A strong water-sensitive oil reservoir dessert intelligent identification method

Abstract

The present application belongs to the technical field of oil and gas field development engineering, and discloses a kind of strong water-sensitive reservoir dessert intelligent identification method.The method comprises: collating, correcting well logging data curve;Determine the single well best clay content calculation method, use single well best clay content calculation method in good hole section, use resistivity-neutron formula method to calculate clay content in poor hole section, determine the single well whole section clay content;Calculate the single well whole section mechanical specific energy E MSE ;Oil layer is established to predict the model of water-sensitive mineral;Determine different hydration degree interval through hydration test;Build multiple prediction models, and verify and screen the best prediction model in prediction accuracy by measuring sample water sensitivity;According to the quantitative prediction result of water-sensitive mineral, determine the strong water-sensitive reservoir dessert.The present application combines water-sensitive intelligent prediction model determined by physical model experiment, considers the wellbore stability factor when predicting clay, and can accurately identify the strength and weakness of water sensitivity and strong water-sensitive reservoir dessert.

Description

Technical Field

[0001] This invention belongs to the field of oil and gas field development engineering technology, and specifically relates to a method for intelligent identification of sweet spots in highly water-sensitive reservoirs. Background Technology

[0002] Highly water-sensitive reservoirs are constrained by their strong water-sensitivity characteristics. Early-stage horizontal well production in these areas exhibits rapid declines in both production and fracturing. Existing reservoir classification standards, based on physical properties as sweet spots, are inadequate for evaluating water-sensitive reservoirs. Furthermore, existing logging data is insufficient to characterize the sweet spots in water-sensitive reservoirs, while specialized water-sensitive detection instruments are expensive. Additionally, some well areas are near-source deposits with severe wellbore collapse, making clay calculations extremely difficult, and methods for calculating water-sensitive minerals in clay that influence water sensitivity are lacking. Therefore, identifying the dominant sweet spots (relatively high oil saturation and relatively weak water sensitivity) in these highly water-sensitive reservoirs is a crucial task and a pressing problem for developers.

[0003] Currently, mainstream methods for predicting water-sensitive mineral content are limited to clay mineral prediction. This primarily involves using GR (well pressure) and sonic curves to infer clay mineral content from the predicted mud content. However, these methods are highly susceptible to wellbore stability and fail to consider that clay content is not necessarily equal to water-sensitive mineral content. Therefore, these methods are prone to instability, and no single method can accurately predict the water-sensitive mineral content within clay minerals. Summary of the Invention

[0004] This invention aims to address the technical problems existing in the prior art by providing an intelligent identification method for sweet spots in highly water-sensitive oil reservoirs. It not only considers wellbore stability factors in clay prediction but also quantitatively characterizes the content of water-sensitive minerals through numerical modeling. Combined with the intelligent prediction model for water sensitivity obtained from physical model experiments, it can accurately identify the strength of water sensitivity, thereby identifying sweet spots in highly water-sensitive oil reservoirs with relatively high oil saturation and relatively weak water sensitivity.

[0005] To achieve the above technical objectives, the present invention adopts the following technical solution:

[0006] A method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs, the method comprising the following steps:

[0007] Step S101: Organize the logging curve data and perform environmental correction on the logging curve data;

[0008] Step S102: Calculate the clay content using multiple methods, and fit the calculated clay content with the measured clay content to determine the optimal clay content calculation method for a single well; then, divide the wellbore into good and poor wellbore sections according to the wellbore environment, and calculate the clay content in the good wellbore section using the optimal clay content calculation method for a single well, while calculating the clay content in the poor wellbore section using the resistivity-neutron formula method, thereby determining the clay content of the entire well section for a single well;

[0009] Step S103: Calculate the mechanical specific energy E of the entire well section based on drilling data. MSE :

[0010]

[0011] In the above formula: E MSE - Mechanical specific energy, MPa; F - Drilling pressure, N; CALI - Well diameter, mm; N - Rotary rotation speed, r / min; ZS - Drilling time, min;

[0012] Step S104: Determine the development status of the oil layer in the block, and establish a "diagenesis-rock breaking" water-sensitive mineral prediction model for each oil layer based on mechanical specific energy and the measured absolute content of water-sensitive minerals.

[0013] Step S105: Take rock samples from the block and conduct hydration tests to determine different hydration degree ranges based on the absolute content of different water-sensitive minerals;

[0014] Step S106: Using the well depth and logging curve data obtained in step S101, the clay content of the entire well section calculated in step S102, the mechanical specific energy in step S103, and the predicted water-sensitive minerals of the entire well section constructed by the oil layer in step S104 as input values, and taking 50%-80% of the total amount of water-sensitive samples obtained from the experiment as the result target, a prediction model is constructed.

[0015] Step S107: Using the remaining sample amount of the total measured water sensitivity of the sample obtained in the experiment as the test data, verify the accuracy of each prediction model in step S106, and select the prediction model with the best prediction accuracy as the water-sensitive intelligent prediction model.

[0016] Step S108: Quantitatively predict water-sensitive minerals using the water-sensitive intelligent prediction model in step S107. Based on the water sensitivity results output by the water-sensitive intelligent prediction model, identify the sweet spot area of ​​the strongly water-sensitive oil reservoir.

[0017] Furthermore, in step S101, environmental correction is performed on the well logging curve data, specifically as follows:

[0018] A correlation equation is established by fitting stable and unstable logging curves. When the unstable logging curve is affected by the environment, a segment of the curve is fitted by the correlation equation to replace the logging curve segment affected by the environment, thereby achieving environmental correction of the logging curve.

[0019] Further, in step S102, multiple methods are used to calculate the clay content, and the calculated clay content is fitted with the measured clay content to determine the optimal clay content calculation method for a single well, specifically including:

[0020] Based on the density, neutron, and resistivity curves, clay content a was calculated using the "neutron-density intersection method," clay content b was calculated using the "resistivity-neutron formula method," and clay content c was calculated using the "Schlumberger chart method." The clay contents a, b, and c were then fitted with the measured clay contents, and the method with the best fit was selected as the optimal method for calculating the clay content of a single well.

[0021] Furthermore, in step S102, the wellbore is divided into good wellbore sections and poor wellbore sections according to the wellbore environment. Specifically, wellbore sections with collapsed mudstone walls are poor wellbore sections, while those with good mudstone walls are good wellbore sections.

[0022] Furthermore, in step S104, the water-sensitive mineral prediction model is constructed using SPSS software.

[0023] Furthermore, in step S105, determining different hydration degree ranges based on the absolute content of different water-sensitive minerals specifically includes:

[0024] If the absolute content of water-sensitive minerals is <4%, it falls within the weak hydration range;

[0025] If the absolute content of water-sensitive minerals is 4%-6%, it falls within the medium hydration range;

[0026] If the absolute content of water-sensitive minerals is >6%, it belongs to the strong hydration range.

[0027] Furthermore, in step S106, the prediction model includes an SVM random number prediction model, a decision tree C5.0 random number prediction model, and a neural network random number prediction model.

[0028] Furthermore, the logging curve data used in step S106 includes at least density, neutron, and resistivity data.

[0029] Furthermore, in step S106, the SVM random number prediction model, the decision tree C5.0 random number prediction model, and the neural network random number prediction model are all constructed using SPSS software.

[0030] Furthermore, in step S108, based on the water sensitivity results "strong / medium / weak" output by the water-sensitive intelligent prediction model, the "weak" value is determined to be a sweet spot region of a strongly water-sensitive reservoir.

[0031] Compared with the prior art, the beneficial effects of the present invention are:

[0032] This invention addresses the technological gap in predicting the content of water-sensitive minerals in clay minerals by proposing, for the first time, an intelligent model for predicting water-sensitive minerals. Using existing well logging curves, depth data, and drilling data with good correlation, linear and nonlinear prediction models were constructed respectively. The model with the best prediction accuracy was selected and verified using measured sample data as the intelligent water-sensitive prediction model. This model then automatically determines the water sensitivity of the reservoir based on the input data and identifies sweet spots in strongly water-sensitive reservoirs based on the prediction results. This invention not only considers wellbore stability factors in clay prediction but also quantitatively characterizes the content of water-sensitive minerals through numerical models. Combined with physical model experiments, the intelligent water-sensitive prediction model can accurately identify the strength of water sensitivity, thereby identifying sweet spots in strongly water-sensitive reservoirs with relatively high oil saturation and relatively weak water sensitivity. Experimental analysis of reservoirs and comparison of actual rock sample detection data with the prediction data of this invention show that the prediction accuracy of the intelligent water-sensitive prediction model is 82%-90%, providing important theoretical guidance for the exploitation of strongly water-sensitive reservoirs. Attached Figure Description

[0033] Figure 1 This is a flowchart of the intelligent identification method for sweet spots in highly water-sensitive reservoirs according to an embodiment of the present invention;

[0034] Figure 2 This is a schematic diagram showing the results of determining clay content using the conventional density curve and neutron curve difference method.

[0035] Figure 3 This is a schematic diagram showing the results of determining clay content using the method of an embodiment of the present invention;

[0036] Figure 4 This is a schematic diagram showing the correlation between the mechanical specific energy of the P3w1 water-sensitive mineral prediction model and the measured water-sensitive minerals in an embodiment of the present invention.

[0037] Figure 5 This is a schematic diagram showing the correlation between the mechanical specific energy of the P3w2 water-sensitive mineral prediction model and the measured water-sensitive minerals in an embodiment of the present invention.

[0038] Figure 6The diagrams above illustrate the water sensitivity prediction results of various prediction models in this invention, where a is a diagram of the water sensitivity prediction results of the SVM random number prediction model; b is a diagram of the water sensitivity prediction results of the decision tree C5.0 random number prediction model; and c is a diagram of the water sensitivity prediction results of the neural network random number prediction model. Detailed Implementation

[0039] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0040] Example 1

[0041] This invention provides a method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs, such as... Figure 1 As shown, the method includes the following steps:

[0042] Step S101: Organize the logging curve data, perform environmental correction on the logging curve data, and restore the true logging data;

[0043] Among these, environmental correction is performed on the well logging curve data, specifically as follows:

[0044] A correlation equation is established by fitting stable and unstable logging curves. When the unstable logging curve is greatly affected by the environment, a segment of the curve is fitted by the correlation equation to replace the logging curve segment affected by the environment, thereby achieving environmental correction of the logging curve.

[0045] Step S102: Based on the density, neutron, and resistivity curves, calculate the clay content a using the "neutron-density intersection method," the clay content b using the "resistivity-neutron formula method," and the clay content c using the "Schlumberger chart method," respectively. Fit the clay contents a, b, and c to the measured clay contents to determine the optimal clay content calculation method for a single well. Then, based on the wellbore environment, classify the mudstone wellbore collapse section into poor wellbore sections, and vice versa. Calculate the clay content in the good wellbore sections using the optimal clay content calculation method for a single well, and in the poor wellbore sections using the resistivity-neutron formula method, thereby determining the total clay content of the entire well section.

[0046] Among them, the "neutron-density intersection method" uses a conventional data intersection chart method to input neutron and density curves into the chart and determine the clay content by judging the location of the input point on the chart.

[0047] The clay content is calculated using the "resistivity-neutron formula method" according to the following formula:

[0048]

[0049] In equation (1) above, VCL is the volume percentage of soil content;

[0050] The "Schlumberger plot method" uses Schlumberger's data intersection plot method to input neutron and density curves into a plot and determine the clay content by identifying the location of the input point on the plot.

[0051] like Figure 2 The diagram shows the results of determining clay content using the conventional density curve and neutron curve difference method. Figure 3 The diagram shows the results of determining clay content using the method of this embodiment of the invention. This embodiment of the invention, which determines the clay content of the entire well section, not only considers the influence of the wellbore environment but also employs an optimal approach to multiple clay prediction methods. Compared to the conventional method of determining clay content through the difference between density curves and neutron curves, the clay prediction accuracy of this embodiment is improved from 76% to 80% compared to the conventional method, and the average absolute error is reduced from 10% to 1.6%.

[0052] Step S103: Calculate the mechanical specific energy (EMSE) of the entire well section based on drilling data.

[0053]

[0054] In equation (2) above: E MSE - Mechanical specific energy, MPa; F - Drilling pressure, N; CALI - Well diameter, mm; N - Rotary rotation speed, r / min; ZS - Drilling time, min.

[0055] Step S104: Determine the development status of the oil layer in the block, and establish a "diagenesis-rock breaking" water-sensitive mineral prediction model for each oil layer based on the measured absolute content of water-sensitive minerals.

[0056] Since the target reservoir in this embodiment of the invention contains two oil layers available for development, it is necessary to construct water-sensitive mineral prediction models for the two oil layers in layers, namely: P3w1 water-sensitive mineral prediction model and P3w2 water-sensitive mineral prediction model.

[0057] This step utilizes two key characteristics: the distribution of water-sensitive minerals is influenced by sedimentation and diagenesis, and the content of water-sensitive minerals affects the rock-breaking efficiency of the drill bit. It combines depth, sequence stratigraphy, mechanical energy (the work done by the drill bit to break a unit volume of rock under drilling pressure and torque during drilling, which reflects the rock-breaking efficiency of the drill bit; a high mechanical energy indicates low rock-breaking efficiency) with measured water-sensitive mineral content and geological characteristics. Based on two main oil layers, a layered "diagenesis-rock-breaking" water-sensitive prediction model is constructed using SPSS software.

[0058] The formula for the P3w1 water-sensitive mineral prediction model is as follows:

[0059]

[0060] The formula for the P3w2 water-sensitive mineral prediction model is:

[0061]

[0062] In equations (3) and (5) above: SM - water sensitivity index; Depth - depth, m; F - drilling pressure, N; CALI - well diameter, mm; N - rotary table speed, r / min; ZS - drilling time, min; SH - clay content; GR - logging curve; DEN - density logging curve; CNL - neutron logging curve; SW - water saturation; SP - spontaneous potential logging curve;

[0063] Among them, the correlation results between the mechanical specific energy of the P3w1 water-sensitive mineral prediction model and the measured water-sensitive minerals are as follows: Figure 4 As shown, the correlation results between the mechanical specific energy of the P3w2 water-sensitive mineral prediction model and the measured water-sensitive minerals are as follows: Figure 5 As shown, the trend fit R of the two water-sensitive mineral prediction models can be observed. 2 The values ​​are 0.88 and 0.72 respectively. Therefore, the mechanical specific energy of the water-sensitive mineral prediction model in this embodiment of the invention has a good correlation with the measured water-sensitive minerals, which is beneficial to improving the prediction accuracy of the constructed model.

[0064] Step S105: Take rock samples from the block and conduct hydration tests, i.e., select core samples with different water-sensitive mineral contents and soak them in water, and determine different hydration degree ranges based on the absolute content of different water-sensitive minerals; in this embodiment of the invention, the following three hydration degree ranges are determined based on the absolute content of water-sensitive minerals being <4% or >6% as the boundary:

[0065] If the absolute content of water-sensitive minerals is <4%, it falls within the weak hydration range;

[0066] If the absolute content of water-sensitive minerals is 4%-6%, it falls within the medium hydration range;

[0067] If the absolute content of water-sensitive minerals is >6%, it belongs to the strong hydration range.

[0068] Step S106: Using the well depth and logging curve data obtained in step S101, the clay content of the entire well section calculated in step S102, the mechanical specific energy in step S103, and the predicted water-sensitive minerals of the entire well section constructed by the oil layer in step S104 as input values, and taking 70% of the total amount of water-sensitive samples obtained from the experiment as the result target, a prediction model is constructed.

[0069] The well logging curve data includes at least density, neutron, and resistivity data. The prediction models constructed in this embodiment include SVM random number prediction models, decision tree C5.0 random number prediction models, and neural network random number prediction models. It should be noted that the prediction models are not limited to those constructed using the above three methods; in other embodiments, prediction models using other methods can also be constructed or added.

[0070] Step S107: Using the remaining 30% of the total amount of water-sensitive samples obtained from the experiment as test data, verify the accuracy of each prediction model in step S106, and select the prediction model with the best prediction accuracy.

[0071] like Figure 6 The figures show the water sensitivity prediction results of various prediction models. Among them, a is a schematic diagram of the water sensitivity prediction results of the SVM random number prediction model, with a prediction accuracy of 77-80%; b is a schematic diagram of the water sensitivity prediction results of the decision tree C5.0 random number prediction model, with a prediction accuracy of 70-85%; and c is a schematic diagram of the water sensitivity prediction results of the network random number prediction model, with a prediction accuracy of 82-90%.

[0072] Comparison revealed that the neural network random number prediction model exhibited the highest prediction accuracy, meeting the required accuracy. Therefore, the neural network random number prediction model with the best prediction accuracy was ultimately selected as the water-sensitive intelligent prediction model for this embodiment of the invention. The comparison between the predicted data from the neural network random number prediction model and the actual rock sample detection data is shown in Table 1 below. The confidence level in the table serves as the basis for the data model's prediction values. The confidence level represents the noise content in the data; a higher confidence level indicates lower noise and higher accuracy of the prediction results. This confidence level data was automatically calculated using SPSS software numerical simulation.

[0073] Table 1. Results of water sensitivity prediction by random numbers from neural networks (70% of samples)

[0074]

[0075]

[0076] Step S108: Quantitatively predict water-sensitive minerals using the water-sensitive intelligent prediction model in step S107, and determine the "weak" value as a sweet spot region of a strong water-sensitive reservoir based on the predicted water sensitivity results of "strong / medium / weak".

[0077] The above description is merely an embodiment of this application and is not intended to limit the invention. Any modifications, equivalent substitutions, and improvements made within the scope of this application should be included within the protection scope of this invention.

Claims

1. A method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs, characterized in that, The method includes the following steps: Step S101: Organize the logging curve data and perform environmental correction on the logging curve data; Step S102: Calculate the clay content using multiple methods, and fit the calculated clay content with the measured clay content to determine the optimal clay content calculation method for a single well; then, divide the wellbore into good and poor wellbore sections according to the wellbore environment, and calculate the clay content in the good wellbore section using the optimal clay content calculation method for a single well, while calculating the clay content in the poor wellbore section using the resistivity-neutron formula method, thereby determining the clay content of the entire well section for a single well; Step S103: Calculate the mechanical specific energy E of the entire well section based on drilling data. MSE : In the above formula: E MSE - Mechanical specific energy, MPa; F - Drilling pressure, N; CALI - Well diameter, mm; N - Rotary rotation speed, r / min; ZS - Drilling time, min; Step S104: Determine the development status of the oil layer in the block, and establish a "diagenesis-rock breaking" water-sensitive mineral prediction model for each oil layer based on mechanical specific energy and measured absolute content of water-sensitive minerals. Step S105: Take rock samples from the block and conduct hydration tests to determine different hydration degree ranges based on the absolute content of different water-sensitive minerals; Step S106: Using the well depth and logging curve data obtained in step S101, the clay content of the entire well section calculated in step S102, the mechanical specific energy in step S103, and the predicted water-sensitive minerals of the entire well section constructed by the oil layer in step S104 as input values, and taking 50%-80% of the total amount of water-sensitive samples obtained from the experiment as the result target, a prediction model is constructed. Step S107: Using the remaining sample amount of the total measured water sensitivity of the sample obtained in the experiment as the test data, verify the accuracy of each prediction model in step S106, and select the prediction model with the best prediction accuracy as the water-sensitive intelligent prediction model. Step S108: Quantitatively predict water-sensitive minerals using the water-sensitive intelligent prediction model in step S107. Based on the water sensitivity results output by the water-sensitive intelligent prediction model, identify the sweet spot area of ​​the strongly water-sensitive oil reservoir.

2. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 1, characterized in that, In step S101, environmental correction is performed on the well logging curve data, specifically as follows: A correlation equation is established by fitting stable and unstable logging curves. When the unstable logging curve is affected by the environment, a segment of the curve is fitted by the correlation equation to replace the logging curve segment affected by the environment, thereby achieving environmental correction of the logging curve.

3. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 1, characterized in that, In step S102, multiple methods are used to calculate the clay content, and the calculated clay content is fitted with the measured clay content to determine the optimal clay content calculation method for a single well. Specifically, this includes: Based on the density, neutron, and resistivity curves, clay content a was calculated using the "neutron-density intersection method", clay content b was calculated using the "resistivity-neutron formula method", and clay content c was calculated using the "Schlumberger chart method". The clay contents a, b, and c were then fitted with the measured clay contents, and the method with the best fit was selected as the optimal method for calculating the clay content of a single well.

4. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 2, characterized in that, In step S102, the wellbore is divided into good wellbore sections and poor wellbore sections according to the wellbore environment. Specifically, wellbore sections with collapsed mudstone walls are poor wellbore sections, while those with good mudstone walls are good wellbore sections.

5. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 1, characterized in that, In step S104, the water-sensitive mineral prediction model is constructed using SPSS software.

6. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 1, characterized in that, In step S105, determining different hydration degree ranges based on the absolute content of different water-sensitive minerals specifically includes: If the absolute content of water-sensitive minerals is <4%, it falls within the weak hydration range; If the absolute content of water-sensitive minerals is 4%-6%, it falls within the medium hydration range; If the absolute content of water-sensitive minerals is >6%, it belongs to the strong hydration range.

7. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 1, characterized in that, In step S106, the prediction model includes an SVM random number prediction model, a decision tree C5.0 random number prediction model, and a neural network random number prediction model.

8. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 7, characterized in that, The logging curve data used in step S106 includes at least density, neutron, and resistivity data.

9. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 7, characterized in that, In step S106, the SVM random number prediction model, the decision tree C5.0 random number prediction model, and the neural network random number prediction model are all constructed using SPSS software.

10. The method for intelligent identification of sweet spots in highly water-sensitive oil reservoirs according to claim 6, characterized in that, In step S108, based on the water sensitivity results "strong / medium / weak" output by the water-sensitive intelligent prediction model, the "weak" value is determined to be a sweet spot region of a strong water-sensitive reservoir.

Images