Method for predicting production of horizontal well in sandstone gas reservoir based on recombination of logging curve characteristics

Abstract

The present invention discloses a method for predicting the production of horizontal wells in sandstone gas reservoirs based on the recombination of logging curve features, including: decomposing the GR curve based on the logging data of the test horizontal well to obtain high-frequency and low-frequency curve data; combining the acoustic logging curve data with the low-frequency curve data to establish a standard for dividing gas-producing reservoirs for reservoir division; calculating the single-point productivity within the high-quality reservoir of the test horizontal well using natural gas test data, combining multiple test horizontal wells, performing regression analysis and fitting on the relationship between different influencing factors and single-point productivity, and then establishing a single-point productivity prediction formula according to the fitting relationship and high-frequency curve data: based on the logging data of the horizontal well to be predicted, using the single-point productivity prediction formula to predict the productivity of the horizontal well to be predicted. The present invention only needs a small amount of natural gas production data of the tested horizontal wells as calibration for the horizontal section of the horizontal well in the sandstone gas reservoir, and can accurately predict the natural gas production of the horizontal section of the newly drilled horizontal well based on the conventional logging curve.

Description

Technical Field

[0001] The present invention relates to the technical field of natural gas exploration and development, and particularly relates to a method for predicting the production of horizontal wells in sandstone gas reservoirs based on the recombination of logging curve characteristics. Background Technique

[0002] In the field of natural gas exploration and development, horizontal wells have become a widely used technology. Horizontal wells can significantly improve the productivity of reservoirs and the recovery rate of natural gas. However, before perforation and fracturing, it is necessary to accurately predict the productivity of the horizontal section of the horizontal well. The accurate prediction of the natural gas productivity of the horizontal section of the horizontal well is crucial for formulating production plans, optimizing resource management, and enhancing economic benefits.

[0003] In the field of natural gas exploration and development, there are various existing methods for predicting natural gas productivity. In the professional direction of oil and gas geology, the physical properties and scale of underground reservoirs are often understood through geological exploration and core analysis, and numerical simulation methods are used to predict the natural gas production of reservoirs. The disadvantages of this method are that it requires professional numerical simulation software, takes a long time, and relies on a large amount of core experimental analysis data. In the professional direction of oil and gas development, the historical production fitting of wells is often carried out through natural gas production data, and further the future natural gas production of wells is predicted. The disadvantage of this method is that it requires the historical production data of the well to be predicted. If the well to be predicted is a newly completed well, the prediction effect is poor. In the professional direction of oil and gas artificial intelligence, machine learning or deep learning algorithms are often used to train models through the production data of a large number of production wells, so as to predict natural gas production. The disadvantage of this method is that it requires a large amount of production data of production wells, and general gas reservoirs cannot meet this condition. Since the above three types of methods all have disadvantages, these disadvantages are more obvious in horizontal wells, resulting in greater limitations in their use. There is an urgent need to develop a method for predicting the production of horizontal wells in sandstone gas reservoirs with high accuracy and few usage limiting conditions. Summary of the Invention

[0004] The purpose of the present invention is to provide a method for predicting the production of horizontal wells in sandstone gas reservoirs based on the recombination of logging curve characteristics. For the horizontal section of the horizontal well in the sandstone gas reservoir, only a small amount of natural gas production data of the tested horizontal wells is required for calibration, and the natural gas production of the horizontal section of the newly drilled horizontal well can be accurately predicted based on the conventional logging curves.

[0005] To achieve the above purpose, the present invention provides the following solution:

[0006] A method for predicting the production of horizontal wells in sandstone gas reservoirs based on the recombination of logging curve characteristics, the method comprising the following steps:

[0007] Obtain the logging data of the horizontal well to be predicted, the logging data of the tested horizontal wells, and the natural gas test data;

[0008] Based on the logging data of the test horizontal well, the GR curve is decomposed to obtain the high-frequency and low-frequency curve data of the test horizontal well;

[0009] The acoustic logging curve data of the test horizontal well is obtained from the logging data of the test horizontal well, combined with the low-frequency curve data of the test horizontal well, a gas-producing reservoir division standard is established, and the high-quality reservoir of the test horizontal well is determined;

[0010] Using the natural gas test data to calculate the single-point productivity in the high-quality reservoir of the test horizontal well, combining multiple test horizontal wells, performing regression analysis and fitting on the relationship between different influencing factors and the single-point productivity, and then establishing a single-point productivity prediction formula according to the fitting relationship and the high-frequency curve data of the test horizontal well:

[0011] Based on the logging data of the horizontal well to be predicted, using the single-point productivity prediction formula, predict the productivity of the horizontal well to be predicted.

[0012] Furthermore, the logging data of the horizontal well to be predicted and the logging data of the test horizontal well both include natural gamma logging curves, deep and shallow resistivity logging curves, acoustic logging curves, and data of porosity, density, water saturation, gas saturation, permeability, and shale content; the natural gas test data includes the test depth range and test flow rate.

[0013] Furthermore, the GR curve decomposition is specifically to perform eigen-decomposition on the GR curve to obtain curves with different frequency components, and according to the curve average relative error standard, recombine to obtain the high-frequency curve and the low-frequency curve.

[0014] Furthermore, the GR curve decomposition specifically includes the following steps:

[0015] S1, determine all the maximum and minimum points of the GR curve, fit the maximum envelope curve GR max and the minimum envelope curve GR min , average the two envelope curves to obtain the average line, and subtract the average line from the original GR curve to obtain a new curve GR h1 without low frequency;

[0016] S2, determine whether GR h1 meets conditions a and b:

[0017]

[0018] In the formula: n represents the number of upper extreme points of GR h1 dimensionless; p represents the number of zero-crossing points of GR h1 dimensionless; h max represents the upper envelope curve of GR h1 in API; h min represents GRh1 Lower envelope, API;

[0019] If not satisfied, use GR h1 Instead of GR 原始 , repeat step S1 until GR h1 is satisfied. If satisfied, at this time GR h1 is GR 原始 The first-order component of the curve is renamed GR c1 , use GR 原始 to subtract GR c1 , obtaining a new curve GR with high-frequency components removed r1 ;

[0020] S3, use GR r1 to replace GR 原始 curve and repeat steps S1 - S2 to obtain the second-order component GR c2 , and repeat this process until the nth-order component GR cn , and when its corresponding nth-order residue GR r is monotonic or constant, the decomposition process stops; finally GR 原始 curve is decomposed into the sum of curves with different frequencies;

[0021] S4, after the GR curve is decomposed, n curves with different frequencies GR ci and a residue trend line GR r are obtained. From GR r and n GR ci curves, synthetic curves from low frequency to high frequency are respectively combined;

[0022] S5, calculate the average relative error value for all synthetic curves and GR 原始 curve;

[0023] When the average relative error value of the curve is less than 0.1 and greater than 0.05, the correlation between the curve and the formation characteristics is relatively high and the high-frequency characteristics are not obvious, and it is more matched with the formation characteristics in terms of spectral characteristics, thus being suitable as the GR 低频 curve;

[0024] S6, the first j (i < j < n) curves can be composed to obtain GR 高频 The composition of the curve is as follows:

[0025] GR 高频 = GR c1 + GR c2 + ··· + GR cj (8)

[0026] In the formula: GR 高频 represents GR 原始High-frequency characteristic curve obtained by curve decomposition and recombination, API;

[0027] For S7, the GR can be obtained by combining the last n - j curves 低频 The components of the curve are:

[0028] GR 低频 = GR cj+1 + GR cj+2 + ··· + GR cn + GR rn (9)

[0029] In the formula: GR 低频 represents the low-frequency characteristic curve obtained by curve decomposition and recombination of GR, API. 原始 Curve decomposition and recombination of low-frequency characteristic curve, API.

[0030] Furthermore, the gas-producing reservoir division standard is expressed as:

[0031]

[0032]

[0033] In the formula: I GR represents the relative value of natural gamma, dimensionless; GR mi , GR ma represent the natural gamma values corresponding to pure sandstone and pure shale in this area, API; GR 低频 represents GR 原始 Curve decomposition and recombination of low-frequency characteristic curve, API; α represents the correlation coefficient, dimensionless; c represents the empirical coefficient, dimensionless, generally 3.7; Δt, Δt min and Δt max respectively represent the acoustic time difference curve response value, response minimum value and response maximum value, μs / m;

[0034] If d6 < α, it is classified as a non-reservoir; if d4 < α < d6, it is classified as a poor reservoir; if α < d4, it is classified as a high-quality reservoir; where d6 represents the first set threshold, d4 represents the second set threshold, and d6 > d4.

[0035] Furthermore, the single-point productivity within the high-quality reservoir of the test horizontal well is calculated using natural gas test data. Combining multiple test horizontal wells, after trend line analysis and modeling of the relationship between different influencing factors and single-point productivity, factors with good correlation effects are selected for regression analysis and fitting. Then, according to the fitting relationship and the high-frequency curve data of the test horizontal well, a single-point productivity prediction formula is established, specifically including:

[0036]

[0037]

[0038] Q 预测 = C1(F1) + C2(F2) + … + C m (F m ) + B + GR 高频 (14)

[0039] Where: Q 125 represents the single - point production capacity, m³ / day; Q t represents the normal test flow rate in the natural gas test data, 10,000 m³ / day; H q represents the thickness of the high - quality reservoir, m; Q1, Q2 to Q n represent the relationship between different factors and production capacity impact, m³ / day; a1 to a n represent the slope values of the trend line, dimensionless; b1 to b n represent the intercept values of the trend line, dimensionless; F1 to F m represent the values of the influencing factors, dimensionless; C1 to C m represent the slope values of the regression analysis, dimensionless; B is the intercept value of the regression analysis, dimensionless; GR 高频 represents GR 原始 the high - frequency characteristic curve of the GR curve decomposition and recombination, API;

[0040] Q 预测 represents the single - factor formula R 2 > 0.8 of the factors, plus the GR 高频 curve - fitted single - point production capacity prediction formula, m³ / day.

[0041] Furthermore, the different influencing factors include shale content, porosity, permeability, gas saturation, and resistivity.

[0042] Furthermore, based on the logging data of the horizontal well to be predicted, using the single - point production capacity prediction formula to predict the production capacity of the horizontal well to be predicted, specifically including:

[0043] Based on the logging data of the horizontal well to be predicted, perform GR curve decomposition to obtain the high - frequency and low - frequency curve data of the horizontal well to be predicted;

[0044] Obtain the acoustic logging curve data of the horizontal well to be predicted from the logging data of the horizontal well to be predicted, combine it with the low - frequency curve data of the horizontal well to be predicted, establish a gas - producing reservoir division standard, and determine the High quality reservoir section of the horizontal well to be predicted;

[0045] Use the single - point production capacity prediction formula to calculate the production capacity of each depth point within the reservoir section of the horizontal well to be predicted, and perform cumulative summation to obtain the High quality overall production capacity of the reservoir section of the horizontal well to be predicted.

[0046] According to the specific embodiments provided by the present invention, the following technical effects are disclosed: The horizontal well production prediction method for sandstone gas reservoirs based on the recombination of logging curve features provided by the present invention first decomposes the GR curve data in the target interval into high-frequency and low-frequency curve data, combines the low-frequency curve with the acoustic logging data to establish a standard for dividing the gas-producing reservoir template to divide the gas-producing reservoir interval. After calculating the single-point productivity in the reservoir interval using the natural gas test data, the relationship between different factors and production is established, and then a productivity prediction formula is established based on the fitting relationship and the high-frequency curve to predict the gas production. It can be seen that the present invention predicts the productivity of the horizontal section of the horizontal well in the sandstone gas reservoir based on conventional logging data and a small amount of natural gas test data. This method does not need to rely on a large amount of experimental analysis data, the required data is easy to obtain, and the calculation cost is low. It can accurately predict the natural gas production of the horizontal section and provide strong technical support for natural gas exploration and development. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0048] Figure 1 It is a flowchart of the horizontal well production prediction method for sandstone gas reservoirs based on the recombination of logging curve features of the present invention;

[0049] Figure 2 It is a schematic diagram of the GR curve decomposition principle of the present invention;

[0050] Figure 3 It is a schematic diagram of the curve component conditions of the present invention;

[0051] Figure 4 It is a schematic diagram of the comparison of the average relative error of the GR curve of the present invention;

[0052] Figure 5 It is a schematic diagram of the decomposition and recombination of the GR curve features of the present invention;

[0053] Figure 6 It is a schematic diagram of the division of the relevant relationship of reservoir classification of the present invention;

[0054] Figure 7 It is a schematic diagram of the influence model of different influencing factors on productivity of the present invention (where a is the relationship diagram between porosity and single-point productivity; b is the relationship diagram between gas saturation and single-point productivity; c is the relationship diagram between porosity multiplied by gas saturation and single-point productivity; d is the relationship diagram between shale content and single-point productivity; e is the relationship diagram between resistivity and single-point productivity; f is the relationship diagram between density and single-point productivity);

[0055] Figure 8Schematic diagram of the production capacity prediction effect of the present invention;

[0056] Figure 9 Characteristic decomposition diagram of the GR curve of Well A1 in the embodiment of the present invention;

[0057] Figure 10 Comparison diagram of the average relative error of the combined curve of Well A1 in the embodiment of the present invention;

[0058] Figure 11 Recombinant diagram of the GR curve in the embodiment of the present invention;

[0059] Figure 12 Comparison of the GR lithologic profile diagram in the embodiment of the present invention;

[0060] Figure 13 Relevant relationship diagram of reservoir classification in the embodiment of the present invention;

[0061] Figure 14 Analysis model of different influencing factors and production in the embodiment of the present invention (wherein, a is the relationship diagram between porosity and single-point production capacity; b is the relationship diagram between gas saturation and single-point production capacity; c is the relationship diagram between porosity multiplied by gas saturation and single-point production capacity; d is the relationship diagram between shale content and single-point production capacity; e is the relationship diagram between resistivity and single-point production capacity; f is the relationship diagram between density and single-point production capacity);

[0062] Figure 15 Comparison diagram of the calculation results of the production capacity of the well to be predicted in the embodiment of the present invention;

[0063] Figure 16 Comparison diagram of the natural gas production capacity prediction result and the test result of Well B2 in the embodiment of the present invention. Detailed implementation manners

[0064] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0065] The purpose of the present invention is to provide a method for predicting the production capacity of the horizontal section of a horizontal well in a sandstone gas reservoir based on conventional logging data and a small amount of natural gas test data. For the horizontal section of a horizontal well in a sandstone gas reservoir, only a small amount of natural gas production data of the tested horizontal wells is required for calibration, and the natural gas production of the horizontal section of the newly drilled horizontal well can be predicted based on the conventional logging curves.

[0066] To make the above objects, features, and advantages of the present invention more obvious and understandable, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0067] As Figure 1 shown, the horizontal well production prediction method for sandstone gas reservoirs based on the recombination of logging curve characteristics provided by the present invention includes the following steps:

[0068] (1) Prepare data

[0069] Prepare the logging data of the horizontal well to be predicted (including logging curves such as natural gamma, deep and shallow resistivity, acoustic wave, etc. and data such as porosity, density, water saturation, and shale content), the logging data of the test horizontal well, and the natural gas test data (including data such as the test depth range and test flow rate). Among them, the test horizontal well refers to a standard well with known production for calibration.

[0070] (2) Curve feature decomposition

[0071] In this step, the GR curve is decomposed and recombined into high-frequency and low-frequency curves, and then the recombined curves are used for geological interpretation, reservoir classification, and calculation of production capacity.

[0072] As Figure 2 shown, the curve feature decomposition step specifically includes:

[0073] S1. Find all the maximum and minimum points of the GR curve within the depth range of the target layer, and fit the maximum envelope line GR max and the minimum envelope line GR min . Average the two envelope lines to obtain an average line, and subtract the average line from the original GR curve to obtain a new curve GR h1 without low frequency;

[0074]

[0075] In the formula: GR 原始 represents the original GR curve, API; GR max represents the natural gamma maximum envelope line, API; GR min represents the natural gamma minimum envelope line, API; GR h1 represents the first-order signal quantity, API;

[0076] S2. As Figure 3 shown, judge whether GR h1 meets the conditions a and b:

[0077]

[0078] In the formula: n represents the number of upper extreme points of GR h1 , dimensionless; p represents the number of zero-crossing points of GR h1 , dimensionless; h max represents the upper envelope line of GR h1 , API; hmin Denote GR h1 Lower envelope, API;

[0079] If not satisfied, use GR h1 Instead of GR 原始 , repeat step S1 until GR h1 Is satisfied. If satisfied, at this time GR h1 Is GR 原始 Rename the first-order component of the curve to GR c1 , use GR 原始 Subtract GR c1 , to obtain a new curve GR without high-frequency components r1 , see formula (3):

[0080] GR r1 = GR 原始 - GR c1 (GR c1 = GR h1 ) (3)

[0081] In the formula: GR c1 Denotes GR 原始 The first-order component decomposed out, API; GR r1 Denotes the first-order remainder, API;

[0082] S3, use GR r1 Instead of GR 原始 Curve, repeat steps S1 - S2 to obtain the second-order component GR c2 , repeat continuously until the nth-order component GR cn , and when its corresponding nth-order remainder GR r Is monotonic or constant, the decomposition process stops, see formula (4); finally GR 原始 The curve is decomposed into the sum of curves with different frequencies, formula (5):

[0083]

[0084]

[0085] In the formula: GR r Denotes the nth-order remainder, representing the average trend or mean value of the curve, API; x1, x2 represent random depth values and x1 < x2, dimensionless; GR r (x1) represents the GR value at depth point x1 r Value, API;

[0086] S4, after the GR curve is decomposed, n curves with different frequencies GR ci And a remainder trend line GR r , from GRr and n GRs ci The curves are combined respectively to obtain curves from low frequency to high frequency, as shown in formula (6):

[0087]

[0088] In the formula: GR n represents the cumulative sum of the nth-order remainder and the nth-order component curves, that is, the GR 原始 curve, API; GR1 to GR n-1 represents the GR curve synthesized by the nth-order remainder and different component curves, API;

[0089] S5. Calculate the average relative error value for all synthesized curves and GR 原始 curves, as shown in formula (7):

[0090]

[0091] In the formula: GR MRE represents the average relative error value of the curve, dimensionless; m represents the number of the depth point in the curve, the number of the top point of the curve is 1, and the number of the bottom depth point is m, dimensionless; represents the value of the i-th depth point of the synthesized GR curve, API; GR 原始i represents GR 原始 the value of the i-th depth point of the curve, API;

[0092] When the average relative error value of the curve is less than 0.1 and greater than 0.05, the correlation between the curve and the formation characteristics is relatively high and the high-frequency characteristics are not obvious. It is more matched with the formation characteristics in terms of spectral characteristics and thus suitable as the GR 低频 curve, as Figure 4 shown;

[0093] S6. The GR 高频 curve can be obtained by combining the first j (i < j < n) curves. The composition of the GR

[0094] GR 高频 = GR c1 + GR c2 + ··· + GR cj (8)

[0095] In the formula: GR 高频 represents the high-frequency characteristic curve obtained by decomposing and recombining the GR 原始 curve, API;

[0096] S7. The GR 低频 curve can be obtained by combining the last n - j curves. The composition of the GR

[0097] GR 低频 = GR cj+1+GR cj+2 + ··· +GR cn +GR rn (9)

[0098] Where: GR 低频 represents the low-frequency characteristic curve of the GR curve decomposition and recombination, API. 原始 The decomposition and recombination of the GR curve into its characteristic curves helps in better identifying and classifying reservoirs. The low-frequency components are used to identify reservoirs, while small-scale variations and heterogeneities within the high-frequency components are excluded. In addition to using the natural gamma ray curve for characteristic decomposition and recombination, this method can also be applied to curves such as the gamma ray without uranium and compensated neutron for rock and mineral analysis.

[0099] Figure 5 (3) Reservoir classification

[0100] Combining the AC curve and the GR

[0101] curve to establish a formula to statistically correlate the correlation coefficient with porosity to establish a comparison template (such as 低频 ) can more accurately establish the reservoir classification criteria for this target area (see Table 1): Figure 6

[0102]

[0103]

[0104] Where: I GR represents the relative value of natural gamma ray, dimensionless; GR mi , GR ma represent the natural gamma ray values corresponding to clean sandstone and clean shale in this area, API; GR 低频 represents the low-frequency characteristic curve of the GR curve decomposition and recombination, API; α represents the correlation coefficient, dimensionless; c represents the empirical coefficient, dimensionless, generally 3.7; Δt, Δt 原始 min max min and Δt max represent the acoustic travel time curve response value, response minimum value, and response maximum value respectively, μs / m;

[0105] Table 1 Statistical schematic table of reservoir classification correlation

[0106] Horizontal section Non-reservoir Poor reservoir High-quality reservoir Correlation <![CDATA[d6<α]]> <![CDATA[d4 < α < d6]]> <![CDATA[α < d4]]>

[0107] As shown in Table 1, if d6 < α, it is classified as a non-reservoir; if d4 < α < d6, it is classified as a poor reservoir; if α < d4, it is classified as a high-quality reservoir; where d6 represents the first set threshold, d4 represents the second set threshold, and d6 > d4. The first set threshold and the second set threshold can be set according to the actual situation.

[0108] (4) Establish the fitting of the analysis formula for the relationship between different factors and production

[0109] Using the correlation coefficient obtained in step (3), repeat step (2) to obtain high-quality reservoirs and non-high-quality reservoirs within the horizontal sections of multiple horizontal wells. For the high-quality reservoir sections, perform single-point production capacity Q 125 calculation, formula (12). Statistically analyze the influence relationship of different curve factors on production and establish an influence relationship model, as Figure 7 shown. Evaluate the model, check the goodness of fit and prediction ability of the model, formula (13). After determining the best model, select the influencing factors for regression analysis fitting and add the GR 高频 curve as the details and variations affecting production capacity in the formation, and fit the production capacity, see formula (14):

[0110]

[0111]

[0112] Q 预测 = C1(F1) + C2(F2) + … + C m (F m ) + B + GR 高频 (14)

[0113] In the formula: Q 125 represents the single-point production capacity, m³ / day; Q t represents the normal test flow rate in the natural gas test data, 10,000 m³ / day; H q represents the thickness of the high-quality reservoir, m; Q1, Q2 to Q n represent the influence relationship between different factors and production capacity, m³ / day; a1 to a n represent the slope values of the trend line, dimensionless; b1 to b n represent the intercept values of the trend line, dimensionless; F1 to F m represent the values of the influencing factors, dimensionless; C1 to C m represent the slope values of the regression analysis, dimensionless; B is the intercept value of the regression analysis, dimensionless; GR 高频 represents the GR 原始 curve's high-frequency characteristic curve after decomposition and recombination, API;

[0114] Q 预测 represents the single-factor formula R 2 > 0.8 factor, plus the GR 高频 curve's single-point production capacity prediction formula after fitting, m³ / day.

[0115] (5) Calculate the production capacity

[0116] The productivity formula obtained in step (4) can be used to predict the reservoir production in the horizontal section by well logging curves. First, after calculating the productivity of all single points in the horizontal section, the productivity of all single points in all reservoir sections is statistically analyzed to achieve accurate prediction of the productivity of the entire reservoir section.

[0117]

[0118] In the formula: Q 总 —— Calculated normal flow of the reservoir section, 10,000 m³ / day; i—— Depth point number of the tested reservoir section, the depth number of the reservoir top is 1, and the depth number of the reservoir bottom is n; Q i —— Calculated single-point productivity at depth point i, Q 预测 , m³ / day. For example Figure 8 It is a schematic diagram of the prediction effect of the reservoir productivity in the horizontal section.

[0119] In view of the characteristics of the sandstone horizontal section reservoir development, after classifying the reservoir types, relationships are established through the different influences of different factors on productivity to conduct production prediction. First, the GR curve is decomposed into characteristic curves of different frequencies, and according to the standard of the average relative error of the curves, it is recombined into the GR 低频 curve and the GR 高频 curve. After establishing a correlation between the GR 低频 curve and the AC curve to obtain a reservoir classification template, the reservoir is classified, and then the single-point productivity in the reservoir section is obtained using natural gas test data. By analyzing and fitting the relationships between different factors and production with multiple wells and adding the GR 高频 curve section, a production formula is obtained. Using this method, 15 horizontal wells that have undergone natural gas testing in a gas field in the Sichuan Basin are selected. Among them, 10 wells are used to establish a model to fit the production formula, and 5 wells are used for production prediction. The average relative error between the prediction results and the actual test results is less than 0.2, and the correlation coefficient is greater than 0.8, indicating that this method has high prediction accuracy and good invention effect.

[0120] Taking 15 horizontal wells in a gas reservoir in the Sichuan Basin as an example, the specific implementation method and implementation effect of the present invention are described.

[0121] ① Prepare data

[0122] Prepare data of 10 natural gas test wells (Well A1 to A 10 Well) and 5 wells to be predicted (Well B1 to B5), including data such as natural gamma, resistivity, acoustic wave curve, porosity, shale content, etc. And the natural gas test data corresponding to these wells in the horizontal section, including data such as the test depth range and test flow rate.

[0123] ② Curve characteristic decomposition

[0124] Taking Well A1 as an example based on logging data, the horizontal section length of Well A1 is determined by combining natural gas test data, and its GR 原始 curve is decomposed into 7 component curves and 1 residual curve, as Figure 9 shown;

[0125] Combining the residual line GR r and the 7 component lines to obtain the GR 原始 curve from low frequency to, see formula (16):

[0126]

[0127] Calculate the average relative error of the 7 curves, as Figure 10 shown. The average relative error of the GR4 curve GR MRE meets the requirements between 0.1 and 0.05. The statistical results are shown in Table 2:

[0128] Table 2 Statistical table of average relative error of curves

[0129]

[0130]

[0131] Combining the decomposed component curves GR c1 , GR c2 and GR c3 to generate the GR 高频 characteristic curve:

[0132] GR 高频 = GR c1 + GR c2 + GR c3 (17)

[0133] Combining the decomposed component curves GR c4 , GR c5 , GR c6 , GR c7 and the residual curve GR r to generate the GR 低频 characteristic curve, see Figure 11 :

[0134] GR 低频 = GR c4 + CR c5 + GR c6 + GR c7 + GR r (18)

[0135] After obtaining the GR 低频 curve composed of low-frequency characteristics, for GR 原始Curve and GR 低频 The curve is used to plot the lithologic profile respectively (such as Figure 12 ). After the GR curve is decomposed and recombined by the curve characteristics, the details and variation characteristics in the formation are eliminated, making the formation composition relatively uniform, so as to better identify and divide the reservoir. 低频

[0136] ③ Reservoir classification

[0137] Figure 13 After dividing the reservoir section according to the lithologic profile, the sandstone reservoir is classified. First, the relative value I of natural gamma is calculated by formula (10) GR , and then the correlation coefficient α between the GR curve and the AC curve in the horizontal section is calculated by formula (11) 低频 . Establish a cross plot of the correlation relationship (α) - porosity (φ) ( ). It can be seen from Table 3 of the statistical results that the α value of the non-reservoir section is greater than 0.372, α between 0.342 and 0.453 is a poor reservoir, and α less than 0.341 is a high-quality reservoir:

[0138] Table 3 Statistical results of reservoir classification correlation

[0139] Horizontal section Non-reservoir Poor reservoir High-quality reservoir Correlation 0.372＜α 0.341＜α＜0.453 α＜0.341

[0140] ④ Establish the fitting of the analysis formula for the relationship between different factors and production

[0141] Repeat step ② to decompose and recombine the curves of 15 wells, use the correlation relationship obtained in step ③ to classify and divide the reservoirs of all wells, and count the data required for factor analysis of the test wells (Well A1 - Well A 10 ), such as reservoir thickness, shale content, porosity, density, resistivity, and natural gas test flow rate, etc., and calculate Q 125 normal flow rate. The detailed data and calculation results are shown in Table 4:

[0142] Table 4 Statistical table of factor analysis

[0143]

[0144]

[0145] Figure 14 Combine shale content, porosity, permeability, gas saturation, and resistivity, etc., and bring them in for the analysis and modeling of the relationship with single-point productivity ( ), and obtain each model formula (19):

[0146]

[0147] ​Where: POR - porosity, %; SG - gas saturation, %; SH - shale content, %; RT - resistivity, Ω·m; PERM - permeability, md.

[0148] Obtain the fitting formula R of the product of shale content, porosity and gas content and production 2 When greater than 0.8, the obtained correlation is high and the effect is good. Select these two factors and production for regression analysis and fitting, and add the GR 高频 curve as the influence characteristics of detailed changes in the formation, and obtain the single-point productivity prediction formula (20):

[0149]

[0150] ⑤ Calculate the productivity effect

[0151] Use the productivity formula Q 预测 After calculating the single-point productivity of the reservoirs of 5 wells to be predicted (Wells B1 - B5), use formula (15) to statistically analyze the single-point productivity of the gas-producing reservoir sections. It is found that the average relative error between the calculation results of the 5 wells to be predicted and the actual test results is 0.13 (Table 5), and the correlation coefficient R 2 is 0.875 ( Figure 15 ) indicating that the prediction result has high accuracy and good correlation effect:

[0152]

[0153]

[0154]

[0155] Where: Q Bj —— Calculated normal flow rate Q of the reservoir section of the well to be predicted 总 , 10,000 m³ / day; j - serial number of the well to be predicted, dimensionless; Q t —— Test normal flow rate in the natural gas test results, 10,000 m³ / day; Q RE —— Relative error of a single well of the well to be predicted, dimensionless; Q MRE —— Average relative error of the well to be predicted, dimensionless; t - number of wells to be predicted, dimensionless.

[0156] Table 5 Comparison table of productivity calculation of wells to be predicted

[0157]

[0158]

[0159] The calculation results are as Figure 16 shown:

[0160] Figure 16Taking the well to be predicted, Well B2, as an example, the test section of Well B2 is from 2601 to 3495 meters. The horizontal section is mainly composed of pure sandstone. The reservoir is divided by decomposing and recombining the GR curve and combining it with the AC curve through curve feature decomposition. Then, after calculating the single-point productivity through the productivity calculation formula, the single-point productivities within the reservoir section are accumulated to obtain the production of all reservoirs within the horizontal section. The coincidence degree between the predicted natural gas production of the horizontal well by the present invention and the actual measured gas production is relatively high, which fully shows that the method of the present invention has good application effects in predicting the gas production of horizontal wells.

[0161] The method for predicting the production of horizontal wells in sandstone gas reservoirs based on the recombination of logging curve features provided by the present invention establishes a productivity prediction method that combines curve feature decomposition and multi-influence factor analysis. Through this method, the accuracy of evaluating the natural gas production in the horizontal section can be improved, so as to ensure the smooth progress of the exploration and development of horizontal wells in sandstone gas reservoirs.

[0162] Specific examples are used in this article to elaborate on the principle and implementation mode of the present invention. The description of the above embodiments is only used to help understand the method and its core idea of the present invention; at the same time, for those of ordinary skill in the art, according to the idea of the present invention, there will be changes in the specific implementation mode and application scope. In summary, the content of this specification should not be construed as a limitation to the present invention.

Claims

1. A method for predicting the production of horizontal wells in sandstone gas reservoirs based on the reconstruction of well logging curve characteristics, characterized in that: The following steps are involved: Obtaining well logging data of the horizontal well to be predicted, well logging data of the test horizontal well, and natural gas test data; Based on the logging data of the test horizontal well, GR curve decomposition is performed to obtain high-frequency and low-frequency curve data of the test horizontal well; the GR curve decomposition is specifically to perform characteristic decomposition on the GR curve to obtain different frequency component curves, and recombinantly obtain high-frequency curves and low-frequency curves according to the average relative error standard of the curves; Acoustic logging curve data of the test horizontal well is obtained from the logging data of the test horizontal well, and combined with the low-frequency curve data of the test horizontal well, a gas-producing reservoir classification standard is established to determine the high-quality reservoir of the test horizontal well; the gas-producing reservoir classification standard is expressed as: Where: I GR Represents the relative value of natural gamma, dimensionless; GR mi , GR ma Indicates the natural gamma value corresponding to pure sandstone and pure mudstone in this area, API; GR 低频 Represents GR 原始 The low-frequency characteristic curve of curve decomposition and reconstruction, API; α represents the correlation coefficient, dimensionless; c represents the empirical coefficient, dimensionless; Δt, Δt min and Δt max Respectively represent the response value, minimum response value and maximum response value of the acoustic time difference curve, μs / m; If d6<α, it is classified as a non-reservoir; if d4<α<d6, it is classified as a poor reservoir; if α<d4, it is classified as a high-quality reservoir; where d6 represents the first set threshold, d4 ​​represents the second set threshold, and d6>d4; The single-point capacity of the high-quality reservoir of the test horizontal well is calculated using the natural gas test data. Combined with multiple test horizontal wells, the relationship between different influencing factors and single-point capacity is regressed and fitted. Then, a single-point capacity prediction formula is established based on the fitting relationship and the high-frequency curve data of the test horizontal well. Specifically, it includes: Q 预测 =C1(F1)+C2(F2)+…+C m (F m )+B+GR 高频 (14) Where: Q 125 Indicates the single-point production capacity, cubic meters per day; Q t Indicates the normal flow rate of natural gas test data, 10,000 cubic meters / day; H q Indicates the thickness of high-quality reservoirs, m; Q1, Q2 to Q n Indicates the relationship between different factors and production capacity, cubic meters / day; a1 to a n Indicates the slope value of the trend line, dimensionless; b1 to b n Represents the trend line intercept value, dimensionless; F1 to F m Indicates the value of the influencing factor, dimensionless; C1 to C m represents the slope value of regression analysis, dimensionless; B represents the intercept value of regression analysis, dimensionless; GR 高频 Represents GR 原始 High-frequency characteristic curve of curve decomposition and reconstruction, API; Q 预测 Indicates the selection of single factor formula R 2 >0.8 factor, plus GR 高频 Single point capacity prediction formula after curve fitting, cubic meters / day; Based on the logging data of the horizontal well to be predicted, the single-point production capacity prediction formula is used to predict the production capacity of the horizontal well to be predicted.

2. The method for predicting production of horizontal wells in sandstone gas reservoirs based on well logging curve feature reconstruction according to claim 1, characterized in that: The logging data of the horizontal well to be predicted and the logging data of the horizontal well to be tested include natural gamma logging curves, deep and shallow resistivity logging curves, sonic logging curves, and data on porosity, density, water saturation, gas saturation, permeability, and mud content; the natural gas test data include test depth range and test flow rate.

3. The method for predicting production of horizontal wells in sandstone gas reservoirs based on well logging curve feature reconstruction according to claim 1, characterized in that: The GR curve decomposition specifically includes the following steps: S1, determine all the maximum and minimum points of the GR curve, and fit the maximum envelope GR max and the minimum envelope GR min , average the two envelopes to get the average line, subtract the average line from the original GR curve to get a new curve GR without low frequency h1 ; S2, determine GR h1 Whether conditions a and b are met: Where: n represents GR h1 The number of upper extreme points, dimensionless; p represents GR h1 The number of zero crossings, dimensionless; h max Represents GR h1 Upper envelope, API; h min Represents GR h1 lower envelope, API; If not satisfied, use GR h1 Replace GR 原始 , repeat step S1 until GR h1 Until it is satisfied, if it is satisfied, then GR h1 GR 原始 The first-order component of the curve is renamed GR c1 , using GR 原始 Subtract GR c1 , and obtain a new curve GR without high-frequency components r1 ; S3, using GR r1 Replace GR 原始 Repeat steps S1-S2 to obtain the second-order component GR c2 , repeat until the nth order component GR cn , and satisfies its corresponding n-order margin GR r The decomposition process stops when it is monotonic or constant; finally GR 原始 The curve is decomposed into the sum of curves of different frequencies; S4, GR curve is decomposed to obtain n curves of different frequencies GR ci and a residual trend line GR r , by GR r and n GR ci The curves are combined to obtain synthetic curves from low frequency to high frequency; S5, for all synthetic curves and GR 原始 The average relative error value of the curve is calculated; When the average relative error value of the curve is less than 0.1 and greater than 0.05, the correlation between the curve and the formation characteristics is high and the high-frequency characteristics are not obvious. The spectral characteristics are more consistent with the formation characteristics and are therefore suitable as GR. 低频 curve; S6. The first j (i < j < n) curves can be combined to obtain GR 高频 The composition of the curves is as follows: GR 高频 =GR c1 +GR c2 +···+GR cj (8) Where: GR 高频 Represents GR 原始 High-frequency characteristic curve of curve decomposition and reconstruction, API; S7, the next nj curves can be composed to get GR 低频 The components of the curve are: GR 低频 =GR cj+1 +GR cj+2 +···+GR cn +GR rn (9) Where: GR 低频 Represents GR 原始 Low-frequency characteristic curve after curve decomposition and reconstruction, API.

4. The method for predicting production of horizontal wells in sandstone gas reservoirs based on well logging curve feature reconstruction according to claim 1, characterized in that: The different influencing factors include mud content, porosity, permeability, gas saturation and resistivity.

5. The method for predicting production of horizontal wells in sandstone gas reservoirs based on well logging curve feature reconstruction according to claim 1, characterized in that: The production capacity of the horizontal well to be predicted is predicted based on the well logging data of the horizontal well to be predicted by using a single-point production capacity prediction formula, specifically including: Based on the logging data of the horizontal well to be predicted, the GR curve is decomposed to obtain the high-frequency and low-frequency curve data of the horizontal well to be predicted; Acquire the acoustic logging curve data of the horizontal well to be predicted from the logging data of the horizontal well to be predicted, combine it with the low-frequency curve data of the horizontal well to be predicted, establish the gas production reservoir division standard, and determine the high-quality reservoir section of the horizontal well to be predicted; The single-point capacity prediction formula is used to calculate the capacity of each depth point in the high-quality reservoir section of the horizontal well to be predicted, and the cumulative sum is performed to obtain the overall capacity of the reservoir section of the horizontal well to be predicted.

Images