A reservoir pressure characterization method under small sample speed conditions

Abstract

The application discloses a reservoir pressure characterization method under a small sample speed condition, which comprises the following steps: establishing a forward algorithm of P-wave and S-wave velocity data, and comparing measured P-wave and S-wave velocities with the forward velocities for quality control; calculating a pore pressure curve, performing quality control on predicted velocity data, and correcting velocity forward parameters; processing post-stack seismic data, constructing seismic attributes sensitive to seismic velocity, and obtaining a high-precision velocity field; and introducing a machine learning velocity field construction algorithm factor into an Eton equation for pore pressure prediction. The application is applied to a scenario of a deficient wellbore velocity, is based on a quality control method based on velocity forward curve and measured pressure and velocity data, and uses a high-precision velocity field construction method of post-stack velocity sensitive seismic attributes, so that the application can solve the actual demand for the high-precision velocity field in the reservoir pressure prediction process, greatly improves the precision and efficiency of the reservoir pressure prediction, and verifies the high efficiency and applicability of the pressure prediction through three-dimensional seismic target area testing.

Description

Technical Field

[0001] This invention belongs to the field of oil and gas field exploration and development technology, and in particular relates to a reservoir pressure characterization method under small sample velocity conditions. Background Technology

[0002] Seismic pressure prediction is widely used in basin dynamics research, deep reservoir evaluation, unconventional reservoir exploration and development, oil and gas reservoir protection during drilling, and prevention of complex downhole accidents. For a long time, seismic pressure prediction has been limited by the accuracy of the velocity field in seismic data processing. However, in practical production applications, the long processing cycle and large workload of pre-stack seismic data have restricted the application of pressure prediction based on seismic velocity fields.

[0003] The current basic status of velocity field construction is as follows:

[0004] First, the prediction of overpressure layers using seismic methods mainly utilizes the characteristic of low seismic velocity anomalies in overpressure layers. Therefore, the accuracy of the seismic velocity field is related to the success or failure of pressure prediction.

[0005] Secondly, inversion methods are valuable in pressure prediction; however, any inversion method, whether pre-stack or post-stack, must be based on an acceptable rock velocity model. Pre-drilling seismic techniques typically cannot detect pressure data in thin, overpressured sandstone and shale formations, which often cause unexpected problems during drilling. Formation pressure prediction based on high-precision velocity field analysis will undoubtedly be widely used in future reservoir engineering.

[0006] Third, seismic processing velocity fields are currently commonly used as the main velocity fields for pressure prediction. However, to obtain a high-precision velocity field using pre-stack seismic processing, detailed interpretation and analysis of velocity points are required, which is time-consuming and costly. Due to technical and workload limitations, it is generally not feasible in the later stages of reservoir prediction and seismic interpretation.

[0007] Fourth, using seismic calibration to obtain wellpoint velocity data and combining it with the stacked velocity field obtained from pre-stack seismic data processing to construct a velocity field is also a common method for establishing a velocity field. Although this method can establish a velocity field for application time-depth conversion, the spatial velocity accuracy is limited due to the sparse velocity analysis points, making it impossible to standardize the spatial variation characteristics of velocity in complex underground reservoir media. The application scenarios of the obtained velocity field will also be limited.

[0008] Fifth, with the rise of big data and artificial intelligence technologies, pressure prediction based on refined velocity analysis will inevitably be the future direction of development. This will involve integrating multi-source velocity fields using deep learning algorithms, pre-stack processing of seismic velocity fields, reservoir AVO attributes, and reservoir-sensitive seismic attributes, employing PCA data dimensionality reduction and compression methods, integrating multi-field velocity resources, and applying machine learning algorithms to optimize velocity models to obtain high-precision pressure prediction fields. Currently, many geophysical companies have released a large amount of commercial pressure prediction software based on velocity field analysis. Most of these software utilize forward modeling guided by classical geophysical theory to obtain different velocity data. However, these methods involve numerous steps and a large workload, making pressure prediction under current seismic interpretation technology conditions only theoretically feasible, with very low technical feasibility.

[0009] Therefore, in order to solve the problem of high-precision reservoir pressure prediction technology based on the fine construction of seismic velocity fields, this application provides a simple and efficient velocity field optimization algorithm. Summary of the Invention

[0010] The problem this invention aims to solve is to provide a reservoir pressure characterization method under small sample velocity conditions. Based on wellbore clay content, porosity, and water saturation data, it performs forward modeling of P-wave and S-wave velocity data in areas with scarce velocity logging, and expands the velocity sampling points. By extracting velocity-sensitive seismic attribute data volumes, it constructs a velocity-sensitive attribute dataset. Based on a deep learning model, it establishes a nonlinear relationship between the wellbore and the attribute dataset. Finally, it applies the constructed high-precision velocity field model to perform high-precision pressure field prediction.

[0011] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is: a method for characterizing reservoir pressure under small sample velocity conditions, comprising the following steps:

[0012] S1: Establish a forward modeling algorithm for wellbore P-wave and S-wave velocity data, and perform quality control by comparing the measured P-wave and S-wave velocities with the forward modeled velocities;

[0013] S2: Perform pore pressure curve calculation, use measured normal pressure and pore pressure measurement point data, optimize Eton equation exponent, perform quality control on predicted velocity data, and correct velocity forward modeling parameters;

[0014] S3: Process the post-stack seismic data, construct seismic attributes sensitive to seismic velocity, select an appropriate number of deep learning hidden structure layers and algorithm factor length, and obtain a high-precision velocity field;

[0015] S4: Introduce a machine learning velocity field construction algorithm factor into the Eton equation to predict pore pressure.

[0016] Furthermore, in S1, based on the evaluation of clay content, porosity, and water saturation in the wellbore, a forward modeling algorithm for P-wave and S-wave velocity data is established. Under conditions of scarce velocity data, forward modeling of velocity curves is performed using well logging physical and electrical parameters to expand the velocity sample data. Based on different lithofacies models, formation fluid parameters, clay velocity parameters, and quartz velocity parameters are set. The forward modeling of P-wave and S-wave curves is performed using the Xu-White model or a self-consistent model, and compared with the measured P-wave and S-wave curves to verify the rationality of the rock forward modeling model parameter settings.

[0017] Furthermore, in S2, the GR curve, mudstone indicator curve, density curve, forward-modeled velocity curve, calculation of wellbore overburden pressure, hydrostatic pressure curve, normal compaction curve, and pore pressure curve are applied using the Eaton equation.

[0018] Furthermore, by integrating forward-modeled P-wave velocity, S-wave velocity, measured static pressure data, and measured pore pressure data, and using an iterative optimization method for wellbore velocity curves for quality control, a joint analysis model for reservoir pressure prediction is established.

[0019] Furthermore, the joint analysis formula for reservoir pressure prediction is as follows:

[0020] d = f(m) + e

[0021]

[0022] Where, d: forward modeling function; Vp: forward modeling P-wave velocity; Vs: normal rock S-wave velocity; ρ: forward modeling density; M: target matrix; ф: porosity; Sc: clay content; SW: water saturation; pp: pore pressure; obp: overburden pressure; hp: hydrostatic pressure; Vforward: forward modeling velocity; Vnct: logging compaction tendency; Exp: Eton index.

[0023] Furthermore, measured P-wave and S-wave velocity data, measured normal pressure, and tested formation pressure were used to calibrate and quality control the forward velocity curves and the static pressure and pore pressure data calculated from the forward velocity curves. Quality control was carried out on the forward velocity curves and pore pressure prediction curves from both velocity and pressure perspectives.

[0024] Furthermore, in S3, a velocity-sensitive attribute dataset is constructed using the P-wave and S-wave velocity reflectivity, fluid factor, post-stack trace integral, and sweet spot attribute obtained from pre-stack AVO inversion, to process the velocity-sensitive attributes of seismic data.

[0025] Furthermore, in S3, the forward velocity curve is used as the learning curve, and the velocity-sensitive attribute is used as the input attribute to establish a nonlinear relationship between velocity data and seismic attributes.

[0026] Furthermore, a deep learning model was trained at the wellbore location, using trace integral data, sweet spot attributes, and pre-stack AVO attributes as input sample data, and the wellbore forward modeling velocity curve as the training objective, to construct the relationship between wellbore velocity data and post-stack seismic attributes. The trained deep learning model was then used to convert trace integrals and sweet spot attributes into a velocity model through a nonlinear correlation method, obtaining a high-precision velocity field, the formula of which is as follows.

[0027]

[0028] Where wi: bias weight; Xi: number of seismic traces of sensitive attribute i; b: bias factor; Vforward: shaft forward modeling speed.

[0029] Furthermore, in S4, the overall process construction equation formula is as follows:

[0030]

[0031] Where Xi: number of seismic traces; Wi: weighting factor; b: deviation factor; pp: pore pressure; obp: buoyancy pressure; hp: hydrostatic pressure; Vnct: compaction tendency; Exp: Eton index.

[0032] The advantages and positive effects of this invention are:

[0033] This invention features a simple process and convenient operation. Addressing scenarios where wellbore velocity data is scarce, it applies rock physics forward modeling to model wellbore velocity data. The rationality of the forward-modeled velocity data is jointly verified using measured pressure data and P-wave and S-wave data. An innovative joint analysis model of wellbore pressure prediction feasibility is created by combining measured pressure data with measured P-wave and S-wave velocity data. Simultaneously, a new method for reservoir pressure characterization is employed, utilizing machine learning algorithms to construct a rapid and high-precision velocity field based on velocity-sensitive seismic attributes such as trace integrals and sweet spot properties. This method, based on a quality control approach using machine learning and velocity forward modeling curves and measured pressure and velocity data, and utilizing post-stack velocity-sensitive seismic attributes to construct a high-precision velocity field, addresses the practical need for high-precision velocity fields in reservoir pressure prediction, significantly improving the accuracy and efficiency of reservoir pressure prediction. Testing in a 3D seismic target area has demonstrated the high efficiency and applicability of this invention for pressure prediction. Attached Figure Description

[0034] Figure 1 This is an overall flowchart of an embodiment of the present invention.

[0035] Figure 2 This is the Pickett interpretation version of the logging saturation in this embodiment of the invention.

[0036] Figure 3 This is a comparison of the forward modeling curves of longitudinal and transverse wave velocities in rock physics according to embodiments of the present invention.

[0037] Figure 4 This is a flowchart of stress prediction based on machine learning in an embodiment of the present invention.

[0038] Figure 5 This is a pore pressure prediction version based on the wellbore, as described in this embodiment of the invention.

[0039] Figure 6 This is a cross-sectional view of pore pressure prediction according to an embodiment of the present invention.

[0040] Figure 7 This is a version of the wellbore fracture pressure prediction based on an embodiment of the present invention.

[0041] Figure 8 This is a cross-sectional view of the predicted rupture pressure according to an embodiment of the present invention.

[0042] Figure 9 This embodiment of the invention establishes a profile based on a machine learning velocity field. Detailed Implementation

[0043] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0044] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0045] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation", "connection" and "linking" should be interpreted broadly, and those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0046] The embodiments of the present invention will be further described below with reference to the accompanying drawings:

[0047] like Figure 1As shown, a method for characterizing reservoir pressure under small sample velocity conditions is characterized by the following steps:

[0048] S1: Establish a forward modeling algorithm for wellbore P-wave and S-wave velocity data, and perform quality control by comparing the measured P-wave and S-wave velocities with the forward modeled velocities. Specifically, the algorithm is first constructed based on wellbore clay content evaluation, porosity evaluation, and water saturation technical evaluation to establish a forward modeling algorithm for P-wave and S-wave velocity data lacking wellbore velocity data. Under the condition of scarce velocity data, forward modeling of velocity curves is performed using well logging physical property parameters and well logging electrical parameters to expand the velocity sample data. According to different lithofacies models, formation fluid parameters, clay velocity parameters, and quartz velocity parameters are set. The Xu-White model or self-consistent model is used to perform forward modeling of P-wave and S-wave curves, and the results are compared with the measured P-wave and S-wave curves to verify the rationality of the rock forward modeling model parameter settings.

[0049] Specifically, during well logging formation evaluation, neutron density cross plots are used to establish the rock physical parameters of dry and wet rock points based on the triangulation method, and the clay content is estimated on this basis; the porosity of the wellbore is evaluated based on the optimization of the three porosity curves of density, neutron and acoustic waves; and the Archie porosity model is established using Picket cross plots.

[0050] like Figure 2 , Figure 3 As shown, forward modeling of P-wave and S-wave velocity curves based on the Xu-White model is performed. Optimized rock physics forward modeling parameters are applied, combined with V_Clay, total porosity, and water saturation parameters, to conduct forward modeling of P-wave and S-wave velocities in wells without measured P-wave and S-wave data. Classical rock geophysical forward modeling methods are used to expand the velocity sample.

[0051] S2: As Figure 4 As shown, the Eton equation based on the wellbore is applied to calculate the pore pressure curve. Measured normal pressure and pore pressure data are used to optimize the Eton equation exponent, further quality control of the predicted velocity data, and correction of the velocity forward modeling parameters. Specifically, the GR curve, mudstone indicator curve, density curve, forward-modeled velocity curve, calculated wellbore overburden pressure, hydrostatic pressure curve, normal compaction curve, and pore pressure curve are calculated using the Eaton equation.

[0052] like Figure 5-8 As shown, measured P-wave and S-wave velocity data, measured normal pressure, and tested formation pressure were used to calibrate and control the forward velocity curves and the static pressure and pore pressure data calculated from the forward velocity curves. From the perspectives of velocity and pressure, the quality control and feasibility analysis of wellbore fracture pressure were carried out from two different dimensions.

[0053] By integrating forward-modeled P-wave velocity, S-wave velocity, measured static pressure data, and measured pore pressure data, accuracy control is achieved during velocity field processing. This allows for more effective dual-parameter control from both the source and application perspectives, optimizing the velocity field. An iterative optimization method for wellbore velocity curves in quality control is employed, establishing a joint analysis model for reservoir pressure prediction. The formula is as follows.

[0054] d = f(m) + e

[0055]

[0056]

[0057] Where, d: forward modeling function; Vp: forward modeling P-wave velocity; Vs: normal rock S-wave velocity; ρ: forward modeling density; M: target matrix; ф: porosity; Sc: clay content; SW: water saturation; pp: pore pressure; obp: overburden pressure; hp: hydrostatic pressure; Vforward: forward modeling velocity; Vnct: logging compaction tendency; Exp: Eton index.

[0058] S3: Process post-stack seismic data to construct seismic attributes sensitive to seismic velocity. Specifically, apply seismic processing velocity fields, reservoir inversion velocity fields, and fuse velocity-sensitive seismic attributes with velocity trend attributes to construct a multi-source velocity-sensitive attribute dataset. Based on this, apply machine learning algorithms, selecting appropriate deep learning hidden structure layers and algorithm factor lengths to obtain a high-precision velocity field. Specifically, using wellbore data as the target curve and multi-source seismic attribute data as input data, perform multi-source velocity field fusion. Obtain a high-precision velocity field, and use the fused velocity field as the primary velocity field to perform reservoir seismic velocity and pressure data conversion.

[0059] A velocity-sensitive attribute dataset was constructed using P-wave and S-wave velocity reflectivity obtained from pre-stack AVO inversion, fluid factor, post-stack trace integral, and sweet spot attribute to process the velocity-sensitive attributes of seismic data.

[0060] Using forward velocity curves as learning curves and velocity-sensitive attributes as input attributes, a nonlinear relationship between velocity data and seismic attributes is established. A trained deep learning model is used to convert trace integrals, sweet spot attributes, and pre-stack AVO attributes into a velocity model through a nonlinear correlation method, thereby obtaining a high-precision velocity field.

[0061] A deep learning model was trained at the wellbore location, using trace integral data, sweet spot attributes, and pre-stack AVO attributes as input sample data, and the wellbore forward modeling velocity curve as the training objective, to construct the relationship between wellbore velocity data and post-stack seismic attributes. The trained deep learning model was then used to convert trace integrals and sweet spot attributes into a velocity model through a nonlinear correlation method, obtaining a high-precision velocity field, such as... Figure 9 As shown. A data compression method based on eigenvalue analysis is applied to compress multi-attribute data. The intrinsic data is constructed as follows:

[0062]

[0063] Where I: eigenma matrix; Xi: number of seismic traces

[0064] S4: A machine learning velocity field construction algorithm factor is introduced into the Eton equation to predict the pore pressure in the target area. The velocity field construction method based on machine learning is as follows. The overall process equation formula is as follows.

[0065]

[0066] Where Xi: number of seismic traces; Wi: weighting factor; b: deviation factor; pp: pore pressure; obp: buoyancy pressure; hp: hydrostatic pressure; Vnct: compaction tendency; Exp: Eton index.

[0067] In summary, this invention addresses the scenario of scarce wellbore velocity data by applying rock physics forward modeling to model wellbore velocity data. The rationality of the forward-modeled velocity data is jointly verified using measured pressure data and P-wave and S-wave data. An innovative joint analysis model of wellbore pressure prediction feasibility is created by combining the forward-modeled velocity data verified by measured pressure data and measured P-wave and S-wave velocity data. Simultaneously, a new method for reservoir pressure characterization is employed, utilizing machine learning algorithms to construct a rapid and high-precision velocity field based on velocity-sensitive seismic attributes such as trace integrals and sweet spot properties. This method, based on a quality control approach using machine learning and velocity forward modeling curves and measured pressure and velocity data, and utilizing post-stack velocity-sensitive seismic attributes to construct a high-precision velocity field, addresses the practical need for a high-precision velocity field in reservoir pressure prediction, significantly improving the accuracy and efficiency of reservoir pressure prediction. Testing in a 3D seismic target area has demonstrated the high efficiency and applicability of this invention for pressure prediction.

[0068] The foregoing has provided a detailed description of one embodiment of the present invention, but this description is merely a preferred embodiment and should not be construed as limiting the scope of the invention. All equivalent variations and modifications made within the scope of the claims of this invention should still fall within the patent coverage of this invention.

Claims

1. A reservoir pressure characterization method for small sample velocity conditions, characterized in that: Includes the following steps, S1: Establish a forward modeling algorithm for wellbore P-wave and S-wave velocity data, and perform quality control by comparing the measured P-wave and S-wave velocities with the forward modeled velocities; S2: Perform pore pressure curve calculation, use measured normal pressure and pore pressure measurement point data, optimize Eton equation exponent, perform quality control on predicted velocity data, and correct velocity forward modeling parameters; S3: Process the post-stack seismic data, construct seismic attributes sensitive to seismic velocity, select the number of deep learning hidden structure layers and algorithm factor length, and obtain a high-precision velocity field; S4: Introduce a machine learning velocity field construction algorithm factor into the Eton equation to predict pore pressure.

2. The method of claim 1, wherein: In S1, based on the evaluation of clay content, porosity, and water saturation in the wellbore, a forward modeling algorithm for P-wave and S-wave velocity data is established. Under the condition of scarce velocity data, forward modeling of velocity curves is performed using well logging physical and electrical parameters to expand the velocity sample data. Based on different lithofacies models, formation fluid parameters, clay velocity parameters, and quartz velocity parameters are set. The forward modeling of P-wave and S-wave curves is performed using the Xu-White model or a self-consistent model, and compared with the measured P-wave and S-wave curves to verify the rationality of the rock forward modeling model parameter settings.

3. A reservoir pressure characterization method under small sample velocity conditions according to claim 1 or 2, characterized in that: In S2, the GR curve, mudstone indicator curve, density curve, forward-modeled velocity curve, overburden pressure, hydrostatic pressure curve, normal compaction curve, and pore pressure curve are calculated using the Eaton equation.

4. The reservoir pressure characterization method under small sample velocity conditions according to claim 3, characterized in that: By integrating forward modeling of P-wave velocity, S-wave velocity, measured static pressure data, and measured pore pressure data, a joint analysis model for reservoir pressure prediction was established using an iterative optimization method for wellbore velocity curves under quality control.

5. The reservoir pressure characterization method under small sample velocity conditions according to claim 4, characterized in that: The joint analysis formula for reservoir pressure prediction is as follows: Where, d: forward modeling function; Vp: forward modeling P-wave velocity; Vs: normal rock S-wave velocity; ρ: forward modeling density; M: target matrix; ф: porosity; Sc: clay content; SW: water saturation; pp: pore pressure; obp: overburden pressure; hp: hydrostatic pressure; Vforward: forward modeling velocity; Vnct: logging compaction tendency; Exp: Eton index.

6. A method for characterizing reservoir pressure under small sample velocity conditions according to claim 1 or 2, characterized in that: The forward velocity curves and the static pressure and pore pressure data calculated from the forward velocity curves were calibrated and quality controlled using measured P-wave and S-wave velocity data, measured normal pressure and tested formation pressure respectively. The forward velocity curves and pore pressure prediction curves were quality controlled from both velocity and pressure aspects.

7. A reservoir pressure characterization method for small sample velocity conditions according to claim 1 or 2, characterized in that: In S3, a velocity-sensitive attribute dataset is constructed using the P-wave and S-wave velocity reflectivity obtained from pre-stack AVO inversion, fluid factors, post-stack trace integrals, and sweet spot attributes to process the velocity-sensitive attributes of seismic data.

8. The reservoir pressure characterization method under small sample velocity conditions according to claim 7, characterized in that: In S3, the forward velocity curve is used as the learning curve, and the velocity-sensitive attribute is used as the input attribute to establish a nonlinear relationship between velocity data and seismic attributes.

9. The reservoir pressure characterization method under small sample velocity conditions according to claim 8, characterized in that: A deep learning model was trained at the wellbore location, using trace integral data, sweet spot attributes, and pre-stack AVO attributes as input sample data, and the forward modeling velocity curve of the wellbore as the training objective, to construct the relationship between wellbore velocity data and post-stack seismic attributes. Using the trained deep learning model, a nonlinear correlation method was employed to convert trace integrals and sweet spot attributes into a velocity model, obtaining a high-precision velocity field. The formula is as follows. V forward = where, wi: bias weight; Xi: the number of seismic traces of sensitive attribute i; b: bias factor; V forward : wellbore forward velocity.

10. A method for characterizing reservoir pressure under small sample velocity conditions according to claim 1 or 2, characterized in that: In S4, the overall process construction equation is as follows: Where Xi: number of seismic traces; Wi: weighting factor; b: deviation factor; pp: pore pressure; obp: buoyancy pressure; hp: hydrostatic pressure; Vnct: compaction tendency; Exp: Eton index.

Images