A 3D modeling method for shale reservoirs based on big data

Abstract

The present invention discloses a shale reservoir three-dimensional modeling method based on big data, and the shale reservoir three-dimensional modeling method includes the following steps: S1, constructing a pre-drilling three-dimensional geological model of the target area by using the Kriging interpolation method, and measuring the downhole data of the drill bit during drilling; S2, transmitting the downhole data to the software platform in real time, and using the software platform to correct the pre-drilling three-dimensional geological model, and obtaining the three-dimensional geological model; S3, using the three-dimensional geological model to display the underground stratum conditions in real time on the display terminal; S4, the staff observes the relationship between the drill bit and the stratum through the display terminal, and makes adaptive adjustments to ensure that the drill bit remains in the target layer. The present invention uses the drilling data to establish the pre-drilling three-dimensional geological model, and corrects the model in real time according to the drilling data, which can ensure the accuracy of the model, provide the model with functions such as slicing, scaling, and rotation, and intuitively display the mutual relationship of the wellbore trajectory in three-dimensional space, which helps to reduce the risk of wellbore collision.

Description

Technical Field

[0001] The present invention relates to the field of shale storage, and in particular to a shale reservoir three-dimensional modeling method based on big data. Background Art

[0002] At present, due to the strong heterogeneity of Longmaxi shale and the thin target reservoir, the drill bit is prone to deviate from the target layer during the drilling process, and geological guidance is particularly important. At present, there is only a geological guidance software with two-dimensional display function in China, which is not intuitive enough to observe and cannot provide a better basis for the adjustment of the drill bit. Therefore, there is an urgent need for a three-dimensional display software platform that can accurately display the underground stratum structure in real time.

[0003] Two-dimensional geosteering generally uses adjacent data as a reference for steering modeling, mainly considering the vertical distribution of formation attributes, and the steering process is relatively simple. The data used is mainly limited to geological and well logging data, which can meet the requirements of horizontal well geosteering to a certain extent; seismic data contains more potential formation and oil and gas information, which can reflect the formation morphology, rock properties and oil and gas reservoir location, etc., and is almost not or rarely used in the current two-dimensional geosteering process. At present, domestic gas fields generally use MD (with GR or GR+RT) or LWD (with GR, RT) combined with comprehensive logging for horizontal well geosteering.

[0004] Two-dimensional geosteering cannot accurately reflect the three-dimensional spatial distribution of the wellbore trajectory, as well as the changes in the stratigraphic structure and reservoir anisotropy around the wellbore trajectory. It cannot accurately describe faults and fractures, and cannot utilize the reservoir inversion results of seismic data. This brings certain uncertainties to geosteering work.

[0005] Currently, no effective solution has been proposed for the problems in the related technologies. Summary of the invention

[0006] In view of the problems in the related technology, the present invention proposes a shale reservoir three-dimensional modeling method based on big data to overcome the above-mentioned technical problems existing in the existing related technology.

[0007] To this end, the specific technical solution adopted by the present invention is as follows:

[0008] A shale reservoir three-dimensional modeling method based on big data, the shale reservoir three-dimensional modeling method comprising the following steps:

[0009] S1. Construct a pre-drilling 3D geological model of the target area by using the Kriging interpolation method, and measure the downhole data of the drill bit during drilling;

[0010] S2, transmitting the downhole data to the software platform in real time, and using the software platform to correct the pre-drilling three-dimensional geological model, and obtain a three-dimensional geological model;

[0011] S3, using the three-dimensional geological model to display the underground strata conditions on the display terminal in real time;

[0012] S4. The staff observes the relationship between the drill bit and the formation through the display terminal and makes adaptive adjustments to ensure that the drill bit remains in the target layer.

[0013] Furthermore, the construction of the Kriging interpolation method includes the following steps:

[0014] Construct variogram;

[0015] Use the known points that are effectively searched in the neighborhood as input points, and calculate the variance function values ​​between any two input points and between the point to be interpolated and all input points;

[0016] Assign values ​​to the K matrix and find its inverse to obtain the inverse matrix of the K matrix;

[0017] The weight coefficient is obtained by multiplying the inverse matrix of the K matrix by the M matrix, and the attribute value of the point to be interpolated is obtained by weighting.

[0018] Furthermore, the K matrix is ​​constructed as follows:

[0019]

[0020] In the formula, γ(v n ,v n ) is the variogram value between the nth known point and the nth known point, and the values ​​in the first n rows and n columns of the K matrix are the variogram values ​​between any two known points.

[0021] Furthermore, the M matrix is ​​constructed as follows:

[0022]

[0023] In the formula, γ(v n ,v n ) is the variogram value between the nth known point and the nth known point, and the first n rows of the M matrix are the variogram values ​​between the point currently to be estimated and each known point.

[0024] Furthermore, the method of constructing a pre-drilling three-dimensional geological model of the target area by the Kriging interpolation method and measuring downhole data of the drill bit during drilling includes the following steps:

[0025] S11, establishing a blank model;

[0026] S12, according to the size of the drilled block, set the length of the three directions of the 3D model and the step length of each grid, and read the adjacent well data and seismic data;

[0027] S13, establishing a three-dimensional geological model before drilling through the attribute values ​​of the points to be interpolated by the Kriging interpolation method, and using the downhole data measured in real time by the drill bit during the drilling process.

[0028] Furthermore, the real-time transmission of downhole data to the software platform, and the use of the software platform to correct the pre-drilling three-dimensional geological model, and obtaining the three-dimensional geological model includes the following steps:

[0029] S21, the software platform receives the downhole data transmitted in real time, and modifies the pre-drilling three-dimensional geological model in real time according to the drilling data;

[0030] S22, slicing, scaling and rotating the three-dimensional geological model before drilling, and continuously correcting the three-dimensional geological model before drilling;

[0031] S23. Continuously modify until the desired three-dimensional geological model is obtained.

[0032] Furthermore, the method of using the three-dimensional geological model to display the underground stratum conditions in real time on the display terminal includes the following steps:

[0033] S31. Construct sedimentary microfacies identification model;

[0034] S32, processing the three-dimensional geological model in the display terminal through the sedimentary microfacies identification model to identify the underground strata in the three-dimensional geological model;

[0035] S33. Display the underground strata in the three-dimensional geological model on the display terminal in a visual manner.

[0036] Furthermore, the construction of the sedimentary microfacies identification model comprises the following steps:

[0037] S311, inputting underground stratum images to the GoogleNet input layer;

[0038] S312, extracting underground stratum image features through convolution operation of the convolution layer and pooling processing of the downsampling layer;

[0039] S313, inputting the fully connected layer to determine the target category of the underground stratum image;

[0040] S314. Convert the output values ​​of multiple classifications into corresponding probability scores through the softmax loss function, and output the classification results.

[0041] Furthermore, the model formula of the GoogleNet is:

[0042]

[0043] In the formula, X is the input signal dimension, Y is the output signal dimension, W is the convolution kernel, b is the bias term, θ is the model weight parameter, and x is the input object.

[0044] Furthermore, the formula of the softmax loss function is:

[0045]

[0046] In the formula, softmax(z i ) is the probability score of the i-th neuron, z i is the output value of the i-th neuron, n is the number of neurons involved in classification, and e is the base of the natural logarithm.

[0047] The beneficial effects of the present invention are:

[0048] 1. The present invention uses the drilling data to establish a three-dimensional geological model before drilling, and corrects the model in real time according to the drilling data, which can ensure the accuracy of the model. At the same time, it provides slicing, scaling, rotation and other functions for the model, intuitively displays the relationship between the wellbore trajectory in three-dimensional space, and helps to reduce the risk of wellbore collision. It combines two-dimensional and three-dimensional geological models to make real-time geological reservoir decisions, and can achieve relatively accurate simulation of extremely complex geological conditions, ensuring that the drill bit is always in the target layer.

[0049] 2. The present invention can realize real-time and accurate display of underground stratum conditions, and at the same time provide functions such as rotary slicing, so as to facilitate ground staff to check and adjust the drill bit and ensure that the drill bit always drills in the target layer.

[0050] 3. The present invention provides the advantages of Kriging interpolation by comparing it with several other interpolation methods. Kriging has great advantages over other methods. While ensuring the interpolation accuracy, the complexity of Kriging interpolation is also improved accordingly, which greatly improves the interpolation efficiency and can be better applied to practical projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] Figure 1 The present invention is a flowchart of a method for three-dimensional modeling of shale reservoirs based on big data according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] To further illustrate each embodiment, the present invention provides drawings, which are part of the disclosure of the present invention and are mainly used to illustrate the embodiments and can be used in conjunction with the relevant descriptions in the specification to explain the operating principles of the embodiments. With reference to these contents, ordinary technicians in the field should be able to understand other possible implementations and advantages of the present invention. The components in the figures are not drawn to scale, and similar component symbols are generally used to represent similar components.

[0054] According to an embodiment of the present invention, a shale reservoir three-dimensional modeling method based on big data is provided.

[0055] The present invention is further described with reference to the accompanying drawings and specific embodiments. Figure 1 As shown, according to the big data-based shale reservoir three-dimensional modeling method according to an embodiment of the present invention, the shale reservoir three-dimensional modeling method includes the following steps:

[0056] S1. Construct a pre-drilling 3D geological model of the target area by using the Kriging interpolation method, and measure the downhole data of the drill bit during drilling;

[0057] In one embodiment, the construction of the Kriging interpolation method includes the following steps:

[0058] Construct variogram;

[0059] Use the known points that are effectively searched in the neighborhood as input points, and calculate the variance function values ​​between any two input points and between the point to be interpolated and all input points;

[0060] Assign values ​​to the K matrix and find its inverse to obtain the inverse matrix of the K matrix;

[0061] The weight coefficient is obtained by multiplying the inverse matrix of the K matrix by the M matrix, and the attribute value of the point to be interpolated is obtained by weighting.

[0062] In one embodiment, the K matrix is ​​constructed as follows:

[0063]

[0064] In the formula, γ(v n ,v n ) is the variogram value between the nth known point and the nth known point, and the values ​​in the first n rows and n columns of the K matrix are the variogram values ​​between any two known points.

[0065] In one embodiment, the M matrix is ​​constructed as follows:

[0066]

[0067] In the formula, γ(v n ,vn ) is the variogram value between the nth known point and the nth known point, and the first n rows of the M matrix are the variogram values ​​between the point currently to be estimated and each known point.

[0068] In one embodiment, the method of constructing a pre-drilling three-dimensional geological model of the target area by the Kriging interpolation method and measuring downhole data of the drill bit during drilling comprises the following steps:

[0069] S11, establishing a blank model;

[0070] S12, according to the size of the drilled block, set the length of the three directions of the 3D model and the step length of each grid, and read the adjacent well data and seismic data;

[0071] S13, establishing a three-dimensional geological model before drilling through the attribute values ​​of the points to be interpolated by the Kriging interpolation method, and using the downhole data measured in real time by the drill bit during the drilling process.

[0072] Specifically, since we want to obtain an unbiased estimate, we must meet the unbiased condition, that is, the weight value of the weight coefficient is 1; at the same time, the regionalized variable needs to meet the second-order equilibrium assumption. Combining the above conditions, we can get the weight coefficient;

[0073] In addition, in recent years, scientific computing visualization technology has developed rapidly and has been widely used in various engineering and computing fields, especially in geological three-dimensional modeling. In order to truly reflect the geological model, we need to visualize large data sets. However, under the existing technical conditions and the limitations of many factors in actual situations, the data points that can be collected are extremely small compared to the point set required for data visualization, so relying solely on limited discrete data points is far from meeting the visualization requirements. In order to solve this problem, interpolation technology has emerged;

[0074] The so-called interpolation technology is to use some corresponding analysis and research models to study the attribute information of these sample points based on the known limited discrete data point information, establish a mapping relationship between its spatial domain and attribute domain, quantify this mapping relationship, and finally estimate the value at the unknown point based on this established mapping relationship. Due to the wide application of interpolation technology, many interpolation methods have also emerged. The existing commonly used interpolation methods are: inverse distance method, radial basis function method, Thiessen polygon method, spline interpolation, Kriging method, etc.

[0075] The basic idea of ​​the inverse distance method is to first calculate the distance between the known point and the unknown point, and then use the inverse of the distance to perform a weighted average on the attribute values ​​of the known points to obtain the unknown attribute value. This method is simple to understand and easy to implement, but it only uses the distance factor as the basis for interpolation, and does not take into account the relationship between the known points themselves, so the estimation accuracy is limited.

[0076] The Thiessen polygon method uses a known point as the center and divides the interpolation area into thousands of polygons. The known points in the sub-area are used as the input of the points to be interpolated in the polygon area for interpolation calculation. This method is intuitive, simple, and easy to understand, but when the sampling points are distributed very unevenly, the polygons formed are prone to singular polygons, resulting in low interpolation accuracy. In addition, the Thiessen polygon method is only applicable to two-dimensional planes, and its application has great limitations.

[0077] As a spatial prediction method widely used in many fields, Kriging was first proposed by South African mining engineer Krige. The object of the Kriging method is the variogram, which can give an estimate of the regionalized variable in a limited area, and this estimate is optimal and unbiased. The basic idea of ​​Kriging interpolation is to first determine the influence of the known point on the valuation point, which is represented by the weight coefficient, and then use the weight coefficient to perform weighted summation of the attribute values ​​of the known points to obtain the attribute value of the interpolated point. Therefore, Kriging interpolation is a linear interpolation method. In addition, the reason why it is said to be the optimal and unbiased estimation method is that its mathematical model is established on the basis of analyzing the regionalized variable and the variogram. From the introduction of the above methods, it can be seen that the Kriging method not only considers the distance factor, but also takes the spatial correlation between the known points as an important theoretical premise for estimation, and quantifies this spatial correlation through the study of the variogram. Therefore, the error of the result obtained by the Kriging interpolation method is very small and more accurate. The complexity of Kriging interpolation is also correspondingly improved, which greatly improves the interpolation efficiency and can be better applied to practical engineering.

[0078] Kriging has great advantages over other methods. While ensuring the interpolation accuracy, the complexity of Kriging interpolation is also increased accordingly.

[0079] S2, transmitting the downhole data to the software platform in real time, and using the software platform to correct the pre-drilling three-dimensional geological model, and obtain a three-dimensional geological model;

[0080] In one embodiment, the real-time transmission of downhole data to the software platform, and the use of the software platform to correct the pre-drilling three-dimensional geological model, and obtaining the three-dimensional geological model includes the following steps:

[0081] S21, the software platform receives the downhole data transmitted in real time, and modifies the pre-drilling three-dimensional geological model in real time according to the drilling data;

[0082] S22, slicing, scaling and rotating the three-dimensional geological model before drilling, and continuously correcting the three-dimensional geological model before drilling;

[0083] S23. Continuously modify until the desired three-dimensional geological model is obtained.

[0084] S3, using the three-dimensional geological model to display the underground strata conditions on the display terminal in real time;

[0085] In one embodiment, the method of using a three-dimensional geological model to display the underground strata conditions on a display terminal in real time includes the following steps:

[0086] S31. Construct sedimentary microfacies identification model;

[0087] S32, processing the three-dimensional geological model in the display terminal through the sedimentary microfacies identification model to identify the underground strata in the three-dimensional geological model;

[0088] S33. Display the underground strata in the three-dimensional geological model on the display terminal in a visual manner.

[0089] In one embodiment, constructing a sedimentary microfacies identification model comprises the following steps:

[0090] S311, inputting underground stratum images to the GoogleNet input layer;

[0091] S312, extracting underground stratum image features through convolution operation of the convolution layer and pooling processing of the downsampling layer;

[0092] S313, inputting the fully connected layer to determine the target category of the underground stratum image;

[0093] S314. Convert the output values ​​of multiple classifications into corresponding probability scores through the softmax loss function, and output the classification results.

[0094] In one embodiment, the model formula of the GoogleNet is:

[0095]

[0096] In the formula, X is the input signal dimension, Y is the output signal dimension, W is the convolution kernel, b is the bias term, θ is the model weight parameter, and x is the input object.

[0097] In one embodiment, the formula of the softmax loss function is:

[0098]

[0099] In the formula, softmax(z i ) is the probability score of the i-th neuron, z i is the output value of the i-th neuron, n is the number of neurons involved in classification, and e is the base of the natural logarithm.

[0100] Specifically, the training process of the convolutional neural network is essentially the optimization and update process of the network weights. Adjusting the GoogleNet parameters according to the well logging image dataset and selecting appropriate hyperparameters are helpful for the retrieval of well logging images.

[0101] S4. The staff observes the relationship between the drill bit and the formation through the display terminal and makes adaptive adjustments to ensure that the drill bit remains in the target layer.

[0102] Specifically, intuitively displaying the mutual relationship of wellbore trajectories in three-dimensional space helps reduce the risk of wellbore collision, ensures the stable progress of construction, and provides stable operation guarantees for construction.

[0103] In summary, with the help of the above technical scheme of the present invention, the present invention uses the drilling data to establish a three-dimensional geological model before drilling, and corrects the model in real time according to the drilling data, which can ensure the accuracy of the model, and provide the model with slicing, scaling, rotation and other functions, intuitively display the relationship between the wellbore trajectory in the three-dimensional space, which helps to reduce the risk of wellbore collision; the two-dimensional and three-dimensional geological models are combined to make real-time geological reservoir decisions, and can simulate the extremely complex geological conditions more accurately, ensuring that the drill bit can always be in the target layer; the present invention can realize real-time and accurate display of underground stratum conditions, and provide functions such as rotation slicing, which is convenient for ground staff to check and adjust the drill bit, ensuring that the drill bit is always drilling in the target layer; the present invention compares with several other interpolation methods to give the advantages of Kriging interpolation. Compared with several other methods, the Kriging method has great advantages. While ensuring the interpolation accuracy, the complexity of Kriging interpolation is also correspondingly improved, which greatly improves the interpolation efficiency, and can be better applied to practical engineering.

[0104] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A 3D modeling method for shale reservoirs based on big data, characterized in that: The shale reservoir three-dimensional modeling method comprises the following steps: S1. Construct a pre-drilling 3D geological model of the target area by using the Kriging interpolation method, and measure the downhole data of the drill bit during drilling; S2, transmitting the downhole data to the software platform in real time, and using the software platform to correct the pre-drilling three-dimensional geological model, and obtain a three-dimensional geological model; S3, using the three-dimensional geological model to display the underground strata conditions on the display terminal in real time; S4. The staff observes the relationship between the drill bit and the formation through the display terminal and makes adaptive adjustments to ensure that the drill bit remains in the target layer; The construction of the Kriging interpolation method includes the following steps: Construct variogram; Use the known points that are effectively searched in the neighborhood as input points, and calculate the variance function values ​​between any two input points and between the point to be interpolated and all input points; Assign values ​​to the K matrix and find its inverse to obtain the inverse matrix of the K matrix; The weight coefficient is obtained by multiplying the inverse matrix of K matrix with M matrix, and the attribute value of the point to be interpolated is obtained by weighting; The K matrix is ​​constructed as follows: In the formula, γ(v n ,v n ) is the variogram value between the n-th known point and the n-th known point, and the values ​​in the first n rows and n columns of the K matrix are the variogram values ​​between any two known points; The M matrix is ​​constructed as follows: In the formula, γ(v n ,v n ) is the variogram value between the n-th known point and the n-th known point, and the first n rows of the M matrix are the variogram values ​​between the point currently to be estimated and each known point; The method of constructing a pre-drilling three-dimensional geological model of the target area by using the Kriging interpolation method and measuring downhole data of the drill bit during drilling comprises the following steps: S11, establishing a blank model; S12, according to the size of the drilled block, set the length of the three directions of the 3D model and the step length of each grid, and read the adjacent well data and seismic data; S13, establishing a three-dimensional geological model before drilling through the attribute values ​​of the points to be interpolated by the Kriging interpolation method, and using the downhole data measured in real time by the drill bit during the drilling process.

2. The method for three-dimensional modeling of shale reservoirs based on big data according to claim 1, characterized in that: The real-time transmission of downhole data to the software platform, and the use of the software platform to correct the pre-drilling three-dimensional geological model, and obtaining the three-dimensional geological model includes the following steps: S21, the software platform receives the downhole data transmitted in real time, and modifies the pre-drilling three-dimensional geological model in real time according to the drilling data; S22, slicing, scaling and rotating the three-dimensional geological model before drilling, and continuously correcting the three-dimensional geological model before drilling; S23. Continuously modify until the desired three-dimensional geological model is obtained.

3. The method for three-dimensional modeling of shale reservoirs based on big data according to claim 1, characterized in that: The method of using the three-dimensional geological model to display the underground stratum conditions on the display terminal in real time includes the following steps: S31. Construct sedimentary microfacies identification model; S32, processing the three-dimensional geological model in the display terminal through the sedimentary microfacies identification model to identify the underground strata in the three-dimensional geological model; S33. Display the underground strata in the three-dimensional geological model on the display terminal in a visual manner.

4. The method for three-dimensional modeling of shale reservoirs based on big data according to claim 3 is characterized in that: The construction of the sedimentary microfacies identification model comprises the following steps: S311, inputting underground stratum images to the GoogleNet input layer; S312, extracting underground stratum image features through convolution operation of the convolution layer and pooling processing of the downsampling layer; S313, inputting the fully connected layer to determine the target category of the underground stratum image; S314. Convert the output values ​​of multiple classifications into corresponding probability scores through the softmax loss function, and output the classification results.

5. The method for three-dimensional modeling of shale reservoirs based on big data according to claim 4, characterized in that: The model formula of the GoogleNet is: In the formula, X is the input signal dimension, Y is the output signal dimension, W is the convolution kernel, b is the bias term, θ is the model weight parameter, and x is the input object.

6. The method for three-dimensional modeling of shale reservoirs based on big data according to claim 5, characterized in that: The formula of the softmax loss function is: In the formula, softmax(z i ) is the probability score of the i-th neuron, z i is the output value of the i-th neuron, n is the number of neurons involved in classification, and e is the base of the natural logarithm.

Images