A method for identifying tight sandstone reservoirs by fusing multi-curve prediction errors

Abstract

The present application relates to a kind of compact sandstone reservoir identification method based on multi-curve fusion prediction error, selects the standardized well logging data including sliding window end point and its previous depth point to form multiple groups of data, fitting to obtain the natural gamma prediction error curve and deep resistivity prediction error curve of target layer section of target block;Deep resistivity prediction error curve is corrected, make the response of corrected deep resistivity prediction error curve and natural gamma prediction error curve to effective gas-bearing reservoir uniform;Natural gamma and corrected deep resistivity weight coefficient are calculated based on geological adaptive weight model, and then fusion prediction error curve is obtained;The data of well calibrated in target block are used to determine the fusion error threshold of effective gas-bearing reservoir;Based on fusion prediction error curve and fusion error threshold, the automatic identification, layering and evaluation of compact sandstone effective gas-bearing reservoir of target layer section of target block are completed.The present application has strong industrialization popularization value.

Description

Technical Field

[0001] This invention belongs to the field of oil and gas exploration and development technology, and in particular relates to a method for identifying tight sandstone reservoirs based on multi-curve fusion prediction error. Background Technology

[0002] The Upper Paleozoic strata of the Yan'an gas field contain a large-scale marine-continental transitional sedimentary system. The sandstone bodies are thin, interbedded frequently, and have complex vertical stacking relationships, resulting in extremely high reservoir heterogeneity. Identifying and accurately characterizing high-quality gas-bearing sandstones is extremely challenging. Traditional well logging interpretation relies on human experience and single-curve characteristics, making it susceptible to interference from factors such as lithology, fluids, formation water salinity, and wellbore conditions. This leads to highly ambiguous reservoir identification and low accuracy, failing to meet the demands of industrial exploration for thin, interbedded, tight sandstone.

[0003] INPEFA technology can amplify formation anomaly information by constructing the prediction error between well logging curves and their trend values, and has been gradually applied to cycle division and reservoir identification. However, existing technologies have the following obvious drawbacks: First, traditional commercial INPEFA is mostly an integral accumulation type, with the value range forcibly normalized to 0-1 and all positive values. After accumulation and smoothing, the signal is significantly weakened, resulting in insufficient sensitivity to thin-layered and weakly anomalous reservoirs, and unclear geological indications. Second, the natural gamma ray (GR) and deep resistivity (RT) responses to effective gas-bearing reservoirs have naturally opposite polarities. Negative anomalies in GR prediction errors indicate sandstone-developed strata, while positive anomalies in RT prediction errors indicate well-gas-bearing strata. The two have contradictory polarities, different amplitude scales, and inconsistent reference surfaces, making direct weighted fusion impossible. Forced fusion easily leads to signal cancellation and reservoir identification distortion, which is a long-standing technical bottleneck in this field. Third, existing multi-curve fusion weights are mostly assigned manually based on experience or calculated using a single variance contribution, which is highly subjective and lacks constraints from prior geological information, resulting in poor applicability to different sedimentary blocks. Fourth, reservoir discrimination thresholds are mostly set manually, without a block-adaptive statistical mechanism, making it difficult to meet the requirements of industrial exploration in terms of identification accuracy. To address the aforementioned technical shortcomings, an innovative technical solution is urgently needed to achieve a breakthrough.

[0004] To address this, the present invention proposes a multi-curve fusion prediction error reservoir identification method that features unified direction, objective weighting, threshold segmentation, and single-point residual type. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the existing technology and solve the technical problems of the traditional INPEFA technology, such as contradictory polarity of logging curve response, subjective fusion weight allocation, poor thin-layer reservoir identification effect, and weak block applicability. It provides a method for identifying tight sandstone reservoirs based on multi-curve fusion prediction error, and realizes high-precision, adaptive, and quantitative identification of high-quality gas-bearing reservoirs in tight sandstone under marine-continental transitional facies and various sedimentary environments.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] A method for identifying tight sandstone reservoirs based on multi-curve fusion prediction error is as follows: Obtain standardized logging data for the target layer of the target block; Using a preset depth point as the end point of the sliding window, standardized logging data containing the end point of the sliding window and its preceding depth points are selected to form multiple sets of data. The fitted prediction normalized value and single-point residual prediction error of each depth point within the sliding window are obtained by fitting. This leads to the natural gamma prediction error curve and deep resistivity prediction error curve of the target layer in the target block. The deep resistivity prediction error curve is corrected to make the response of the corrected deep resistivity prediction error curve and the natural gamma prediction error curve to the effective gas-bearing reservoirs consistent. The natural gamma and the corrected deep resistivity weight coefficient are calculated based on the geological adaptive weight model, and then the fusion prediction error curve is obtained. Using data from calibrated wells within the target block, determine the fusion error threshold for effective gas-bearing reservoirs; Based on the fusion prediction error curve and fusion error threshold, the automatic identification, stratification and evaluation of effective gas-bearing reservoirs in tight sandstone of the target block and target segment are completed.

[0008] The specific process for obtaining the standardized well logging data is as follows: Well logging data of the target block and target layer are collected, including natural gamma and deep resistivity data; outlier removal, missing value completion and extreme value normalization are performed on the well logging data to obtain standardized well logging data; The formula for extreme value normalization is as follows: (1) In the formula: For the first Normalized values ​​of logging data at depth points, with values ​​ranging from [0,1]; For the first Original logging values ​​at each depth point; and These are the minimum and maximum values ​​of logging data for the entire well section of the target block, respectively.

[0009] The specific process for the natural gamma prediction error curve and the deep resistivity prediction error curve is as follows: Set the length of the sliding window The value is an integer between 10 and 30; using a preset depth point as the endpoint of the sliding window, standardized logging data containing the endpoint of the sliding window and its preceding depth points are selected to form multiple sets of data, the number of which is related to the length of the sliding window. equal; The slope was obtained by fitting the data using the least squares method from multiple sets of data. The intercept b is calculated using the following formula: (2) (3) In the formula: The length of the sliding window; The depth point number within the sliding window; For the first in the sliding window Normalized values ​​of logging data at each depth point; b is the slope of the fitted line; b is the intercept of the fitted line. Then calculate the first [number] within the sliding window. Normalized values ​​of the fitted prediction at each depth point and single-point residual prediction error: (4) (5) In the formula: For the first in the sliding window Normalized values ​​of the fitted predictions for each depth point; For the first in the sliding window Single-point residual prediction error at each depth point; This leads to the target block target segment number. Normalized values ​​of the fitted prediction at each depth point and single-point residual prediction error: (15) (16) In the formula: For the first Normalized values ​​of the fitted predictions for each depth point; For the first Single-point residual prediction error at each depth point; This leads to the natural gamma prediction error curve and deep resistivity prediction error curve of the target layer in the target block.

[0010] The specific process for correcting the deep resistivity prediction error curve is as follows: the deep resistivity prediction error curve is reversed in polarity, its amplitude is calibrated, and it is aligned with the baseline to obtain the corrected deep resistivity prediction error curve.

[0011] The specific process of correcting the deep resistivity prediction error curve is as follows: (6) In the formula: For the first The deep resistivity prediction error after correction at each depth point; For the first Single-point residual prediction error of deep resistivity at a depth point; This is the amplitude calibration coefficient; Baseline calibration coefficients; This leads to the obtained correction curve for deep resistivity prediction. in, Amplitude calibration coefficient The specific calculation process is as follows: (7) Baseline calibration coefficient The specific calculation process is as follows: (8) In the formula: The natural gamma prediction error curve; The deep resistivity prediction error curve; The standard deviation of the natural gamma prediction error curve; The standard deviation of the original prediction error curve for deep resistivity; This is the arithmetic mean of the natural gamma prediction error curve; This is the arithmetic mean of the original prediction error curves for deep resistivity.

[0012] It also includes the amplitude calibration coefficient when the target block is a marine-continental transitional facies tight sandstone. Baseline calibration coefficient ; .

[0013] The process for obtaining the natural gamma and deep resistivity weighting coefficients is as follows: (9) (10) , (11) (12) (13) In the formula: The variance contribution of the natural gamma prediction error curve; To correct the variance contribution of the deep resistivity prediction error curve; The variance of the natural gamma prediction error curve; To correct the variance of the deep resistivity prediction error curve; Geological consistency coefficient; This represents the Pearson correlation coefficient used to calculate the natural gamma prediction error curve and the corrected deep resistivity prediction error curve. The standard deviation of the raw natural gamma ray logging data; The standard deviation of the original deep resistivity logging data; The dominant lithology coefficient; This is the formation fluid correction factor; Porosity; The content of clay; This represents the average porosity within the target block. This represents the average mud content within the target block; The natural gamma weighting coefficient; To correct the deep resistivity weighting coefficient.

[0014] The specific process for obtaining the fusion prediction error curve is as follows: (14) In the formula: To fuse the prediction error curves.

[0015] The specific process for obtaining the fusion error threshold is as follows: Select calibration wells covering at least one type of stratum within the target block, and match the fusion prediction error with the depth of core physical properties and gas production capacity data; Statistically analyze the proportion of effective gas-bearing reservoirs and production capacity patterns according to the error interval, and then determine the fusion error threshold that meets the requirements of effective gas-bearing reservoirs.

[0016] The technical effects of this invention are as follows: This invention abandons the traditional integral-cumulative INPEFA model and adopts a single-point residual prediction error, which significantly improves the sensitivity of response to thin and weakly anomalous reservoirs, providing clear geological indications. Addressing the industry-wide problem of opposite polarity in well logging curve responses, it presents an original formation response consistency mapping model. Through a triple mechanism of polarity reversal, amplitude calibration, and baseline alignment, it achieves effective fusion of multiple curves, solving the problems of signal cancellation and identification distortion. A geologically adaptive weighting model is constructed, integrating data features and prior geological information. The weighting calculation is entirely objective and without subjective intervention, making it applicable to various sedimentary blocks. Using block-specific statistical thresholds improves the adaptability and accuracy of reservoir identification. Field verification shows a reservoir identification accuracy of no less than 92%, demonstrating strong industrial application value. Attached Figure Description

[0017] Figure 1 This is a flowchart of the present invention.

[0018] Figure 2 This is a schematic diagram comparing the natural gamma prediction error curve, the deep resistivity prediction error curve, and the fused prediction error curve in an embodiment of the present invention.

[0019] Figure 3 This is a statistical analysis chart of the fusion error threshold calibration in a specific application case of the present invention.

[0020] Figure 4 This is a comprehensive result diagram of actual well reservoir identification in a specific application case of the present invention. Detailed Implementation

[0021] A method for identifying tight sandstone reservoirs based on multi-curve fusion prediction error is presented below.

[0022] Step 1: Well Logging Data Acquisition and Preprocessing: Collect natural gamma ray (GR), deep resistivity (RT), and porosity continuously at the target well depth. mud content Conventional logging data is processed by removing outliers, filling in missing values, and normalizing extreme values ​​according to formula (1) to obtain standardized logging data.

[0023] Step 2: Obtain the natural gamma prediction error curve and deep resistivity prediction error curve of the target layer in the target block: Set the length of the sliding window The value is an integer between 10 and 30; using a preset depth point as the endpoint of the sliding window, standardized logging data containing the endpoint of the sliding window and its preceding depth points are selected to form multiple sets of data, the number of which is related to the length of the sliding window. equal; Based on equations (2) and (3), the slope is fitted using the least squares method for multiple sets of data. And the intercept b, and then according to equation (4) the first digit within the sliding window. Normalized value of fitted prediction at each depth point Then, based on equation (5), the single-point residual prediction error of this depth point is obtained; Adjust the preset depth point, and then obtain the target layer of the target block according to equation (15). Normalized value of fitted prediction at each depth point Based on equation (16), the single-point residual prediction error at this depth point is obtained. ; Finally, the natural gamma prediction error curve and the deep resistivity prediction error curve were obtained.

[0024] Step 3: Correct the deep resistivity prediction error curve: Statistical calculations were performed on the logging data of the target well section, and the amplitude calibration coefficient was obtained according to equation (7). The baseline calibration coefficients are obtained according to equation (8). Therefore, based on equation (6), we obtain the first... Deep resistivity prediction error after depth point correction This leads to the obtained correction curve for deep resistivity prediction.

[0025] Step 4: Construct the fusion prediction error curve: The variance contribution of the natural gamma prediction error curve is obtained according to equation (9). and the variance contribution of the corrected deep resistivity prediction error curve The geological consistency coefficient is obtained according to equation (10). The lithological dominant coefficient is obtained according to equation (11). Then, the natural gamma weighting coefficient is obtained according to equation (12). The corrected deep resistivity weighting coefficient is obtained according to equation (13). Then, the fusion prediction error curve is obtained according to equation (14).

[0026] Step 5: Determine the fusion error threshold: Select calibration wells within the target block that cover at least one type of stratum, and match the fusion prediction error with the depth of core physical properties and gas production data; statistically analyze the proportion of effective gas-bearing reservoirs and production patterns according to the error interval, and then determine the fusion error threshold that meets the requirements of effective gas-bearing reservoirs.

[0027] Step 6: Automatic identification, stratification and evaluation of effective gas-bearing reservoirs in tight sandstone of the target block: reservoirs that meet the fusion error threshold are considered effective gas-bearing reservoirs.

[0028] Specific Application Case—Taking the Y177 well in the eastern part of the Yan'an Gas Field as the research object, this study focuses on the identification of tight sandstone reservoirs in the 2620.0m–2680.0m section of the Shanxi Formation. This section features well-developed interbedded sandstone and mudstone, with alternating sandstone and mudstone interlayers. The reservoir as a whole exhibits characteristics of low porosity, low permeability, and strong heterogeneity. Based on measured data, the mudstone section has a higher mud content, ranging from 25% to 42%, while the mudstone section has a mud content that decreases to 8%–22%. The overall porosity ranges from 1.3% to 9.5%, with the high-quality sandstone section having a porosity concentrated between 6.5% and 9.5%, and the mudstone section generally having a porosity below 2.0%. The overall permeability ranges from 0.018 to 0.683 mD, with relatively high permeability in the core sandstone section and extremely low permeability in the mudstone section. Due to the frequent interbedded sandstone and mudstone layers in this section, the effective gas-bearing reservoir is thin and difficult to identify. This invention's method is used for the precise identification of effective gas-bearing reservoirs in tight sandstone.

[0029] A method for identifying tight sandstone reservoirs based on multi-curve fusion prediction error is presented below.

[0030] Step 1: Well Logging Data Acquisition and Preprocessing: Collect natural gamma ray (GR), deep resistivity (RT), and porosity continuously at the target well depth. mud content A total of 481 sets of conventional logging data were collected. Abnormal jump points caused by logging instrument errors or wellbore collapse were removed. Linear interpolation was used to complete the logging data for missing depth points, ensuring the data sequence was continuous and complete. The pre-processed raw logging data were normalized according to equation (1) to eliminate dimensional differences. Statistically, the original natural gamma ray (GR) values ​​for the target well section ranged from 28 API to 137 API, and the original deep resistivity (RT) values ​​ranged from 15 Ω·m to 683 Ω·m. After extreme value normalization, all data were normalized to the [0,1] interval for easier subsequent prediction error calculation. This step corresponds to... Figure 1 The well logging data preprocessing stage.

[0031] Step 2: Obtain the natural gamma prediction error curve and deep resistivity prediction error curve of the target layer in the target block: Set the length of the sliding window The value is an integer between 10 and 30; using a preset depth point as the endpoint of the sliding window, standardized logging data containing the endpoint of the sliding window and its preceding depth points are selected to form multiple sets of data, the number of which is related to the length of the sliding window. equal; In this specific application case, the sliding window length W=15 is set. Taking the 2649.5m depth point as an example, 15 sets of normalized data points are selected from the 2649.5m depth point and the preceding depth point. The slope is calculated using the least squares fitting formula. And the intercept b, thus obtaining the first within the sliding window Normalized value of fitted prediction at each depth point This leads to the single-point residual prediction error at that depth point; The natural gamma prediction error at this depth point was calculated. = -0.055, a negative value indicates that the lithology at this depth is pure and sandstone is well-developed; the deep resistivity prediction error at this depth point. =0.059, a positive value indicates high resistivity and good gas content. The two have opposite response polarities and cannot be directly weighted and combined. Adjust the preset depth point to obtain the natural gamma prediction error curve and deep resistivity prediction error curve of the target layer of the target block.

[0032] Step 3: Correct the deep resistivity prediction error curve: Statistical calculations were performed on the logging data of the target well section to obtain the amplitude calibration coefficient. Approaching 1, baseline calibration coefficient Approaching 0, it conforms to the geological characteristics of dense sandstone in the marine-continental transitional facies. To simplify calculations and ensure accuracy, it is simplified to... This leads to the correction of the deep resistivity prediction error curve.

[0033] Step 4: Construct the fusion prediction error curve: calculate ; calculate ; Combined with the average porosity within the target block The average mud content within the target block Calculations yielded , ; and then calculate to obtain , ; This leads to the fusion prediction error curve. For details of the fusion prediction error curve, please refer to [link / reference]. Figure 2 .

[0034] Step 5: Determine the fusion error threshold: Twenty wells with complete data from the Upper Paleozoic strata of the Yan'an Gas Field were selected, covering all types of effective gas-bearing reservoirs, dry layers, and mudstone formations. The fusion prediction error was deeply matched with core properties and gas testing production data. The proportion of effective gas-bearing reservoirs and production patterns were statistically analyzed according to the error interval. Verification showed that when the fusion prediction error... When the effective gas-bearing reservoir ratio is less than -0.04, the effective gas-bearing reservoir ratio shall not be less than 92%, and the daily gas production during testing shall not be less than 0.5 × 10⁻⁴. 4 m 3 It fully meets the criteria for effective gas-bearing reservoir development; therefore, the target fusion error threshold is determined to be [value missing]. < -0.04; Figure 3 A statistical chart for fusion error threshold calibration is generated, with fusion prediction error as the horizontal axis and the proportion of effective gas-bearing reservoirs and average daily gas production as the vertical axis, visually displaying the changes in fusion prediction error threshold. When the threshold is less than -0.04, the reservoir quality and productivity show a significant upward trend, proving that the fusion error threshold setting is scientific, reasonable, and in line with reality.

[0035] Step 6: Automatic identification, stratification, and evaluation of effective gas-bearing reservoirs in tight sandstone within the target block: Based on the fusion prediction error curve and fusion error threshold, the target well section of well Y177 was identified point by point: the fusion prediction errors of the 2647.3m~2651.2m and 2643.8m~2646.6m sections were both less than -0.04, and were determined to be effective gas-bearing reservoirs with a cumulative effective thickness of 6.7m. The remaining sections were determined to be dry layers or non-reservoir mudstone. For the complete reservoir identification results, curve correspondences, and stratification annotations, please refer to the appendix. Figure 4 .

[0036] Gas testing confirmed that the daily gas production of this well section reached 0.61 × 10⁻⁶. 4 m 3 / d, core analysis showed a porosity of 7.7% and a gas saturation of 78.2%. The identification results were completely consistent with the gas production capacity and core analysis results, and the reservoir identification accuracy reached 92%, meeting the requirements for on-site exploration and development evaluation.

[0037] The logging data of the target formations from 15 wells in the study area were entered into Excel in a standardized manner. A VBA macro program was then used to automatically complete steps 1-6 above, enabling batch and accurate identification of reservoirs in multiple wells in the region, and providing reliable data support for horizontal well trajectory design and drilling target selection.

Claims

1. A method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors, characterized in that, The method is as follows: Obtain standardized logging data for the target layer of the target block; Using a preset depth point as the end point of the sliding window, standardized logging data containing the end point of the sliding window and its preceding depth points are selected to form multiple sets of data. The fitted prediction normalized value and single-point residual prediction error of each depth point within the sliding window are obtained by fitting. This leads to the natural gamma prediction error curve and deep resistivity prediction error curve of the target layer in the target block. The deep resistivity prediction error curve is corrected to make the response of the corrected deep resistivity prediction error curve and the natural gamma prediction error curve to the effective gas-bearing reservoirs consistent. The natural gamma and the corrected deep resistivity weight coefficient are calculated based on the geological adaptive weight model, and then the fusion prediction error curve is obtained. Using data from calibrated wells within the target block, determine the fusion error threshold for effective gas-bearing reservoirs; Based on the fusion prediction error curve and fusion error threshold, the automatic identification, stratification and evaluation of effective gas-bearing reservoirs in tight sandstone of the target block and target segment are completed.

2. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 1, characterized in that, The specific process for obtaining the standardized well logging data is as follows: Well logging data of the target block and target layer are collected, including natural gamma and deep resistivity data; outlier removal, missing value completion and extreme value normalization are performed on the well logging data to obtain standardized well logging data; The formula for extreme value normalization is as follows: In the formula: For the first Normalized values ​​of logging data at depth points, with values ​​ranging from [0,1]; For the first Original logging values ​​at each depth point; and These are the minimum and maximum values ​​of logging data for the entire well section of the target block, respectively.

3. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 1, characterized in that, The specific process for the natural gamma prediction error curve and the deep resistivity prediction error curve is as follows: Set the length of the sliding window The value is an integer between 10 and 30; using a preset depth point as the endpoint of the sliding window, standardized logging data containing the endpoint of the sliding window and its preceding depth points are selected to form multiple sets of data, the number of which is related to the length of the sliding window. equal; The slope was obtained by fitting the data using the least squares method from multiple sets of data. The intercept b is calculated using the following formula: In the formula: The length of the sliding window; The depth point number within the sliding window; For the first in the sliding window Normalized values ​​of logging data at each depth point; b is the slope of the fitted line; b is the intercept of the fitted line. Then calculate the first [number] within the sliding window. Normalized values ​​of the fitted prediction at each depth point and single-point residual prediction error: In the formula: For the first in the sliding window Normalized values ​​of the fitted predictions for each depth point; For the first in the sliding window Single-point residual prediction error at each depth point; This leads to the natural gamma prediction error curve and deep resistivity prediction error curve of the target layer in the target block.

4. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 1, characterized in that, The specific process for correcting the deep resistivity prediction error curve is as follows: the deep resistivity prediction error curve is reversed in polarity, its amplitude is calibrated, and it is aligned with the baseline to obtain the corrected deep resistivity prediction error curve.

5. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction error according to claim 1, characterized in that, The specific process for correcting the deep resistivity prediction error curve is as follows: In the formula: For the first The deep resistivity prediction error after correction at each depth point; For the first Single-point residual prediction error of deep resistivity at a depth point; This is the amplitude calibration coefficient; Baseline calibration coefficients; This leads to the obtained correction curve for deep resistivity prediction.

6. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 5, characterized in that, The Amplitude calibration coefficient The specific calculation process is as follows: ; Baseline calibration coefficient The specific calculation process is as follows: ; In the formula: The natural gamma prediction error curve; The deep resistivity prediction error curve; The standard deviation of the natural gamma prediction error curve; The standard deviation of the original prediction error curve for deep resistivity; This is the arithmetic mean of the natural gamma prediction error curve; This is the arithmetic mean of the original prediction error curves for deep resistivity.

7. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 5, characterized in that, It also includes the amplitude calibration coefficient when the target block is a marine-continental transitional facies tight sandstone. Baseline calibration coefficient ; .

8. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 1, characterized in that, The process of obtaining the natural gamma and deep resistivity weighting coefficients is as follows: ， In the formula: The variance contribution of the natural gamma prediction error curve; The variance of the natural gamma prediction error curve; To correct the variance of the deep resistivity prediction error curve; Geological consistency coefficient; This represents the Pearson correlation coefficient used to calculate the natural gamma prediction error curve and the corrected deep resistivity prediction error curve. The standard deviation of the raw natural gamma ray logging data; The standard deviation of the original deep resistivity logging data; The dominant lithology coefficient; This is the formation fluid correction factor; Porosity; The content of clay; This represents the average porosity within the target block. This represents the average mud content within the target block; The natural gamma weighting coefficient; To correct the deep resistivity weighting coefficient.

9. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 8, characterized in that, The specific process for obtaining the fusion prediction error curve is as follows: In the formula: To fuse the prediction error curves.

10. The method for identifying tight sandstone reservoirs based on multi-curve fusion prediction errors according to claim 1, characterized in that, The specific process for obtaining the fusion error threshold is as follows: Select calibration wells covering at least one type of stratum within the target block, and match the fusion prediction error with the core physical properties and gas production capacity data; Statistically analyze the proportion of effective gas-bearing reservoirs and production capacity patterns according to the error interval, and then determine the fusion error threshold that meets the requirements of effective gas-bearing reservoirs.

Images