A method for improving inversion result quality by using small work area inversion

Abstract

The present application relates to the technical field of seismic interpretation, and more particularly to a method for improving inversion result quality by using small work area inversion. The method comprises the following steps: determining a target study area and a target well, obtaining a well logging curve of the target well, obtaining a predicted reservoir distribution curve by performing sensitivity analysis on the well logging curve, dividing the target study area into multiple inversion work areas, simulating waveforms indicated by the predicted reservoir distribution curve, establishing a spatial evaluation sample and an initial model according to the simulation result, obtaining a prior probability distribution function and a seismic wave impedance, obtaining a likelihood function in cooperation with the initial model, establishing a posterior probability distribution function according to the prior probability distribution function and the likelihood function, performing random disturbance on model parameters of the posterior probability distribution function, performing distribution sampling on the multiple inversion work areas, and obtaining and recording inversion results. The method provided by the present application effectively improves the inversion result quality by weakening the instability of the corresponding relationship between the well sample reservoir and the seismic waveform characteristics.

Description

Technical Field

[0001] This invention relates to the field of seismic interpretation technology, and in particular to a method for improving the quality of inversion results using small-area inversion. Background Technology

[0002] Reservoir prediction is a crucial component of comprehensive geological evaluation and seismic data interpretation, playing a vital role in oil and gas exploration and development. It is essential for well location assessment, trap evaluation, and reservoir characterization. Seismic inversion is one of the core technologies for reservoir prediction. Based on raw seismic data, this technique integrates seismic and well logging data, matching, combining, and imaging horizontally continuous seismic data with vertically high-resolution well logging data to infer the spatial structure and spatial distribution of subsurface rock strata.

[0003] Currently, seismic waveform indication inversion is commonly used for reservoir prediction. Seismic waveform indication inversion improves inversion resolution, reduces the impact of seismic noise, and utilizes seismic waveforms instead of variograms for statistical analysis. This allows for rapid calculations and multiple simulations, fully leveraging the lateral variations of seismic waveforms to reflect the phase transition characteristics of the reservoir space. Furthermore, it enables the analysis of high-frequency structural characteristics of the vertical lithological assemblages, better reflecting the concept of phase control. The seismic waveform indication inversion method is a true well-seismic combined high-frequency simulation method, transforming inversion results from completely random to progressively deterministic. It also does not have strict requirements on the uniformity of well location distribution, significantly improving the accuracy and applicability of reservoir inversion.

[0004] Currently, seismic inversion technology is playing an increasingly important role in well geological operations. During its continuous iteration and optimization, it has been found that the selection of the number of well samples plays a crucial role in seismic waveform indication inversion, directly affecting the accuracy of reservoir prediction results. Specifically, in actual production, the reservoir prediction results obtained from well samples used in the study area are not absolutely related to the number of samples; in fact, in some blocks, the accuracy of reservoir prediction results decreases as the number of well samples increases. Analysis of seismic waveforms and reservoirs in the work area reveals that the correspondence between reservoirs and seismic waveform characteristics in some well samples is unstable. Therefore, the selection of well samples influences the quality of inversion results to a certain extent. Thus, a method for improving the quality of inversion results using small-area inversion is proposed. Summary of the Invention

[0005] (a) Technical problems to be solved This invention provides a method for improving the quality of inversion results by utilizing small-area inversion, in order to overcome the problems that existing seismic drilling tracking inversion technology cannot fully adapt to the deteriorating reservoir quality of production blocks year by year, the narrow, thin and scattered characteristics of reservoir sand bodies, resulting in low drilling success rate and inability to guarantee production capacity.

[0006] (II) Technical Solution To address the above problems, this invention provides a method for improving the quality of inversion results using small work area inversion, comprising: Step S1: Determine the target study area and identify multiple target wells within the target study area, and obtain well logging data and seismic data from the target wells; Step S2: Obtain the logging curve of the target well based on the logging data and seismic data determined in Step S1, perform sensitivity analysis on the logging curve, and obtain the predicted reservoir distribution curve of the target study area based on the analysis results; Step S3: Obtain the basic data of the target study area determined in step S1, and divide the target study area into multiple inversion work areas based on the basic data of the target study area; Step S4: Use the predicted reservoir distribution curve determined in step S2 to perform waveform indication simulation on the target well determined in step S1, establish a spatial estimation sample based on the waveform indication simulation results, and establish an initial model based on the spatial estimation sample. Step S5: Obtain the prior probability distribution function based on the well logging data determined in Step S1, obtain the seismic impedance based on the seismic data, and obtain the likelihood function through a filter in conjunction with the initial model determined in Step S4. Step S6: Establish a posterior probability distribution function based on the prior probability distribution function and likelihood function obtained in step S5, randomly perturb the model parameters of the posterior probability distribution function, and perform distribution sampling on the multiple inversion work areas determined in step S2 using the MH sampling algorithm to obtain and record the inversion results.

[0007] Preferably, in step S1, the target study area is a production capacity block, and the target well is an inversion well within the production capacity block.

[0008] Preferably, in step S2, the logging curve includes a wave impedance curve and several other lithology sensitivity curves; the predicted reservoir distribution curve is a pseudo-wave impedance curve, which can perform waveform indication inversion for the target study area.

[0009] Preferably, the process of determining the pseudo-wave impedance curve is as follows: the wave impedance curve is used to distinguish sandstone and mudstone in the target study area, a discrimination standard for the wave impedance curve is preset, and if the discrimination of the actual wave impedance curve is lower than the discrimination standard of the wave impedance curve, multiple other lithology sensitivity curves are compared, the curve with the highest lithology sensitivity is selected, and the wave impedance curve and the curve with the highest lithology sensitivity are combined to form the pseudo-wave impedance curve.

[0010] Preferably, the discrimination standard of the wave impedance curve is 90%. If the discrimination of the actual wave impedance curve is less than 90%, the curve with the highest lithological sensitivity is selected. In the fitted pseudo-wave impedance curve, the wave impedance curve is a low-frequency curve, and the curve with the highest lithological sensitivity is a high-frequency curve.

[0011] Preferably, in step S3, the basic data includes the structural distribution characteristics of the target study area, well locations, and platform distribution.

[0012] Preferably, in step S4, the spatial estimation sample is the logging curve of the inverted well in the target well that has a low frequency and a structural similarity greater than 80% through a dual-variable comparison of waveform similarity and spatial distance.

[0013] Preferably, in step S4, the process of establishing the initial model is as follows: by predicting the reservoir distribution curve through singular value decomposition, dynamic clustering analysis of seismic trace waveforms is performed on the target well to establish the mapping relationship between the seismic waveform structure and the well logging curve structure, spatial estimation samples of seismic phases with different waveform structures are generated, and the common structure in the statistical samples is determined as the initial model.

[0014] Preferably, in step S6, the inversion result is the set of the maximum solutions within the solution set range of the corresponding posterior probability distribution function in each inversion work area.

[0015] (III) Beneficial Effects This invention provides a method for improving the quality of inversion results using small-area inversion. The method involves identifying a target study area and multiple target wells within that area, obtaining the logging curves of these target wells, performing sensitivity analysis on the logging curves to obtain the predicted reservoir distribution curves for the target study area, dividing the target study area into multiple independent reservoir prediction inversion work areas, using the predicted reservoir distribution curves to perform waveform indication simulation on the target wells to establish an initial model, obtaining the prior probability distribution function and seismic impedance, obtaining the likelihood function in conjunction with the initial model, establishing the posterior probability distribution function based on the prior probability distribution function and the likelihood function, randomly perturbing the model parameters of the posterior probability distribution function, and performing distribution sampling on multiple inversion work areas to obtain and record the inversion results. This method reduces the instability of the correspondence between well sample reservoirs and seismic waveform characteristics. Compared to conventional large-area inversion, this work area establishment mode effectively improves sample consistency, significantly improves overall inversion efficiency and reservoir prediction accuracy, and achieves better drilling encounter results. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for improving the quality of inversion results using small work area inversion, as described in an embodiment of the present invention. Figure 2This is a schematic diagram of the inversion work area range selection in an embodiment of the present invention, wherein A is a schematic diagram of the first inversion work area range selection, B is a schematic diagram of the second inversion work area range selection, C is a schematic diagram of the third inversion work area range selection, D is a schematic diagram of the fourth inversion work area range selection, E is a schematic diagram of the fifth inversion work area range selection, and F is a schematic diagram of the sixth inversion work area range selection. Figure 3 These are inversion result diagrams of different ranges of inversion work areas in embodiments of the present invention, wherein A is the inversion result diagram of the first inversion work area, B is the inversion result diagram of the second inversion work area, C is the inversion result diagram of the third inversion work area, D is the inversion result diagram of the fourth inversion work area, E is the inversion result diagram of the fifth inversion work area, and F is the inversion result diagram of the sixth inversion work area. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Figure 1 This is a flowchart illustrating a method for improving the quality of inversion results using small work area inversion, as described in an embodiment of the present invention. Figure 1 As shown, this invention provides a method for improving the quality of inversion results using small-area inversion, comprising: Step S1: Determine the target study area and identify multiple target wells within the target study area, and obtain well logging data and seismic data from the target wells; Step S2: Obtain the logging curve of the target well based on the logging data and seismic data determined in Step S1, perform sensitivity analysis on the logging curve, and obtain the predicted reservoir distribution curve of the target study area based on the analysis results; Step S3: Obtain the basic data of the target study area determined in step S1, and divide the target study area into multiple inversion work areas based on the basic data of the target study area; Step S4: Use the predicted reservoir distribution curve determined in step S2 to perform waveform indication simulation on the target well determined in step S1, establish a spatial estimation sample based on the waveform indication simulation results, and establish an initial model based on the spatial estimation sample. Step S5: Obtain the prior probability distribution function based on the well logging data determined in Step S1, obtain the seismic impedance based on the seismic data, and obtain the likelihood function through a filter in conjunction with the initial model determined in Step S4. Step S6: Establish a posterior probability distribution function based on the prior probability distribution function and likelihood function obtained in step S5, randomly perturb the model parameters of the posterior probability distribution function, and perform distribution sampling on the multiple inversion work areas determined in step S2 using the MH sampling algorithm to obtain and record the inversion results.

[0019] In this method, in step S1, the target study area is the production capacity block, and the target well is the inversion well within the production capacity block.

[0020] In practical applications, in step S2, the logging curves include wave impedance curves and several other curves for identifying lithological sensitivity; the predicted reservoir distribution curve is a pseudo-wave impedance curve, which can perform waveform indication inversion for the target study area.

[0021] In this method, a discrimination standard for the wave impedance curve is preset. If the discrimination of the actual wave impedance curve is lower than the discrimination standard, multiple other lithology sensitivity curves are compared, and the curve with the highest lithology sensitivity is selected. The wave impedance curve and the curve with the highest lithology sensitivity are combined to form a pseudo-wave impedance curve.

[0022] In practical applications, the discrimination standard of the wave impedance curve is 90%. If the discrimination of the actual wave impedance curve is less than 90%, the curve with the highest sensitivity to lithology is selected. In the fitted pseudo-wave impedance curve, the wave impedance curve is a low-frequency curve, and the curve with the highest sensitivity to lithology is a high-frequency curve.

[0023] It should be noted that in this method, in step S3, the basic data includes the structural distribution characteristics, well locations, and platform distribution of the target study area.

[0024] In practical applications, in step S4, the spatial estimation sample is the logging curve of the inverted well in the target well that has a low frequency and a structural similarity greater than 80% through the comparison of waveform similarity and spatial distance.

[0025] In this method, in step S4, the initial model is established as follows: by predicting the reservoir distribution curve through singular value decomposition, dynamic clustering analysis of seismic trace waveforms is performed on the target well to establish the mapping relationship between the seismic waveform structure and the well logging curve structure, spatial estimation samples of seismic phases with different waveform structures are generated, and the common structure in the statistical samples is determined as the initial model.

[0026] It should be noted that in practical applications, in step S6, the inversion result is the set of the maximum solutions within the solution set range of the corresponding posterior probability distribution function in each inversion work area.

[0027] This invention provides a method for improving the quality of inversion results using small-area inversion, which can reduce the possibility of unstable correspondence between well sample reservoirs and seismic waveform characteristics, thereby improving the quality of inversion results. Figure 2 and Figure 3 As shown, the working principle of this method of improving the quality of inversion results by using small work areas is described in detail below: Step 1: Determine the target study area and identify multiple target wells within the target study area, and obtain well logging data and seismic data from the target wells; Step 2: Obtain the logging curves of the target well based on the logging and seismic data, perform sensitivity analysis on the logging curves, and obtain the predicted reservoir distribution curves of the target study area based on the analysis results; Step 3: Obtain basic data for the target study area, and divide the target study area into multiple inversion work areas based on the basic data; Step 4: Use the predicted reservoir distribution curve to perform waveform indication simulation on the target well, establish spatial estimation samples based on the waveform indication simulation results, and establish an initial model based on the spatial estimation samples; Step 5: Obtain the prior probability distribution function based on well logging data, obtain the seismic impedance based on seismic data, and obtain the likelihood function through a filter in conjunction with the initial model; Step 6: Establish the posterior probability distribution function based on the prior probability distribution function and the likelihood function. Randomly perturb the model parameters of the posterior probability distribution function, and use the MH sampling algorithm to perform distribution sampling on multiple inversion work areas to obtain and record the inversion results.

[0028] In this embodiment, the target study area is Platform 11 of Block 47, and the target wells are 100 wells within the platform. Waveform indication inversion within different ranges is used to predict the Putaohua oil reservoir. The inversion work area range is increased from small to large. Comparative tests are conducted to analyze the impact of the inversion work area range on the quality of the inversion results.

[0029] In practical applications, sensitivity analysis is performed on the logging curves of the inversion well samples in the target study area. If it is found that the wave impedance curve has poor distinguishability between sandstone and mudstone, a new curve can be obtained by using the characteristic curve reconstruction method with the curve with greater lithological identification sensitivity to predict the reservoir distribution. The wave impedance curve is used as the low-frequency curve, and the curve with greater lithological identification sensitivity is used as the high-frequency curve. The pseudo-wave impedance curve is synthesized in the frequency domain, and the waveform indication inversion is performed using the pseudo-wave impedance curve. The pseudo-wave impedance curve is the predicted reservoir distribution curve.

[0030] In this embodiment, the target study area is divided into multiple inversion work areas according to the structural distribution characteristics and well and platform distribution. Wells with similar low-frequency structures are selected as spatial estimation samples by using waveform similarity and spatial distance as dual variables. The spatial estimation samples enhance the certainty of the low-frequency range of the inversion results, while constraining the range of high-frequency values, making the inversion results more certain.

[0031] In practical applications, based on the predicted reservoir distribution curve, the dynamic clustering analysis of the seismic trace waveforms near the target well is achieved through singular value decomposition. The mapping relationship between the seismic waveform structure and the well logging curve structure is established, and well logging curve sample sets with different types of waveform structures are generated. The common structure in the statistical samples is determined as the initial model. The well logging curve sample set is the spatial estimation sample, and different types of waveform structures represent different types of seismic phases.

[0032] In this embodiment, a priori probability distribution functions are obtained from well logging data based on petrophysical cross-analysis of sandstone and mudstone. Seismic impedance is acquired from seismic data, and matched filtering is performed using an initial sample model to obtain the likelihood function. For example, several wells in the study area may have similar seismic facies characteristics, and are considered to have similar sedimentary environments and sand-mud assemblages. In this case, high-frequency components may differ, but low-frequency components share commonalities, with the common frequency band of the well logging curves exceeding the effective frequency band of the seismic data. Based on this, the determinism of the low-frequency inversion results increases, while also constraining the high-frequency range and reducing the randomness of the inversion.

[0033] In practical applications, the prior probability distribution function and the likelihood function distribution are regarded as the objective function. The posterior probability distribution function is established using the Bayesian theoretical framework. By randomly perturbing the model parameters, the MH sampling algorithm is used to perform distribution sampling, so that the maximum solution of the posterior probability distribution function can be effectively realized. The random solutions that have been effectively realized several times are regarded as the expected value output to obtain the inversion result.

[0034] In practical applications, this embodiment has been used in the drilling tracking inversion process of production capacity blocks such as Changpu 47 and Changpu 434. The target area was divided into 16 independent reservoir prediction inversion work areas according to the structural distribution characteristics and the distribution of well locations and platforms. Compared with conventional large-area inversion, this work area establishment model effectively improved the consistency of samples, and the overall inversion efficiency and reservoir prediction accuracy were greatly improved. Over the years, it has participated in the drilling operation of more than 240 wells and achieved good drilling results.

[0035] Furthermore, in this embodiment, through comparison, the inversion results of the small work area show a natural spatial distribution and high correlation with seismic activity, while the inversion results of the large work area in the waveform stability zone between wells are discontinuous and point-like, with poor correlation with wave group characteristics. The subsequent well sand body prediction accuracy is also slightly lower than that of the small work area. This suggests that the seismic wave group characteristics of different regions will vary significantly, and the correlation between seismic wave group characteristics and well logging curves will also differ considerably. Inversion within the same work area will reduce the correlation between well-seismic samples, affecting the quality of the inversion results.

[0036] This invention provides a method for improving the quality of inversion results using small-area inversion. The method involves identifying a target study area and multiple target wells within that area, obtaining the logging curves of these target wells, performing sensitivity analysis on the logging curves to obtain the predicted reservoir distribution curves for the target study area, dividing the target study area into multiple independent reservoir prediction inversion work areas, using the predicted reservoir distribution curves to perform waveform indication simulation on the target wells to establish an initial model, obtaining the prior probability distribution function and seismic impedance, obtaining the likelihood function in conjunction with the initial model, establishing the posterior probability distribution function based on the prior probability distribution function and the likelihood function, randomly perturbing the model parameters of the posterior probability distribution function, and performing distribution sampling on multiple inversion work areas to obtain and record the inversion results. This method reduces the instability of the correspondence between well sample reservoirs and seismic waveform characteristics. Compared to conventional large-area inversion, this work area establishment mode effectively improves sample consistency, significantly improves overall inversion efficiency and reservoir prediction accuracy, and achieves better drilling encounter results.

[0037] The above embodiments are only used to illustrate the present invention and are not intended to limit the present invention. Those skilled in the art can make various changes and modifications without departing from the spirit and scope of the present invention. Therefore, all equivalent technical solutions also fall within the scope of the present invention, and the patent protection scope of the present invention should be defined by the claims.

Claims

1. A method for improving the quality of inversion results using small work area inversion, characterized in that, include: Step S1: Determine the target study area and identify multiple target wells within the target study area, and obtain well logging data and seismic data from the target wells; Step S2: Obtain the logging curve of the target well based on the logging data and seismic data determined in Step S1, perform sensitivity analysis on the logging curve, and obtain the predicted reservoir distribution curve of the target study area based on the analysis results; Step S3: Obtain the basic data of the target study area determined in step S1, and divide the target study area into multiple inversion work areas based on the basic data of the target study area; Step S4: Use the predicted reservoir distribution curve determined in step S2 to perform waveform indication simulation on the target well determined in step S1, establish a spatial estimation sample based on the waveform indication simulation results, and establish an initial model based on the spatial estimation sample. Step S5: Obtain the prior probability distribution function based on the well logging data determined in Step S1, obtain the seismic impedance based on the seismic data, and obtain the likelihood function through a filter in conjunction with the initial model determined in Step S4. Step S6: Establish a posterior probability distribution function based on the prior probability distribution function and likelihood function obtained in step S5, randomly perturb the model parameters of the posterior probability distribution function, and use the MH sampling algorithm to perform distribution sampling on the multiple inversion work areas determined in step S2, and obtain and record the inversion results.

2. The method for improving the quality of inversion results using small work area inversion according to claim 1, characterized in that, In step S1, the target study area is a production capacity block, and the target well is an inversion well within the production capacity block.

3. The method for improving the quality of inversion results using small work area inversion according to claim 1, characterized in that, In step S2, the logging curves include wave impedance curves and multiple other lithology sensitivity curves; the predicted reservoir distribution curve is a pseudo-wave impedance curve, which can perform waveform indication inversion for the target study area.

4. The method for improving the quality of inversion results using small work area inversion according to claim 3, characterized in that, The process of determining the pseudo-wave impedance curve is as follows: the wave impedance curve is used to distinguish the sandstone and mudstone in the target study area. A discrimination standard for the wave impedance curve is preset. If the discrimination of the actual wave impedance curve is lower than the discrimination standard, multiple other lithological sensitivity curves are compared, and the curve with the highest lithological sensitivity is selected. The wave impedance curve and the curve with the highest lithological sensitivity are combined to form the pseudo-wave impedance curve.

5. The method for improving the quality of inversion results using small work area inversion according to claim 4, characterized in that, The discrimination standard of the wave impedance curve is 90%. If the discrimination of the actual wave impedance curve is less than 90%, the curve with the highest lithological sensitivity is selected. In the fitted pseudo-wave impedance curve, the wave impedance curve is a low-frequency curve, and the curve with the highest lithological sensitivity is a high-frequency curve.

6. The method for improving the quality of inversion results using small work area inversion according to claim 1, characterized in that, In step S3, the basic data includes the structural distribution characteristics, well locations, and platform distribution of the target study area.

7. The method for improving the quality of inversion results using small work area inversion according to claim 1, characterized in that, In step S4, the spatial estimation sample is the logging curve of the inverted well in the target well, which is low-frequency and has a structural similarity greater than 80% through the comparison of waveform similarity and spatial distance.

8. The method for improving the quality of inversion results using small work area inversion according to claim 1, characterized in that, In step S4, the initial model is established as follows: by predicting the reservoir distribution curve through singular value decomposition, dynamic clustering analysis of seismic waveforms is performed on the target well, a mapping relationship between the seismic waveform structure and the well logging curve structure is established, spatial estimation samples of seismic phases with different waveform structures are generated, and the common structure in the statistical samples is determined as the initial model.

9. The method for improving the quality of inversion results using small work area inversion according to claim 1, characterized in that, In step S6, the inversion result is the set of the maximum solutions within the solution set range of the corresponding posterior probability distribution function in each inversion work area.

Images