Well logging identification method and system for ultra-deep carbonatite fault control reservoir type

Abstract

The invention discloses a logging identification method for fault control reservoir types of ultra-deep carbonatite, which comprises the following steps of: acquiring a plurality of logging wells with acquired electric imaging data and reservoir types corresponding to the plurality of logging wells determined by the acquired electric imaging data, and simultaneously acquiring conventional logging data of the logging wells, associating different reservoir types with the conventional logging data to form a conventional logging data set marked with the reservoir types; a logging model is constructed, the logging model is trained by using the obtained conventional logging data set marked with the reservoir type, the input of the trained logging model is conventional logging data, and the output of the trained logging model is the reservoir type corresponding to the conventional logging data; and inputting the conventional logging data of the non-electric imaging data of the to-be-logged well into the well logging model obtained by training, wherein the output reservoir type is the reservoir type of the to-be-logged well. The method is used in a well without electricity measurement imaging and is used for quickly and accurately identifying the reservoir type.

Description

Technical Field

[0001] The present invention relates to the technical field of reservoir type identification, and particularly to a logging identification method for ultra-deep carbonate fault-controlled reservoir types. Background Art

[0002] When evaluating carbonate reservoirs, reservoir types play a crucial role. The reservoir storage performances of different reservoir types vary greatly, and their logging response characteristics are also different. Electrical imaging can not only intuitively and quickly identify reservoir types, but also judge the effective shape of the reservoir in combination with other logging information.

[0003] Chinese Patent "A Method for Identifying Reservoirs by Combining Electrical Imaging with Reef Flat Geological Model" with Patent No. CN102011583B proposes a method for identifying reservoirs by combining electrical imaging with reef flat geological model, which requires obtaining electrical imaging data for subsequent analysis and identification.

[0004] Chinese Patent "A Method for Identifying and Interpreting the Three-Dimensional Structure of Paleo-Karst Carbonate Reservoirs" with Patent No. CN103529475B proposes a method for identifying and interpreting the three-dimensional structure of paleo-karst carbonate reservoirs, which also requires obtaining electrical imaging data for subsequent analysis and identification.

[0005] The researches related to the above patents all require obtaining qualified electrical imaging data. Due to the characteristics of deep burial, high temperature and high pressure of the reservoirs in the Shunbei area, the limited effective electrical imaging data was obtained in the early stage due to reasons such as the instrument not meeting the construction requirements or the complex well conditions.

[0006] Therefore, it has become an urgent technical problem for those skilled in the art to urgently provide a method that can accurately identify the carbonate fault-controlled reservoir type in wells without electrical imaging. Summary of the Invention

[0007] To solve the above technical problems, the present invention provides a method for identifying fault-controlled reservoir types, which is used in wells without measured electrical imaging and is used to quickly and accurately identify reservoir types. The specific technical solutions are as follows:

[0008] A logging identification method for ultra-deep carbonate fault-controlled reservoir types includes the following steps:

[0009] Step S1: Obtain multiple wells with collected electrical imaging data and the reservoir types corresponding to the multiple wells determined by the collected electrical imaging data. At the same time, obtain the conventional logging data of these wells, and associate different reservoir types with the conventional logging data to form a set of conventional logging data marked with reservoir types;

[0010] Step S2: Build a logging model. Use the set of conventional logging data marked with reservoir types obtained in Step S1 to train the logging model. The input of the trained logging model is conventional logging data, and the output is the reservoir type corresponding to the conventional logging data.

[0011] Step S3: Input the conventional logging data of the resistivity imaging data of the well to be measured into the logging model trained in Step S2. The output reservoir type is the reservoir type of the well to be measured.

[0012] Preferably, the conventional logging data includes density, acoustic travel time, natural gamma, resistivity, and element content.

[0013] Preferably, the conventional logging data includes resistivity.

[0014] Preferably, the reservoir types include cavernous type, vuggy type, fracture-vuggy type, and fractured type.

[0015] Preferably, in Step S2, the logging model is a trained machine learning model or neural network model. During training, when the logging model is trained iteratively to a certain number of times or the loss function converges, the trained logging model is obtained.

[0016] The technical effects achieved are:

[0017] Establish a method for identifying the reservoir types of carbonate fault-controlled reservoirs, which is used in wells without resistivity imaging to quickly and accurately identify reservoir types, providing good support for the completion plans of new carbonate wells and the reservoir reconstruction plans of old wells.

[0018] The present invention also provides a logging identification system for ultra-deep carbonate fault-controlled reservoir types, including a data acquisition module, a training module, and an identification module. The data acquisition module and the identification module are both electrically connected to the training module, where:

[0019] The data acquisition module is used to acquire multiple wells with resistivity imaging data collected and the reservoir types corresponding to the multiple wells determined by the resistivity imaging data collected. At the same time, the conventional logging data of these wells is acquired, and different reservoir types and conventional logging data are correlated to form a set of conventional logging data marked with reservoir types.

[0020] The training module is used to build a logging model. Use the set of conventional logging data marked with reservoir types obtained in Step S1 to train the logging model. The input of the trained logging model is conventional logging data, and the output is the reservoir type corresponding to the conventional logging data.

[0021] An identification module is configured to input the conventional logging data of the electric imaging data of the well to be measured into the logging model trained in step S2, and the output reservoir type is the reservoir type of the well to be measured.

[0022] It has the same technical effects as above. Description of the Drawings

[0023] Figure 1 It is a schematic flow chart of a logging identification method for ultra-deep carbonate fracture-controlled reservoir types provided;

[0024] Figure 2 It is a block diagram of a logging identification system for ultra-deep carbonate fracture-controlled reservoir types mentioned;

[0025] Figure 3 It is a schematic diagram of typical fracture-type reservoir characteristics of vertical wells and horizontal wells;

[0026] Figure 4 It is a schematic diagram of typical well cave-type reservoir characteristics.

[0027] Figure 1-2 The reference numerals in the figures are as follows:

[0028] 1 Data acquisition module, 2 Training module, 3 Identification module. Detailed Embodiments

[0029] To solve the above technical problems, the present invention provides a logging identification method for ultra-deep carbonate fracture-controlled reservoir types, including the following steps:

[0030] Step S1: Obtain the logging of multiple wells with acquired electric imaging data and the reservoir types corresponding to the multiple logging determined by the acquired electric imaging data. At the same time, obtain the conventional logging data of these wells, and associate different reservoir types with the conventional logging data to form a set of conventional logging data marked with reservoir types;

[0031] Step S2: Construct a logging model, use the set of conventional logging data marked with reservoir types obtained in step S1 to train the logging model. The input of the trained logging model is the conventional logging data, and the output is the reservoir type corresponding to the conventional logging data;

[0032] Step S3: Input the conventional logging data of the electric imaging data of the well to be measured into the logging model trained in step S2, and the output reservoir type is the reservoir type of the well to be measured.

[0033] Establish a method for identifying carbonate fracture-controlled reservoir types, which is used in wells without measured electric imaging to quickly and accurately identify reservoir types. It provides good support for the completion plan of new carbonate wells and the reservoir reconstruction plan of old wells.

[0034] In a specific embodiment, the conventional logging data includes density, acoustic travel time, natural gamma ray, resistivity, and element content.

[0035] In a specific embodiment, the conventional logging data includes resistivity.

[0036] Determine the main reservoir types developed in this block based on the acquired resistivity imaging data of different strips in Shunbei; in different well types, associate different reservoir types with conventional logging response characteristics, mainly the resistivity morphology, to form a set of non-resistivity imaging data marked with reservoir types. Here, different well types include vertical wells and deviated wells, and the resistivity morphology is mainly divided into a double-track type with a large difference between deep and shallow laterologs, a box type with a small difference between deep and shallow laterologs, and a peak type with basically no difference between deep and shallow laterologs. Use resistivity imaging to clarify the reservoir types corresponding to different resistivity morphologies in different well types. As Figure 3 and 4 shown, for example, in Well Shunbei 501 vertical well, when the resistivity shows obvious double-track characteristics with a large difference between deep and shallow laterologs, or in Well Shunbei 71 inclined horizontal well, when the resistivity shows a peak shape with basically no difference between deep and shallow laterologs, the developed reservoir type is fracture type; in vertical wells or deviated wells, such as Well Shunbei 7 sidetrack well, when the resistivity between deep and shallow laterologs shows a significant decrease in the box type and the acoustic travel time, neutron, and density increase significantly, the developed reservoir type is cave type.

[0037] The technical effects brought by this technical solution: Apply this method to wells without resistivity imaging, and effectively and reasonably judge the reservoir types developed in the well according to the different resistivity morphologies corresponding to different well types and the three porosity curves (acoustic travel time, density, and neutron).

[0038] Among them, the reservoir types include cave type, vuggy type, fracture-vuggy type, and fracture type.

[0039] Among them, in step S2, the logging model is a trained machine learning model or neural network model. When the logging model is trained and iterated a certain number of times or the loss function converges during training, a trained logging model is obtained.

[0040] The present invention also provides a logging identification system for ultra-deep carbonate rock fault-controlled reservoir types, including a data acquisition module 1, a training module 2, and an identification module 3. The data acquisition module 1 and the identification module 3 are both electrically connected to the training module 2, where:

[0041] The data acquisition module 1 is used to acquire multiple logged wells with acquired resistivity imaging data and the reservoir types corresponding to the multiple logged wells determined by the acquired resistivity imaging data, and at the same time acquire the conventional logging data of these logged wells, and associate different reservoir types with the conventional logging data to form a set of conventional logging data marked with reservoir types;

[0042] A training module 2, configured to build a logging model, and use the set of conventional logging data marked with reservoir types obtained at step S1 to train the logging model. The input of the trained logging model is conventional logging data, and the output is the reservoir type corresponding to the conventional logging data.

[0043] An identification module 3, configured to input the conventional logging data of the electric imaging data of the well to be measured into the logging model trained in step S2, and the output reservoir type is the reservoir type of the well to be measured.

Claims

1. A logging identification method for ultra-deep carbonate fault-controlled reservoir types, characterized in that, It includes the following steps: Step S1: Obtain multiple logging wells with acquired electrical imaging data and the reservoir types corresponding to the multiple logging wells determined by the acquired electrical imaging data. At the same time, obtain the conventional logging data of these logging wells, and associate different reservoir types with the conventional logging data to form a set of conventional logging data marked with reservoir types; Step S2: Construct a logging model, and use the set of conventional logging data marked with reservoir types obtained in Step S1 to train the logging model. The input of the trained logging model is conventional logging data, and the output is the reservoir type corresponding to the conventional logging data; Step S3: Input the conventional logging data without electrical imaging data of the well to be measured into the logging model trained in Step S2, and the output reservoir type is the reservoir type of the well to be measured.

2. The logging identification method for the ultra-deep carbonate fault-controlled reservoir type according to claim 1, wherein The conventional logging data includes density, acoustic travel time, natural gamma, resistivity, and element content.

3. The logging identification method for the ultra-deep carbonate fault-controlled reservoir type according to claim 1, characterized in that, The conventional logging data includes resistivity.

4. The logging identification method for the ultra-deep carbonate fault-controlled reservoir type according to claim 1, wherein The reservoir types include cavernous, vuggy, fracture-vuggy, and fractured.

5. The logging identification method for the ultra-deep carbonate fault-controlled reservoir type according to claim 1, characterized in that, In Step S2, the logging model is a trained machine learning model or neural network model. During training, when the logging model is trained iteratively to a certain number of times or the loss function converges, the trained logging model is obtained.

6. A logging identification system for ultra-deep carbonate fault-controlled reservoir types, characterized in that, It includes a data acquisition module, a training module, and an identification module. The data acquisition module and the identification module are both electrically connected to the training module, where: The data acquisition module is used to obtain multiple logging wells with acquired electrical imaging data and the reservoir types corresponding to the multiple logging wells determined by the acquired electrical imaging data. At the same time, obtain the conventional logging data of these logging wells, and associate different reservoir types with the conventional logging data to form a set of conventional logging data marked with reservoir types; The training module is used to construct a logging model, and use the set of conventional logging data marked with reservoir types obtained in Step S1 to train the logging model. The input of the trained logging model is conventional logging data, and the output is the reservoir type corresponding to the conventional logging data; The identification module is used to input the conventional logging data without electrical imaging data of the well to be measured into the logging model trained in Step S2, and the output reservoir type is the reservoir type of the well to be measured.

Images