Shallow sea gravity flow channel reservoir porosity prediction method

Abstract

The invention relates to the technical field of oil and gas exploration and development, in particular to a shallow sea gravity flow channel reservoir porosity prediction method which comprises the following steps: S1, determining a target research area; s2, collecting logging curve data of the reservoir of the target research area; s3, taking the logging curves of density, sound wave, resistivity, gamma and neutron porosity as input, taking the actually measured core porosity as output, and constructing and training a depth feedforward neural network model; and S4, predicting the reservoir porosity by using the deep feed-forward neural network model. The defect that the porosity of the shallow sea gravity flow channel reservoir is difficult to accurately predict in the prior art is overcome, the depth feed-forward neural network model is constructed and trained by using the logging curve data, the nonlinear mapping relation between the logging curve and the porosity is established, the reservoir porosity can be accurately predicted for the logging curve by using the model, and the prediction accuracy of the reservoir porosity is improved. And the method is suitable for predicting the porosity of various types of reservoirs.

Description

Technical Field

[0001] The present invention relates to the technical field of oil and gas exploration and development, and more specifically, to a method for predicting the porosity of a shallow sea gravity flow channel reservoir. Background Art

[0002] Porosity and permeability are key parameters for evaluating reservoir quality and storage capacity. In oil and gas exploration and development, porosity and permeability prediction is an important step in sweet spot reservoir evaluation.

[0003] The shallow-water gravity flow deposits in the Yinggehai Basin have loose reservoirs, mainly medium-high porosity and medium-permeability reservoirs, and low-permeability reservoirs are locally developed. The driving types are mainly elastic drive and weak edge water drive. For example, the Yinger 2 reservoir in the Dongfang area of ​​the Yinggehai Basin is mainly medium-high porosity and medium-high permeability, and the lithology is mainly quartz sandstone, with low feldspar and rock debris content. Diagenesis and sedimentation jointly control the reservoir characteristics. Such reservoir gravity flow deposits change rapidly laterally. Due to shallow burial and fast sedimentation, the reservoir is weakly compacted and generally has the characteristics of medium-low porosity and medium-low permeability, and the porosity has obvious heterogeneity.

[0004] For the porosity prediction of the above-mentioned reservoirs, the conventional prediction method is to use core sample test data, establish a rock physics model or cross analysis, and establish a statistical relationship between porosity and logging curves for prediction. Since the rock physics modeling method for predicting porosity and permeability parameters requires more test parameters, when sufficient rock physics parameters cannot be provided, the reliability of the prediction results is low, and the fitting curve obtained by the cross plot of logging curves is too simple and cannot reflect the complex changes in porosity and permeability. Therefore, conventional methods are difficult to accurately predict the porosity of shallow-water gravity flow channel reservoirs. Summary of the invention

[0005] In view of the problem that the porosity prediction methods in the above-mentioned prior art are difficult to accurately predict the porosity of shallow-water gravity flow channel reservoirs, the present invention provides a porosity prediction method for shallow-water gravity flow channel reservoirs, which can improve the accuracy of porosity prediction results.

[0006] In order to solve the above technical problems, the technical solution provided by the present invention is:

[0007] A method for predicting reservoir porosity in a shallow sea gravity flow channel comprises the following steps:

[0008] S1: Determine the target study area;

[0009] S2: Collecting well logging curve data of the reservoir in the target research area, wherein the well logging curve data includes density, acoustic wave, resistivity, gamma, neutron porosity and measured core porosity logging curves;

[0010] S3: using the well logging curves of density, acoustic wave, resistivity, gamma and neutron porosity as input and the measured core porosity as output, to construct and train a deep feedforward neural network model;

[0011] S4: Predicting reservoir porosity using the deep feedforward neural network model to obtain a porosity prediction value.

[0012] In the above technical solution, a deep feedforward neural network model is constructed and trained using well logging data to establish a nonlinear mapping relationship between well logging and porosity. The established deep feedforward neural network model can accurately predict porosity based on well logging and is suitable for predicting porosity of various types of reservoirs.

[0013] Preferably, in step S1, after determining the target study area, the diagenetic characteristics and pore structure of the reservoir in the target study area are also analyzed. By analyzing the diagenetic characteristics and pore structure of the reservoir, the inherent connection and law of the well logging curve data can be better understood, which is helpful for the subsequent selection of more appropriate neural network structure and parameters, so that the neural network model can better adapt to the complexity of reservoir characteristics and improve prediction performance.

[0014] Preferably, in step S1, the diagenetic characteristics and pore structure of the reservoir are identified through core and thin section observations.

[0015] Preferably, in step S2, the logging curve data also includes the measured core permeability logging curve; in step S3, the measured core permeability logging curve is also used as input; after step S4, there is also step S5: according to the fitting relationship between the measured core porosity and the measured core permeability, the permeability prediction value is calculated through the porosity prediction value. Permeability is one of the key parameters in reservoir evaluation, which can reflect the flow ability of fluid in the reservoir. After predicting the porosity, the permeability prediction value is calculated according to the porosity prediction value, which helps to more accurately evaluate the oil and gas content and fluidity of the reservoir.

[0016] Preferably, in the step S2, a cross plot of the measured core porosity and the measured core permeability is drawn according to the two, and a fitting relationship between the measured core porosity and the measured core permeability is determined according to the cross plot.

[0017] Preferably, in step S2, the lithology of the reservoir is also interpreted by analyzing the logging curve of the reservoir; in step S5, the linear relationship between the measured core porosity and the measured core permeability is different under different lithologies. That is to say, when the reservoir contains a variety of different lithology layers, there is not a single linear relationship between porosity and permeability, but a segmented linear relationship. When predicting the permeability later, different lithology layers use different calculation formulas to calculate the predicted permeability value, which can improve the calculation accuracy of the predicted permeability value.

[0018] Preferably, in step S2, after collecting the well logging curve data of the reservoir in the target study area, the abnormal values ​​in the data are first removed, and then the step S3 is entered. This can improve the prediction accuracy of the model.

[0019] Preferably, in step S2, the well logging curve data of the reservoir in the target study area is also normalized, that is, the data is converted to a unified scale so that different data can be compared and weighted, and the influence of the dimension can be eliminated, so that the model can better understand and process the data, thereby improving the training efficiency and prediction performance of the model.

[0020] Preferably, in step S3, the specific steps of constructing and training the deep feedforward neural network model are:

[0021] S31: Determine the model structure, including the input layer, hidden layer, output layer, the number of neurons in each layer, activation function and optimization algorithm;

[0022] S32: dividing the collected logging curve data into a training set, a test set and a validation set;

[0023] S33: training the model using the training set, and during the training process, continuously updating the network weights through a back propagation algorithm to optimize the network structure;

[0024] S34: using the validation set to evaluate model performance, monitor overfitting during training, and then adjust model parameters according to the validation results;

[0025] S35: using the test set data to evaluate the trained model and calculate the error between the predicted porosity and the measured core porosity;

[0026] S36: If the error meets the standard, the trained model is used as the desired deep feedforward neural network model, otherwise return to step S31 to readjust the model structure.

[0027] Preferably, in step S36, if the error is ≤10%, the trained model is used as the desired deep feedforward neural network model; otherwise, the process returns to step S31 to readjust the model structure.

[0028] Beneficial effects of the present invention:

[0029] 1. The present invention uses well logging curve data to construct and train a deep feedforward neural network model, and establishes a nonlinear mapping relationship between the well logging curve and the porosity. Through this model, the reservoir porosity can be accurately predicted based on the well logging curve.

[0030] 2. The present invention is applicable to the prediction of various types of reservoir porosity, and is conducive to in-depth research on the development characteristics and distribution laws of porosity and permeability of shallow-sea gravity flow channels.

[0031] 3. The present invention also fits the relationship between porosity and permeability according to the measured porosity and the measured permeability. Based on this relationship, the predicted permeability value can be calculated by the predicted porosity value, which helps to more accurately evaluate the oil and gas content and fluidity of the reservoir. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 It is a flow chart of a shallow sea gravity flow channel reservoir porosity prediction method;

[0033] Figure 2 This is a comparison of the sedimentary facies and porosity and permeability characteristics of Well A in the Eastern Area;

[0034] Figure 3 This is the porosity prediction result diagram of a well section in the Dongfang area;

[0035] Figure 4 This is the predicted curve of porosity and permeability of Well A in the Eastern Area;

[0036] Figure 5 It is the cross plot of measured core porosity and measured core permeability;

[0037] Figure 6 It is a flow chart of the principle of deep feedforward neural network model to predict porosity;

[0038] Figure 7 This is a comparison chart of the porosity prediction results of Well A in the Eastern Area and the measured core porosity;

[0039] Figure 8 This is a comparison chart of the permeability prediction results of Well A in the Oriental Area and the measured core permeability. DETAILED DESCRIPTION

[0040] The drawings are only for illustrative purposes and cannot be construed as limiting the present invention. To better illustrate the present embodiment, some parts of the drawings may be omitted, enlarged, or reduced, and do not represent the size of the actual product. For those skilled in the art, it is understandable that some well-known structures and their descriptions may be omitted in the drawings. The positional relationships described in the drawings are only for illustrative purposes and cannot be construed as limiting the present invention.

[0041] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "long", "short" and the like indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, they are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. Therefore, the terms describing the positional relationship in the drawings are only used for illustrative purposes and cannot be understood as limitations on this patent. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0042] The technical solution of the present invention is further described in detail below through specific embodiments and in conjunction with the accompanying drawings:

[0043] Example 1

[0044] This embodiment is a first embodiment of a shallow sea gravity flow channel reservoir porosity prediction method. Figure 1 As shown, it includes the following steps:

[0045] S1: Determine the target research area. In this embodiment, the second section of the Yinggehai gravity flow reservoir in the Dongfang area is taken as the target research area.

[0046] S2: Collect well logging data of the reservoir in the target study area. The well logging data include density, acoustic wave, resistivity, gamma, neutron porosity and measured core porosity logging curves.

[0047] S3: Build and train a deep feedforward neural network model using the well logging curves of density, acoustic wave, resistivity, gamma and neutron porosity as input and the measured core porosity as output.

[0048] S4: Use the deep feedforward neural network model to predict reservoir porosity and obtain the porosity prediction value.

[0049] Furthermore, in step S1, after determining the target study area, the diagenetic characteristics and pore structure of the reservoir in the target study area are also analyzed. By analyzing the diagenetic characteristics and pore structure of the reservoir, the intrinsic connection and regularity of the logging curve data can be better understood, which is helpful for the subsequent selection of more suitable neural network structures and parameters, so that the neural network model can better adapt to the complexity of reservoir characteristics and improve the prediction performance. In this embodiment, the shallow-water gravity flow sandstone reservoir of the second section of Ying in the target study area is mainly quartz sandstone. The overall burial depth of the sandstone reservoir of the second section of Ying is shallow, the compaction effect is weak, and the cementation has a destructive effect on the physical properties of the reservoir; the dissolution effect increases the pore volume, which has a certain positive effect on the physical properties of the reservoir; the favorable reservoir is a channelized lobe sedimentary microfacies, and there are differences between different types of channel sand bodies. Specifically, Figure 2 shown.

[0050] Specifically, in step S1, the diagenetic characteristics and pore structure of the reservoir are identified through core and thin section observations.

[0051] The effect verification of this example: The porosity of Well A in the Eastern Area is compared with the measured results. Figure 3 As shown in Figure 2, the porosity prediction results of a well section in the Dongfang area are as follows: Figure 4 It can be seen that the porosity distribution is consistent with the shallow sea gravity flow distribution trend revealed by drilling and seismic data, which verifies the effectiveness of the method of this embodiment.

[0052] Beneficial effects of this embodiment:

[0053] 1. Use well logging data to build and train a deep feedforward neural network model and establish a nonlinear mapping relationship between well logging and porosity. This model can accurately predict reservoir porosity based on well logging.

[0054] 2. The method of this embodiment is applicable to the prediction of the porosity of various types of reservoirs, and is conducive to in-depth research on the development characteristics and distribution laws of porosity and permeability of shallow-water gravity flow channels.

[0055] Example 2

[0056] This embodiment is a second embodiment of a shallow sea gravity flow channel reservoir porosity prediction method. Based on the embodiment 1, this embodiment further supplements step S2.

[0057] In step S2, the logging curve data also includes the measured core permeability logging curve; in step S3, the measured core permeability logging curve is also used as input; after step S4, there is also step S5: according to the fitting relationship between the measured core porosity and the measured core permeability, the permeability prediction value is calculated through the porosity prediction value. Permeability is one of the key parameters in reservoir evaluation, which can reflect the flow ability of fluid in the reservoir. After predicting the porosity, the permeability prediction value is calculated based on the porosity prediction value, which helps to more accurately evaluate the oil and gas content and fluidity of the reservoir.

[0058] Further, in step S2, a cross-plot of the measured core porosity and the measured core permeability is drawn, as shown in FIG. Figure 5 As shown in FIG. 1 , the fitting relationship between the measured core porosity and the measured core permeability is determined based on the cross plot.

[0059] Furthermore, in step S2, the reservoir lithology is interpreted by analyzing the reservoir logging curve; in step S5, the linear relationship between the measured core porosity and the measured core permeability is different for different lithologies. Figure 5 It can be seen that there is a piecewise linear correlation feature between porosity and permeability. When the porosity is less than 25%, the dry layer and the water layer are dominant, and the slope of the linear relationship between the two is small; when the porosity is greater than 25%, the gas-bearing sandstone is dominant, and the slope of the linear fitting curve increases. It can be seen that when the reservoir contains a variety of different lithology layers, there is not a single linear relationship between porosity and permeability, but a piecewise linear relationship. When predicting the permeability later, different lithology layers use different calculation formulas to calculate the predicted permeability values, which can improve the accuracy of the calculation results of the permeability prediction values.

[0060] The other features, working principles and beneficial effects of this embodiment are consistent with those of Embodiment 1.

[0061] Example 3

[0062] This embodiment is the third embodiment of a shallow sea gravity flow channel reservoir porosity prediction method. Based on the embodiment 1, this embodiment further supplements step S2 and step S3.

[0063] In step S2, after collecting the well logging curve data of the reservoir in the target study area, the abnormal values ​​in the data are first removed, and then step S3 is entered. This can improve the prediction accuracy of the model.

[0064] Furthermore, in step S2, the well logging curve data of the reservoir in the target study area is normalized, that is, the data is converted to a unified scale so that different data can be compared and weighted, and the influence of the dimension can be eliminated, so that the model can better understand and process the data, thereby improving the training efficiency and prediction performance of the model.

[0065] Furthermore, combined with Figure 1 and Figure 6 As shown, in step S3, the specific steps of constructing and training the deep feedforward neural network model are:

[0066] S31: Determine the model structure, including the input layer, hidden layer, output layer, the number of neurons in each layer, activation function and optimization algorithm;

[0067] S32: dividing the collected logging curve data into a training set, a test set and a validation set;

[0068] S33: Use the training set to train the model. During the training process, the network weights are continuously updated through the back propagation algorithm to optimize the network structure.

[0069] S34: Use the validation set to evaluate model performance, monitor overfitting during training, and then adjust model parameters based on the validation results;

[0070] S35: Use the test set data to evaluate the trained model and calculate the error between the predicted porosity and the measured core porosity;

[0071] S36: If the error meets the standard, the trained model is used as the desired deep feedforward neural network model, otherwise return to step S31 to readjust the model structure.

[0072] Further, in step S36, if the error is ≤10%, the trained model is used as the desired deep feedforward neural network model; otherwise, the process returns to step S31 to readjust the model structure. Figure 7 As shown in Figure 2, the smaller the GR (gamma) value, the lower the mud content and the corresponding porosity increases. Figure 7 It can be seen that the porosity prediction results have an obvious linear relationship with the core measured data, and the calculated mean square error is 8.58%, which meets the error standard.

[0073] Furthermore, under the premise that the error result of the porosity prediction value meets the standard, the permeability curve is calculated using the fitting relationship between porosity and permeability. Figure 8 As shown, the mean square error of the predicted permeability is 32 mD.

[0074] The other features, working principles and beneficial effects of this embodiment are consistent with those of Embodiment 2.

[0075] In the specific contents of the above-mentioned specific implementation methods, the various technical features can be combined in any non-contradictory manner. In order to make the description concise, not all possible combinations of the above-mentioned technical features are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0076] Obviously, the above embodiments of the present invention are merely examples for clearly explaining the present invention, and are not intended to limit the implementation methods of the present invention. For those skilled in the art, other different forms of changes or modifications can be made based on the above description, and it is not necessary and impossible to list all the implementation methods here. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the claims of the present invention.

Claims

1. A method for predicting reservoir porosity in shallow sea gravity flow channels, characterized in that: The following steps are involved: S1: Determine the target study area; S2: Collecting well logging curve data of the reservoir in the target research area, wherein the well logging curve data includes density, acoustic wave, resistivity, gamma, neutron porosity and measured core porosity logging curves; S3: using the well logging curves of density, acoustic wave, resistivity, gamma and neutron porosity as input and the measured core porosity as output, to construct and train a deep feedforward neural network model; S4: Predicting reservoir porosity using the deep feedforward neural network model to obtain a porosity prediction value.

2. The method for predicting reservoir porosity of shallow sea gravity flow channel according to claim 1, characterized in that: In step S1, after the target study area is determined, the diagenetic characteristics and pore structure of the reservoir in the target study area are analyzed.

3. The method for predicting reservoir porosity of shallow sea gravity flow channel according to claim 2, characterized in that: In step S1, the diagenetic characteristics and pore structure of the reservoir in the target study area are identified through core and thin section observations.

4. The method for predicting reservoir porosity of shallow sea gravity flow channel according to claim 1, characterized in that: In step S2, the logging curve data also includes a measured core permeability logging curve; in step S3, the measured core permeability logging curve is also used as input; after step S4, there is a step S5: according to the fitting relationship between the measured core porosity and the measured core permeability, the permeability prediction value is calculated through the porosity prediction value.

5. The method for predicting reservoir porosity of shallow sea gravity flow channel according to claim 4, characterized in that: In the step S2, a cross plot of the measured core porosity and the measured core permeability is drawn according to the two, and a fitting relationship between the measured core porosity and the measured core permeability is determined according to the cross plot.

6. A shallow sea gravity flow channel reservoir porosity prediction method according to claim 5, characterized in that: In the step S2, the lithology of the reservoir is also interpreted by analyzing the logging curve of the reservoir; in the step S5, the linear relationship between the measured core porosity and the measured core permeability is different under different lithologies.

7. The method for predicting reservoir porosity of shallow sea gravity flow channel according to claim 1, characterized in that: In the step S2, after collecting the well logging curve data of the reservoir in the target study area, the abnormal values ​​in the data are first removed, and then the step S3 is entered.

8. The method for predicting reservoir porosity of shallow sea gravity flow channel according to claim 1, characterized in that: In the step S2, the well logging curve data of the reservoir in the target study area is also normalized.

9. A shallow sea gravity flow channel reservoir porosity prediction method according to any one of claims 1 to 8, characterized in that: In step S3, the specific steps of constructing and training the deep feedforward neural network model are: S31: Determine the model structure, including the input layer, hidden layer, output layer, the number of neurons in each layer, activation function and optimization algorithm; S32: dividing the collected logging curve data into a training set, a test set and a validation set; S33: training the model using the training set, and during the training process, continuously updating the network weights through a back propagation algorithm to optimize the network structure; S34: using the validation set to evaluate model performance, monitor overfitting during training, and then adjust model parameters according to the validation results; S35: using the test set data to evaluate the trained model and calculate the error between the predicted porosity and the measured core porosity; S36: If the error meets the standard, the trained model is used as the desired deep feedforward neural network model, otherwise return to step S31 to readjust the model structure.

10. A shallow sea gravity flow channel reservoir porosity prediction method according to claim 9, characterized in that: In step S36, if the error is ≤10%, the trained model is used as the desired deep feedforward neural network model; otherwise, the process returns to step S31 to readjust the model structure.

Images