring.

[0052] refer to Figure 2The strip acoustic template constructed in this invention exhibits a three-stage time evolution characteristic within a 3600-second construction period. The horizontal axis represents time in seconds, with four key time nodes marked: 0 seconds, 120 seconds, 720 seconds, and 3600 seconds. The vertical axis represents the relative amplitude of acoustic energy, with a normalized range from 0 to 1.0 and a scale interval of 0.2. Starting from the initial point, in the rapid establishment phase from 0 to 120 seconds, the acoustic energy linearly rises from the baseline level to a peak of 1.0. This phase corresponds to the accumulation process of early disturbance caused by the pump and initial acid entering the high-permeability conductive strip. In the high-energy maintenance phase from 120 to 720 seconds, the acoustic energy remains at the peak level of 1.0, forming a plateau segment, indicating that the acid continues to flow within the high-permeability conductive strip, generating stable flow noise and fracture surface friction signals. During the slow decay phase from 720 to 3600 seconds, the acoustic energy linearly decreases from 1.0 to 0.15, a decay of 85%, corresponding to the process of reduced construction intensity, flowback, and gradual weakening of flow. The 120-second and 720-second phase boundaries are marked with vertical dashed lines in the figure, and key points are marked with solid dots. This three-stage evolution pattern reflects the typical characteristics of energy "rapid establishment, sustained maintenance, and slow decay" during acid fracturing, consistent with the flow disturbance patterns in actual high-permeability conductive strips, providing a standardized time evolution template for subsequent orthogonal matching tracking strip selection.

[0053] Regarding the temperature component, the time evolution of the strip temperature template was first defined. The 3600-second timeframe was divided into three consecutive phases: From 0 to 300 seconds, the temperature linearly decreased from the baseline level to 2 degrees Celsius below the baseline, indicating that cold acid or temperature-differential fluid preferentially penetrated the high-permeability flow strip, rapidly changing the temperature near the strip; from 300 to 1200 seconds, the temperature remained 2 degrees Celsius below the baseline, indicating that a continuous cold or low-temperature fluid existed stably within the strip; from 1200 to 3600 seconds, the temperature linearly recovered from 2 degrees Celsius below the baseline to 0.2 degrees Celsius below the baseline, indicating that the construction intensity weakened, thermal equilibrium gradually recovered, but slight residual anomalies still existed near the strip. This design emphasizes duration and recovery speed, enabling the temperature template to distinguish between ordinary flow with only short-term disturbances and high-permeability flow strips with a sustained low-temperature zone.

[0054] In the radial direction, a relative influence factor of [1.00, 0.85, 0.72, 0.60, 0.50, 0.42, 0.35, 0.29, 0.24, 0.20, 0.17, 0.15] is set for the strip temperature template to reflect that when cold fluid is injected along the strip, its influence on the temperature near the wellbore is the greatest, and its influence on the outer region gradually weakens, while retaining a certain degree of influence at the far end to characterize the strip connectivity. During template construction, the temperature-time evolution curve is multiplied ring by ring by the relative influence factor and summed radially to obtain the temperature-time curve at the corresponding well depth. Then, a moving average with a length of 3 well depths is used in the well depth direction to simulate the longitudinal smoothing characteristics of the temperature field under 1-meter spatial sampling, so that the resolution and smoothness of the template match the actual fiber-optic distributed temperature monitoring data.

[0055] For each candidate high-permeability conductivity strip, the acoustic template and temperature template of that strip are cascaded according to a unified observation sequence, forming a joint template of the strip composed of acoustic and temperature components at each time point. To ensure comparability between different strips, each joint template is normalized by scaling the 95th percentile value of the template data to 1. This normalization makes the matching process more dependent on time patterns and relative changes, and less affected by absolute amplitude, which helps to adopt a unified selection standard across different well depths and circumferential directions. Through the above process, joint templates can be generated for 240×36 candidate high-permeability conductivity strips. All joint templates constitute a joint response template library. Each template maintains a unified standard in terms of time length, composition, and amplitude processing, and can be directly compared with a unified observation sequence.

[0056] After obtaining a unified observation sequence and a joint response template library, an orthogonal matching pursuit-based strip selection and residual update method is used to select sparse strips. This process aims at the unified observation sequence, treats the joint response template library as a set of candidate interpretation strips, and through round-by-round selection and updating, makes the linear combination of a small number of strip joint templates approximate the unified observation sequence. Finally, the selected strips are mapped to a set of high permeability units in space, forming a sparse permeability field.

[0057] In the initial stage, the amplitude of the unified observation sequence is normalized for each well depth. For example, the 95th percentile of the joint response of each well depth can be used as the scaling benchmark to ensure that the numerical ranges of each well depth are at similar levels, thus obtaining the initial residual sequence. This approach is designed to avoid the overall amplitude of some well depths being too large due to background noise or local conditions, which could affect the lateral comparison between strips. This ensures that the strip selection process is based primarily on matching morphology and relative changes rather than absolute amplitude.

[0058] In each round of strip selection, the similarity between each strip joint template in the joint response template library and the current residual sequence at its corresponding well depth is calculated. Similarity can be determined by comparing the trends, peak times, and degree of overlap of the two over a time span. For example, the ratio of the sum of their pointwise products over a 3600-second timeframe to their respective energies can be used to measure their overall consistency. The strip with the highest similarity is selected as the entry strip for this round, indicating that its response characteristics best explain the main structure in the current residual. The reason for prioritizing the strip with the highest similarity is based on the idea of ​​"gradually capturing the most interpretable component" in compressed sensing. A strip is used to first absorb the most concentrated energy portion of the residual, allowing subsequent strip selections to supplement only the remaining features based on what has already been explained.

[0059] After identifying the selected strips for this round, the combined template of the strips is amplitude aligned at the well depth where the strips are located. Its 95th percentile value is scaled to match the 95th percentile value of the current residuals at that well depth, ensuring that the amplitude level of the combined template matches the dominant energy of the residuals. This amplitude alignment allows the template to participate in residual cancellation with appropriate intensity without altering its temporal morphology and relative structure, improving the interpretation effect after stacking. After amplitude alignment, the aligned combined template is subtracted point-by-point from the current residuals at the corresponding well depths to obtain a new residual sequence. This operation is equivalent to treating the selected strips as interpretable components already "extracted" from the observations, retaining the uninterpreted portions for the next iteration.

[0060] To reduce redundant interpretations across different well depths, orthogonalization is introduced. In the new residual sequence, a moving average of five well depths is applied to the two well depths above and below the well depth of the selected well in this round, smoothing out strong local fluctuations into a more uniform background component. This process weakens the diffusion effect of the same well depth on adjacent well depths, allowing subsequent wells to be selected to cover more uncovered depth and circumferential sections, avoiding multiple wells repeatedly fitting the same observation structure in the same area, and improving the discriminability between wells.

[0061] The orthogonal matching pursuit strip selection and residual update process is repeated. In each round, the similarity of all unselected strips is recalculated based on the updated residuals. The process continues, selecting the optimal strip, performing amplitude alignment, residual updates, and smoothing at neighboring well depths, until a termination condition is met. The termination condition can include one of two cases: first, the root mean square value of the new residual sequence decreases to less than 10% of the root mean square value of the initial residual sequence across all well depths and time ranges, indicating that the combination of a small number of strips has explained most of the observed response; second, the number of iterations reaches 50, indicating that a sufficiently sparse explanatory set has been obtained within the predetermined upper limit of the number of strips. This design ensures both a controllable number of strips and sufficient interpretability of the selected strip set for the observed data, aligning with the compressed sensing approach of "representing the main information using a finite sparse basis."

[0062] refer to Figure 3 This invention employs an orthogonal matching pursuit strip selection method to study the decreasing pattern of the root mean square value of the residuals during the iterative process. The horizontal axis represents the iteration number, from 0 to 50, with a scale interval of 5 iterations. The vertical axis represents the normalized value of the root mean square value of the residuals, ranging from 0 to 1.0, with the initial residual set to 1.0. The curve starts at iteration 0, with a residual value of 1.0, and exhibits a clear rapid decreasing characteristic as iterations progress. In the rapid decreasing phase of the first 10 iterations, the residual rapidly decreases from 1.0 to 0.35, a decrease of 65%, and the strips selected in each iteration during this phase significantly contribute to interpreting the observed response. In the stable convergence phase from iterations 10 to 25, the residual further decreases from 0.35 to 0.15, with the rate of decrease slowing down, indicating that the main strips have been identified. After iteration 25, the residual continues to decrease slowly to around 0.09 and tends to stabilize. The residual threshold of 0.1, corresponding to 10% of the initial residual, is marked with a horizontal dashed line in the figure and serves as one of the iteration termination conditions. Key points in the first 15 rounds of the curve are marked with solid circles, representing the selected high-permeability conduction bands in the corresponding rounds. The entire curve exhibits an exponential decay characteristic, verifying the effectiveness of the compressed sensing approach of "representing the main information with a small number of sparse bases." The annotation below the figure indicates that the number of selected bands ranges from 15 to 25, reflecting the core characteristic that the sparse permeability field explains most of the observed responses using a finite number of bands.

[0063] After completing the orthogonal matching tracking strip selection, all selected strips are mapped back to well perimeter modeling units to form a sparse permeability field. For each selected strip, within its covered 12 radial rings and corresponding isoangular sectors, the corresponding well perimeter modeling unit is marked as a high-permeability unit and assigned an equivalent permeability factor of 20; for well perimeter modeling units not covered by any selected strips, an equivalent permeability factor of 1 is assigned. In the well depth direction, connectivity merging is performed on adjacent well depths where selected strips appear consecutively. When the consecutive occurrence length reaches 3 well depths, the strip is marked as a complete and continuous high-permeability channel; when the consecutive occurrence length is 1 to 2 well depths, it is retained as a local high-permeability cluster. The resulting sparse permeability field spatially represents a distribution structure where a small number of continuous or locally clustered high-permeability units coexist with large-scale matrix units, which is the permeability inversion result used for subsequent numerical verification.

[0064] In one alternative implementation, the time phase length or temperature drop of the strip combined template can be adjusted to adapt to different acidizing systems or different injection regimes. For example, the high-energy sonic phase can be extended to 900 seconds or the minimum temperature can be set 3 degrees Celsius below the baseline. As long as the generated strip combined template still maintains the overall characteristics of "rapid response, stable continuity, and gradual recovery," a sparse permeability field can still be generated through the same orthogonal matching tracking strip selection and residual update process. In another alternative implementation, the comprehensive similarity of multiple well depths can be considered simultaneously during strip selection to prioritize strips with continuous response along the well depth direction, thereby enhancing the sensitivity to continuous flow channels. All of the above alternative implementations adjust parameters within a predetermined framework, and technicians can choose according to formation characteristics and construction data.

[0065] Step 4: Perform numerical verification on the generated sparse permeability field and output the corresponding sparse permeability field.

[0066] In one embodiment, the generated sparse permeability field is numerically verified, and the corresponding sparse permeability field is output when a preset consistency condition is met. This includes a complete process of constructing a near-wellbore two-dimensional annular fan grid based on the sparse permeability field, performing flow and energy transfer calculations, generating a simulated joint response sequence, and comparing it with a unified observation sequence. First, the spatial pattern used for numerical calculation is determined based on the sparse permeability field obtained in step 3. The sparse permeability field has been marked on 240 well depths, 12 radial rings, and 36 isoangular sectors, with each wellbore modeling unit assigned an equivalent permeability multiple of 20 or 1. To highlight the connectivity between the high-permeability conductive strips and the matrix region in the near-wellbore plane during numerical calculation, the above information is projected into a near-wellbore two-dimensional annular fan grid to characterize the distribution of conductivity in the circumferential and radial directions of the wellbore. Specifically, an annular fan unit set consisting of 12 radial rings and 36 isoangular sectors is constructed with the wellbore as the center. Each annular fan unit corresponds to a radial thickness and an angular range. For each pair of radial rings and equiangular sectors, the proportion of layers marked as high-permeability units across 240 well depths is statistically analyzed. When the proportion of high-permeability layers reaches a preset threshold (e.g., greater than or equal to 0.1), the ring-sector unit is set as a high-permeability unit in the two-dimensional ring-sector grid, corresponding to an equivalent permeability factor of 20. When the proportion of high-permeability layers is lower than the threshold, the ring-sector unit is set as a matrix unit, corresponding to an equivalent permeability factor of 1. In this way, the three-dimensional sparse permeability field is compressed along the well depth direction into a two-dimensional ring-sector pattern that can represent the distribution of the main flow channels. On the one hand, this preserves the circumferential and radial connectivity of the high-permeability flow strips; on the other hand, it makes subsequent flow and energy transfer calculations more concentrated and efficient in the near-well plane.

[0067] Next, a numerical simulation scenario was established on a two-dimensional annular fan grid near the wellbore. The annular fan elements corresponding to the inner boundary of the wellbore were set to a constant pressure of 30 MPa, and the outermost radial ring corresponding to the outer boundary was set to a constant pressure of 20 MPa. This setting creates a stable radial pressure difference, making the direction of fluid diffusion outward or replenishment from the wellbore clear. This can drive the high-permeability conduction channels and matrix regions to generate differentiated flow behavior on the two-dimensional annular fan plane, thus reflecting the effect of the sparse permeability field in the simulation results. The fluid viscosity was set to 1 cP to represent the equivalent flow characteristics of single-phase fluid in the near-well section during acidizing and flowback processes. The total simulation duration was set to 3600 s, consistent with the time range of the aforementioned unified observation sequence, ensuring that the simulation time axis corresponds point-by-point with the field observation time axis during numerical verification.

[0068] refer to Figure 4This invention illustrates the spatial distribution of the sparse permeability enhancement field generated on a near-wellbore two-dimensional annular fan grid. The central circular area represents the wellbore, surrounded by an annular fan unit grid consisting of 12 radial rings and 36 equiangular sectors, covering the entire near-wellbore area. Each annular fan unit corresponds to a spatial location with a radial thickness and an angular range. The annular fan units marked with cross-diagonal fill lines are high-permeability conductive strip units, with an equivalent permeability multiple of 20. They are mainly distributed in the radial direction of sectors numbered 3, 8, 15, 22, 28, and 31, forming continuous or locally concentrated strip-like structures extending from the wellbore towards the surrounding rock. The blank annular fan units are matrix region units, with an equivalent permeability multiple of 1, occupying the majority of the grid area. The high-permeability conductive strip units exhibit a sparse distribution in the circumferential direction, forming obvious radial channels only in a few locations. Radially, they are mainly concentrated within the first 10 radial rings, reflecting the most significant acid etching permeability enhancement effect in the near-wellbore section. The lower left corner of the figure includes a legend clearly indicating the different filling patterns of the high-permeability guiding strips (K×20) and the matrix region (K×1). The lower right corner of the figure indicates key parameters: 12 radial ring layers, each 0.5m thick; 36 equiangular sectors, each with a 10-degree angle; 6 to 8 selected strips; and a sparsity of less than 5%. This two-dimensional ring-sector grid distribution visually demonstrates the core characteristics of a sparse permeability field, namely, the spatial pattern of a few high-permeability units coexisting with a large area of ​​matrix units, providing a clear permeability field input for subsequent numerical verification and yield enhancement assessment.

[0069] Numerical calculations employ an explicit time-progression approach. The characteristic of explicit time-progression is that within each time step, the pressure and temperature updates of each annular sector unit depend only on the state of the current unit and its adjacent units from the previous time step. This results in a clear computational path and facilitates the direct introduction of the equivalent permeability factor from the sparse permeability field while maintaining the spatially discrete structure. Specifically, within each time step, the fluid transport between units is calculated based on the equivalent permeability factor of each annular sector unit, the connectivity area with adjacent units, the radial and circumferential distances, and the current pressure difference. High-permeability units, due to their larger equivalent permeability factors, achieve higher fluid fluxes under the same pressure gradient, thus gradually forming preferred flow paths during time progression. For temperature field updates, fluid transport between annular sector units is considered an explicit energy-carrying process. High-permeability units exchange a larger volume of fluid per unit time, therefore their temperature changes are closer to the temperature of the injected fluid. Matrix units, with their smaller flow rates, experience slower temperature changes. By repeatedly advancing the process up to 3600 s, the pressure distribution, velocity distribution, and temperature distribution that evolve over time on a two-dimensional annular grid can be obtained.

[0070] To maintain consistency with the unified observation sequence, a simulated joint response sequence needs to be extracted from the numerical results. Since fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring are installed on the outer wall of the wellbore, they are most sensitive to the response of the near-wellbore annular sector. Therefore, at each time step, an acoustic energy proxy is constructed from the velocity and pressure difference information of each equiangular sector on the innermost radial ring. For example, at each time point, the changes in fluid velocity and pressure difference in each equiangular sector on the innermost ring can be weighted and superimposed to form a single acoustic disturbance intensity value, representing the acoustic energy level at the fiber location at that moment. This processing makes the acoustic energy proxy directly related to the degree of near-wellbore fluid disturbance. When the high-permeability guiding strips in the sparse permeability field provide a low-resistance path for the fluid, the velocity fluctuations in the corresponding sector and direction increase, and the accumulated acoustic energy proxy increases accordingly, thus manifesting as an energy rise, plateau, and attenuation process in the simulation results. For the temperature response, the temperature values ​​of each equiangular sector of the innermost radial ring are extracted at each time point, and then area-weighted or simply averaged to obtain the equivalent temperature change curve at the outer wall of the wellbore. This curve is used to simulate the temperature evolution recorded by fiber-optic distributed temperature monitoring. The acoustic energy proxy and temperature change curve obtained in this way correspond to the physical sensitive locations in actual monitoring, allowing the numerical simulation results to be directly used for comparative analysis.

[0071] After obtaining the time-varying acoustic energy surrogate quantity and temperature change curves, a simulated joint response sequence is constructed using the same processing procedure as in step 2. Specifically, this includes: normalizing and smoothing the acoustic energy surrogate quantity time series to ensure its time resolution and noise level are consistent with the acoustic envelope component in the unified observation sequence; baseline-aligning and smoothing the temperature change curve to maintain the same time step and processing method as the temperature change component in the unified observation sequence; and cascading the processed acoustic energy surrogate quantity and temperature change value at each time point to obtain the simulated joint response sequence. Because the construction process strictly aligns the time range, time step, and preprocessing rules, the simulated joint response sequence and the unified observation sequence have a one-to-one structural correspondence and can be directly used for difference evaluation.

[0072] The goal of numerical verification is to examine whether the sparse permeability field generated in step 3 can reproduce the key dynamic characteristics reflected in the monitoring data under a simplified physical scenario. To this end, three evaluation indicators were extracted from the unified observation sequence and the simulated joint response sequence to compare their consistency. The first indicator is the duration of the time rise phase, i.e., the time it takes for the joint response to steadily rise from the baseline level to near the peak level from the pump start-up time, used to characterize the establishment speed of the flow channel; the more connected the high-permeability flow strips, the shorter the simulated rise phase is usually. Consistency with field observations indicates that the overall flow capacity setting of the high-permeability channels in the sparse permeability field is reasonable. The second indicator is the peak occurrence time, i.e., the time when the joint response reaches its maximum value, reflecting the overall response lag in the acidification and backflow processes. A simulated peak time close to the peak time of the unified observation sequence indicates a high degree of matching between the spatial distribution of the high-permeability region and the dynamic process in the sparse permeability field. The third indicator is the peak band thickness in the well depth direction. This is achieved by identifying the length of continuous well depth intervals in the unified observation sequence where the joint response near the peak time is significantly higher than the background level, and then extracting the corresponding well depth interval thickness in the simulated joint response sequence using the same method. The close approximation of the two indicates that the sparse permeability field reasonably characterizes the connectivity length of the high-permeability conductive strips in the well depth direction. For the three indicators, an allowable deviation range of ±10% is set. When all three indicators in the simulated joint response sequence are within this allowable deviation range, it can be determined that the sparse permeability field generated in step 3 is consistent with the unified observation sequence in key dynamic characteristics, possessing the reliability of being the output of permeability inversion results for acid fracturing in carbonate reservoirs.

[0073] When the above conditions are met, the sparse permeability enhancement field generated in step 3 is directly output, and the corresponding selected strip set, well perimeter modeling unit marking results, and boundary conditions and simulation parameters verified by numerical verification are recorded as the data basis for subsequent production enhancement effect evaluation and acid fracturing parameter optimization. Since the numerical verification process is constrained by a unified observation sequence, as long as the three key indicators are all within the allowable deviation range, it can be confirmed that the sparse permeability enhancement field maintains the sparsity of the strip structure and is consistent with the monitoring data in terms of overall response, taking into account both physical rationality and computational efficiency.

[0074] In one optional implementation, segmented injection intensity settings can be applied to different isotropic sectors during the two-dimensional annular grid flow calculation. For example, sectors corresponding to the perforation cluster orientation in actual construction can be assigned higher near-wellbore injection fluxes, making the response of high-permeability conductive strips closer to the actual injection direction distribution. In another optional implementation, the simulation duration can be adjusted to 1800s or 5400s based on on-site temperature and pressure conditions, while simultaneously extracting corresponding time windows from the unified observation sequence to ensure consistent comparison benchmarks. Technicians can adjust specific numerical settings while maintaining the overall framework described above to adapt to different well conditions and construction regimes, while using the same numerical verification logic and output criteria to obtain the corresponding sparse permeability enhancement field.

[0075] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result according to substantially the same method falls within the scope of the present invention. Therefore, the scope of the present invention is defined only by the appended claims.

Claims

1. A method for inverting the permeability enhancement rate of carbonate rock thermal reservoirs through acid fracturing, characterized in that, Includes the following steps: Step 1: Install fiber-optic distributed acoustic monitoring and fiber-optic distributed temperature monitoring composite cables in the target well section to collect monitoring data, and establish a fixed spatial discretization of the well perimeter modeling unit formed by the combination of well depth, radial ring and equiangular sector Cartesian combination. Step 2: Process the collected monitoring data to construct a unified observation sequence; Step 3: Construct and establish a joint response template library corresponding to the well perimeter modeling unit, and based on the compressed sensing concept, perform orthogonal matching tracking strip selection and residual update with the constructed unified observation sequence as the target to generate a sparse permeability field; Step 4: Perform numerical verification on the generated sparse permeability field and output the corresponding sparse permeability field.

2. The method according to claim 1, characterized in that, In step 1, fiber-optic distributed acoustic monitoring uses a measurement distance of 10m, a spatial sampling interval of 1m, and a time sampling frequency of 1000Hz; fiber-optic distributed temperature monitoring uses a spatial sampling interval of 1m and a time sampling period of 1s; a well depth range with a total length of 240m is formed by extending 120m upwards and 120m downwards from the center of the perforation cluster; a near-well perimeter region with a radius of 6m is constructed starting from the outer wall of the wellbore; the aforementioned well depth range is divided into 240 well depth layers with a thickness of 1m, the aforementioned near-well perimeter region is divided into 12 radial rings with a radial thickness of 0.5m, and the circumferential region is divided into 36 equiangular sectors with an angle of 10 degrees; the well perimeter modeling unit is formed by a Cartesian combination of the 240 well depth layers, the 12 radial rings, and the 36 equiangular sectors.

3. The method according to claim 1, characterized in that, Step 2 includes: performing bandpass filtering on the raw gathers of the acquired fiber-optic distributed acoustic monitoring at each well depth, setting a lower passband limit of 40Hz and an upper passband limit of 160Hz; calculating the Hilbert envelope and performing a moving average on the filtered gathers, setting a moving average window of 3s; performing baseline correction on the acquired fiber-optic distributed temperature monitoring sequence at each well depth, taking the average value of the baseline 30s before zero time, and then smoothing it according to a 3s moving average window; at each well depth, concatenating the acoustic envelope sequence and the temperature sequence point by point according to the time index to form a joint response sequence; stacking the joint response sequences from top to bottom in the well depth direction to form a unified observation sequence.

4. The method according to claim 1, characterized in that, Step 3 includes the following steps: Strip candidate generation step: In each well depth and each isoangular sector, 12 radial rings in the current well depth are connected in series to form a high-permeability conductive strip candidate, resulting in 240×36 high-permeability conductive strip candidate strips; Joint response template library construction step: For each generated high-permeability conductive strip candidate strip, a corresponding acoustic template and temperature template are generated and cascaded to form a strip joint template to construct a joint response template library; Orthogonal matching tracking strip selection and residual update step: Using the constructed unified observation sequence as the target sequence and the constructed joint response template library as the candidate set, strip selection and residual update are performed; Sparse permeability field generation: Based on all selected strips, a sparse permeability field is generated.

5. The method according to claim 4, characterized in that, The steps for constructing the joint response template library include: setting the time evolution of the acoustic template, including dividing 3600s into three continuous acoustic phases: the first acoustic phase is from 0s to 120s, with the amplitude linearly increasing to 100% of the peak value; the second acoustic phase is from 120s to 720s, with the amplitude remaining at 100% of the peak value; and the third acoustic phase is from 720s to 3600s, with the amplitude linearly decreasing to 15% of the peak value; and setting the time evolution of the temperature template, including dividing 3600s into three... The system consists of three consecutive temperature phases: the first phase (0s to 300s) during which the temperature linearly decreases to 2 degrees Celsius below the baseline; the second phase (300s to 1200s) during which the temperature remains 2 degrees Celsius below the baseline; and the third phase (1200s to 3600s) during which the temperature linearly rises back to 0.2 degrees Celsius below the baseline. The generated acoustic template is cascaded with the generated temperature template to form a strip joint template, and the strip joint template is normalized by scaling its 95th percentile value to 1.

6. The method according to claim 5, characterized in that, The joint response template library construction steps also include: setting the radial amplitude factor of the acoustic template: for the 12 radial rings within the candidate high-permeability conductive strip, a relative amplitude factor array [1.00, 0.70, 0.50, 0.35, 0.25, 0.18, 0.13, 0.09, 0.06, 0.04, 0.03, 0.02] is sequentially assigned; at the corresponding well depth, the set acoustic template time evolution is multiplied by the relative amplitude factor ring by ring along the radial direction and then summed along the radial direction, and a longitudinal smoothing effect caused by a 10m measurement distance is applied. Specifically, this includes: using a moving average with a length of 10 well depths in the well depth direction to obtain the level... The acoustic template for the cascade is set; the radial influence factor of the temperature template is set: for the 12 radial rings within the candidate high-permeability conductive strip, the relative influence factor array [1.00,0.85,0.72,0.60,0.50,0.42,0.35,0.29,0.24,0.20,0.17,0.15] is assigned sequentially; at the corresponding well depth, the time evolution of the set temperature template is multiplied by the aforementioned relative influence factor ring by ring along the radial direction and then summed along the radial direction, and the longitudinal smoothing effect caused by 1m spatial sampling is applied, specifically including: using a moving average with a length of 3 well depths in the well depth direction to obtain the temperature template for cascading.

7. The method according to claim 4, characterized in that, The orthogonal matching pursuit-based strip selection and residual update steps include: normalizing the constructed unified observation sequence strip by strip according to well depth to obtain an initial residual sequence; in each round, calculating the correlation coefficient between the aforementioned initial residual sequence and each joint template in the joint response template library at the corresponding well depth, using the average value of the calculated correlation coefficient over time as the similarity; selecting the strip with the highest similarity from the candidate set as the selected strip for this round, and removing the template corresponding to the selected strip from the candidate set; in the well depth where the selected strip is located, linearly scaling the strip joint template by amplitude so that the 95th percentile value of the strip joint template is consistent with the 95th percentile value of the initial residual at the corresponding well depth; subtracting the aligned strip joint template from the initial residual at the corresponding well depth point by point to obtain a new residual sequence; and performing a moving average of 5 well depths on the new residual sequence for the two well depths above and below the well depth where the selected strip is located, to achieve orthogonalization.

8. The method according to claim 7, characterized in that, The termination criteria for the orthogonal matching pursuit strip selection and residual update steps include: the root mean square value of the calculated new residual sequence over all well depths and all times; termination when the calculated root mean square value is less than 10% of the root mean square value of the initial residual sequence; or termination when the number of rounds reaches 50.

9. The method according to claim 4, characterized in that, The steps for generating a sparse permeability field include: for all selected strips, marking the corresponding well perimeter modeling units as high permeability units within the 12 radial rings and corresponding isoangular sectors covered by the selected strips, and assigning them an equivalent permeability multiple of 20; assigning an equivalent permeability multiple of 1 to unmarked well perimeter modeling units; and performing connectivity merging on adjacent well depths where the selected strips appear consecutively in the well depth direction, with the rule that consecutive occurrences of 3 well depths are treated as continuous strips, and those with a length of 1 to 2 well depths are treated as isolated strips.

10. The method according to claim 1, characterized in that, Step 4 includes: establishing a near-wellbore two-dimensional annular fan grid on the generated sparse permeability field, setting the inner boundary of the wellbore to a constant pressure of 30 MPa, the outer boundary to a constant pressure of 20 MPa, the fluid viscosity to 1 cP, and the simulation duration to 3600 s; using explicit time-progression to perform flow and energy transfer calculations, extracting the acoustic energy proxy and temperature change curves at the fiber optic location, and generating a simulated joint response sequence; comparing the generated simulated joint response sequence with the constructed unified observation sequence on three indicators: duration of the time rise phase, peak occurrence time, and peak band thickness in the well depth direction, with an allowable deviation of ±10% for each of the three; when the aforementioned allowable deviation conditions are met, outputting the sparse permeability field generated in Step 3.

Images