(1) 3D seismic data processing technology for ultra-low permeability thin interbedded reservoirs.
Three-dimensional seismic data of Zhou 20 1 experimental area 1998 ~ 1999 were collected and processed. The signal-to-noise ratio and resolution of Fuyang oil layer in this study are relatively low, which can not meet the needs of field testing. Therefore, the original data of 100km2 is selected for prestack time migration to improve the prediction effect of structure and reservoir. The processing sampling rate is 2ms, the processing length is 6s, the original warehouse is 25m×50m, and the reprocessing warehouse is 25m×25m.
In view of the characteristics of the original seismic data in this area, such as multi-wave, 50Hz and many outliers, thin interlayer, many faults and complex structure of Fuyang oil layer, in addition to the conventional geometric diffusion compensation, surface consistent amplitude processing and compensation, surface consistent abnormal amplitude (outlier) suppression, surface consistent automatic residual static correction, three-dimensional DMO stacking, zero-phase deconvolution and successive separation of seismic records to improve the resolution, amplitude-preserving frequency-raising processing technology is adopted.
1. A new method of seismic data processing
(1) Pre-stack frequency division denoising can effectively suppress noise and improve data fidelity.
According to the distribution law and area of signal and noise in different frequency bands, using frequency division technology to suppress noise can effectively protect high-frequency weak signals and low-frequency information, suppress abnormal amplitude and improve denoising fidelity. The linear interference in this area mainly appears in the low frequency band below 16Hz; The abnormal noise in the middle and deep layers is mainly distributed in the near offset channel of 15 ~ 40 Hz, which is narrow and quite different from the reflected signal in the same frequency band. Abnormal noise is also distributed in the frequency band above 40Hz, but the energy is weak. Frequency division processing technology is adopted to identify linear interference in low frequency band, accurately detect and suppress interference, and protect high frequency signals from being affected; For low-speed interference in high frequency band, it is only suppressed in high frequency band to protect signals in middle and low frequency bands. Specific methods: first, for single shot records with serious linear interference and wide distribution range, the frequency division detection and frequency division suppression technology of linear interference are adopted to ensure the high fidelity of seismic data; Secondly, for irregular abnormal noise, the methods of frequency division detection and frequency division suppression are adopted to eliminate the noise under the condition of small signal distortion and further improve the quality of data processing; Thirdly, the application of time domain single frequency interference suppression technology can effectively remove 50Hz industrial electrical interference, keep other frequency components from being destroyed, and improve signal fidelity; Fourthly, for seismic records with relatively developed surface waves, adaptive surface wave suppression technology is adopted. This method only suppresses surface waves, is faithful to the low-frequency components of effective signals and other information, and has strong adaptability and stable effect.
(2) High-frequency velocity analysis, that is, high-frequency components of weak energy are superimposed to pick up high-precision velocity estimation values.
In areas with complex fault blocks or thin interbeds, different reflection layers or groups of reflection layers correspond to different velocities, and high-frequency data can obtain more accurate velocity values. In the velocity spectrum of low frequency band and dominant signal-to-noise ratio band, the accuracy and resolution of velocity are not as good as those of high frequency band. High-frequency velocity analysis can pick up weak high-frequency components and stack them well to get high-precision and high-resolution velocity values, and the accuracy of velocity estimation is higher than that of conventional processing methods.
This processing * * * carried out four velocity analysis, and finally chose DMO velocity analysis as the official stacking velocity.
2. Analysis of therapeutic effect
In the whole processing process, according to the characteristics of the original data, a reasonable processing flow is designed, and the processing parameters of each step are analyzed in detail, which greatly improves the quality of the profile. First, from the overall effect of the profile, the signal-to-noise ratio is high and the resolution is moderate; Second, the seismic reflection wave features of Fuyang oil layer are outstanding, which can be traced continuously, with clear and reliable breakpoints and sections, and clear and identifiable reflection structure (Figure 6-2); Thirdly, compared with the original result profile, the frequency band of the processed result profile is broadened, the main frequency is increased by about 15Hz, the resolution is improved, the interlayer information of the target layer is rich, and the low-frequency component of the effective wave is well preserved, which provides a guarantee for the subsequent reservoir prediction.
Figure 6-2 Profile of Final Processing Results
(2) Seismic prediction method for ultra-low permeability thin interbedded reservoirs.
Judging from the current research situation of seismic reservoir prediction, there are two main technical ways to realize reservoir prediction: one is the lateral reservoir prediction technology based on seismic attribute analysis to realize reservoir plane distribution prediction; The second is seismic inversion technology to realize three-dimensional spatial prediction of reservoirs. As far as Fuyang oil layer in the periphery of Daqing is concerned, the thickness of single sandstone in each producing layer is basically below 5m, which is generally 1 ~ 2 m. With the resolution of current seismic data, it is still very difficult to directly interpret a single sand body on the seismic profile, so the lateral prediction of reservoir is still based on the cumulative thickness of sandstone in each oil layer.
1. lateral reservoir prediction by seismic attribute analysis.
Seismic attribute analysis is an important means of lateral reservoir prediction. The purpose of seismic attribute analysis is to extract hidden information from seismic data according to seismic attributes, and transform these information into information related to lithology, physical properties or reservoir parameters, which can directly serve geological interpretation or reservoir engineering, and can qualitatively predict and analyze the distribution characteristics and laws of reservoirs on the plane. It consists of two parts, namely, seismic attribute optimization and prediction. Prediction can be not only the prediction of oil-gas bearing property, lithology or lithofacies, but also the prediction of reservoir parameters. Usually, the seismic attribute values of the whole target interval are used for lateral reservoir prediction, such as main frequency and maximum amplitude. Because this seismic attribute represents the integrity of the reservoir, it is necessary to describe the vertical development series in detail in the oilfield development stage, especially the description of thin interbedded sandstone and mudstone reservoirs, so the results of this lateral reservoir prediction method will inevitably have certain errors at specific points in the plane. Therefore, the analysis of reservoir lateral prediction results should focus on the regularity and trend of reservoir lateral integrity in a certain plane range.
Under the current data and technical conditions (limited data frequency band, insufficient information and defects in the method itself), it is necessary to effectively analyze the utilization of seismic information, establish the statistical relationship between seismic attributes and reservoir characteristics, and screen out effective information suitable for reservoir and oil and gas prediction in the work area, so as to obtain more reliable prediction results.
The main pay zones in Zhou 20 1 test area are F Ⅰ and F Ⅱ reservoirs. Using Geoframe software, 28 kinds of plane seismic attribute parameters are extracted from F ⅰ and F ⅱ reservoirs respectively. By drawing the intersection of seismic attributes and cumulative sandstone thickness, most wells that represent the overall trend of this area and have good correlation with seismic attributes are used to predict the cumulative sandstone thickness. In reservoir prediction, the correlation between the accumulated sandstone thickness of F-I reservoir group and seismic attributes is increased from 265,438+0.65,438+0% to 85.4% after excluding interference wells. It is found that the average positive amplitude and three bandwidth attributes of F ⅰ oil group are highly correlated with sandstone thickness. Based on the average positive amplitude and three bandwidth attributes, the cumulative thickness distribution of sandstone in F ⅰ oil group is predicted by multi-attribute artificial neural network method. There is no correlation between the seismic attributes of F ⅱ oil group and the cumulative thickness of sandstone, but the analysis is related to the undeveloped sandstone of F ⅱ oil group as a whole.
2. Using seismic inversion method to predict the vertical distribution characteristics of main sand bodies.
It is difficult for seismic data to form good response characteristics to thin interbedded reservoirs, so it is necessary to improve vertical resolution through joint inversion of well and earthquake to achieve the purpose of fine description of small layers. In order to better describe the distribution characteristics of sand bodies in Zhou 20 1 well area, Jason software is used for inversion processing. Jason software mainly has three post-stack inversion modules, namely InverTrace, Inver-Mod and StatMod, which correspond to three popular inversion methods: sparse pulse inversion, logging constrained inversion and random inversion. Theoretically, the inversion resolution of the inversion profile obtained by these three inversion methods increases in turn.
Before inversion, 5 1 well involved in inversion was standardized. Combined with the analysis of drilling and logging data in the study area, the reservoir in this area has the characteristics of high wave impedance (low acoustic wave) and low gamma. Using acoustic logging curves to make synthetic records and comprehensively calibrate horizons is the basis for establishing inversion reservoir geological framework model. On the basis of unit test, the inversion flow of three modules, sparse pulse inversion, seismic feature inversion and random inversion, is determined, and the fine description of main target layers is realized.
(1) constrained sparse pulse inversion
Constrained sparse pulse inversion is an inversion method based on fast trend constrained pulse inversion algorithm. This method can be used in areas with few wells or many wells, but only wave impedance inversion can be done. In the reservoir inversion in this area, from the wave impedance profile, the relatively large reservoir development section has obvious wave impedance response display, but the thin reservoir can not be distinguished. This is because the inversion method is mainly based on seismic data, and the acoustic wave and density logging curves only limit the wave impedance trend and range, so the resolution of wave impedance profile depends on the resolution of seismic data. The results are mainly used to determine the general distribution of reservoirs, and on this basis, the interpretation horizons are further encrypted, which improves the calibration accuracy of synthetic records and provides more detailed models and synthetic seismic records for subsequent seismic feature inversion and random inversion.
(2) Seismic feature inversion
Seismic feature inversion technology is a model-based logging parameter inversion technology under seismic data constraints. The core idea is that all kinds of data on seismic traces are interrelated, and any data in the same model layer can be obtained by weighting the data of other traces. Therefore, the interpretation results of seismic data are combined with logging data to generate a fine initial geological model, which makes full use of the information of geology, logging data and seismic data. Principal component analysis and model estimation are carried out on logging data and seismic data. Synthetic record data volume is generated by interpolating and extrapolating synthetic records near the well, and optimized by certain constraints, so that the initial model and seismic data can achieve the best match. When the error between the synthetic recorded data volume and the actual seismic data volume meets the accuracy requirements, the spatial weight distribution is obtained and the weight coefficient volume is formed. The weight coefficient is applied to other types of logging curves to obtain the attribute data volume of the logging curve, such as wave impedance, interval velocity, resistivity, porosity and so on.
This method is suitable for areas with high degree of exploration and development, and a certain number of wells are needed to ensure the quality of inversion results. By this method, a series of attribute data volumes such as wave impedance and resistivity are obtained in this area. From the profile, its resolution is higher than that obtained by sparse pulse inversion.
(3) Stochastic simulation and stochastic inversion
Stochastic simulation and stochastic inversion methods use geostatistics to randomly simulate heterogeneous reservoirs. This technology is also used in areas with high degree of exploration and development and clear understanding of reservoir development characteristics, and for stochastic simulation of reservoir physical parameters (such as porosity and permeability). ) and lithology simulation.
In the stochastic inversion technology, firstly, based on constrained sparse pulse inversion and seismic feature inversion, the approximate range of sand body distribution in X, Y and Z directions is preliminarily determined through full geological analysis of reservoir development characteristics and vertical and horizontal distribution laws. Using inoue wave impedance and inversion wave impedance data volume, histogram analysis and variation analysis are carried out, and then simulation calculation is carried out by combining seismic data volume and wavelet, and the result of simulation calculation is the result of inversion.
The above three inversion methods are applied to the inversion of Zhou 20 1 block respectively, and the three inversion methods are interlocking, and the data obtained in the previous step are applied to the next inversion. Theoretically, the resolution and prediction accuracy should be improved gradually, but from the inversion effect, the resolution obtained by constrained sparse pulse inversion is lower and the results obtained by seismic feature inversion are more reliable. Because of the complexity of reservoirs in this area and the low understanding of reservoir development characteristics in this area, the effects of stochastic simulation and stochastic inversion are not ideal. Therefore, optimizing the inversion results of seismic characteristics will be the main basis for comprehensive reservoir interpretation in the next step.
3. Comprehensive interpretation of sand body
Using the inversion results, the sand body is comprehensively described, and the sandstone thickness and effective thickness in this area are predicted on the wave impedance body and resistivity body inverted by seismic characteristics. The key of sand body description is to determine its boundary, and the scope of sand body is determined by the size of color mark in inversion profile. The main methods to accurately determine the boundary of sand bodies are as follows: firstly, the oil-bearing sand bodies of well 5 1 in the study area are compared and carefully analyzed, and finally a suitable chromatogram is determined to describe the sand bodies in the whole work area; Secondly, for single sand body and target interval, the wave impedance values of sandstone, mudstone and transitional lithology are counted vertically, and then the sand body boundary is determined horizontally according to this relationship and the reflection characteristics of rock strata. Considering that there are some differences in composition and velocity characteristics between different sand bodies in different wells and even sand bodies in different horizons in the same well, the inversion profile shows subtle differences in the color code corresponding to each sand body. Therefore, it is necessary to combine logging data with actual geological conditions to finally complete the specific description and interpretation of sand bodies.
On the inversion profile, the description method is centered on the well point and extends around. According to horizon calibration, the sand bodies are compared on the north-south profile and the east-west profile of the well, and the approximate range of sand bodies is determined. On this basis, the matching between the plane distribution of sand bodies and paleogeomorphic deposits is analyzed, and then it is tracked around the well for more detailed manual interpretation until the distribution range of sand bodies can be clearly determined and the final plane distribution of sandstone can be obtained. On the basis of sand body interpretation, the effective thickness and plane distribution of Zhou 20 1 well area are determined by using resistivity attribute body, and the final effective thickness prediction result is obtained by using known wells for correction, which serves as the basis for well pattern design.
(3) Tracking prediction method while drilling.
In the process of drilling, new drilling results are added to the inversion data volume at any time for inversion tracking prediction, so as to further deepen the understanding of the reservoir, improve the reservoir prediction results and optimize the well location operation.
In September, 2005, the data of 25 new wells were loaded into the inversion data volume, and the rolling prediction of seismic data was carried out under the condition of high well pattern density, which gradually reduced the multiplicity of seismic data, approached the real scale of river sand bodies and improved the accuracy of reservoir prediction. Through the analysis of the development of newly drilled oil layers, combined with the understanding of seismic attributes and a new round of seismic inversion tracking prediction, it is confirmed that the reservoir development in the east of Zhou 20 1 well area is poor. Combined with the research results of reservoir tracking and prediction, the injection-production well pattern is further improved, and the supplementary adjustment scheme of Zhou 20 1 well area is deployed in time, and 27 wells are designed, including 3 horizontal wells.
In the drilling process of Zhou 20 1 well area, under the guidance of the theory of "practice, understanding, re-practice, re-understanding", a total of * * 5 1 well was drilled through iterative inversion, tracking prediction, rolling drilling and other scientific methods, including 48 vertical wells and 3 horizontal wells (Figure 6-3).
Figure 6-3 Bitmap of Completed Drilling in 20 1 Test Area of Sanzhao Depression.