Deep reservoir quality prediction method based on diagenetic facies geochemical coupling

Abstract

The application relates to the technical field of oil and gas exploration and development, and is a deep reservoir quality prediction method based on rock facies-geochemical coupling, which comprises the following steps: obtaining geochemical parameters and physical property parameters of a layer section to be predicted; taking the geochemical parameters and the physical property parameters of the layer section to be predicted as inputs, inputting a reservoir quality coupling prediction model, and outputting a porosity calculation value and a reservoir quality grade prediction result of the layer section to be predicted; and the reservoir quality coupling prediction model comprises a multivariate regression model related to the geochemical parameters, the physical property parameters and the porosity and a reservoir quality grade classification model. The application eliminates the multi-solution property of a traditional reservoir quality prediction method from a geological mechanism by means of strong coupling modeling of rock facies and geochemical parameters, and significantly improves the deep and super-deep reservoir quality prediction accuracy. Compared with the prior art, the application can significantly improve the drilling rate of high-quality reservoirs.

Description

Technical Field

[0001] This invention relates to the field of oil and gas exploration and development technology, and is a method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling. Background Technology

[0002] As my country's oil and gas exploration and development continues to advance into deeper and ultra-deeper reservoirs, the Tarim Basin and Junggar Basin in Xinjiang have become the most important strategic replacement areas for ensuring national energy security. The deep and ultra-deep oil and gas reservoirs developed in these regions are generally characterized by large burial depths, high formation temperatures, high formation pressures, complex lithologies, rapid lateral changes, extremely high heterogeneity, multiple superimposed tectonic movements, and frequent hydrothermal activity, significantly increasing the difficulty of reservoir quality prediction.

[0003] Currently, traditional reservoir quality prediction methods mainly rely on core observation, conventional logging curves, seismic attribute inversion, and geological experience analogy. These methods have certain applicability in shallow and intermediate reservoir exploration, but they have significant shortcomings in deep and ultra-deep conditions: First, seismic data has a weak response to dissolution pores, microfractures, hydrothermal alteration range, and mineral filling degree within deep reservoirs, making it difficult to effectively distinguish between effective reservoirs and tight sections, resulting in a low drilling rate for high-quality reservoirs; Second, traditional research methods based on diagenetic facies are mostly limited to qualitative description and cannot be quantitatively linked to key physical properties such as porosity and permeability, making them difficult to directly apply to exploration deployment; Third, methods that rely solely on geochemical data for reservoir evaluation lack geological genetic constraints, are prone to multiple interpretations, and cannot accurately reflect the true formation and evolution process of reservoirs; Finally, existing technologies fail to integrate diagenetic evolution stages, geochemical tracer indicators, reservoir physical properties, and spatial distribution patterns, making it difficult to form a systematic, quantitative, and scalable deep reservoir quality prediction technology system. Therefore, developing a method for predicting the quality of deep reservoirs that is constrained by geological mechanisms and based on geochemical indicators has become a key technical problem that urgently needs to be solved in the efficient exploration of ultra-deep oil and gas in Xinjiang.

[0004] Patent application CN120703842A discloses a method for predicting high-quality reservoirs in deep-buried matrix-type tight sandstone. The method includes obtaining production data and reservoir physical parameters from the target area; plotting cumulative frequency and probability distribution maps of porosity and permeability based on the production data and reservoir physical parameters; determining the inflection point of porosity, the first inflection point of permeability, and the second inflection point of permeability based on the cumulative frequency and probability distribution maps of porosity and permeability, respectively; selecting several samples from the target area and determining the upper limit point of permeability based on these samples; classifying reservoir types based on the inflection points of porosity, the first inflection point of permeability, the second inflection point of permeability, and the upper limit point of permeability, and identifying high-quality reservoirs. This paper primarily uses porosity and permeability to classify reservoirs in deep-to-ultra-deep tight sandstone.

[0005] Patent application CN116738517A discloses a reservoir prediction method, apparatus, electronic equipment, and medium. The method includes: determining reservoir elastic parameters of a target area based on well logging data and seismic gather data; determining pore space structure parameters of each reservoir in the target area based on the reservoir elastic parameters; and predicting whether each reservoir is a target reservoir based on the pore space structure parameters. This document primarily relies on pore space structure parameters for reservoir evaluation and prediction, without coupling geological and geochemical parameters. Summary of the Invention

[0006] This invention provides a method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling, which can effectively solve the problems of insufficient prediction accuracy and strong ambiguity in deep and ultra-deep reservoir quality prediction technologies.

[0007] One of the technical solutions of this invention is achieved through the following measures: a method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling, comprising: Obtain geochemical and physical parameters of the segment to be predicted; Using the geochemical and physical property parameters of the section to be predicted as input, the reservoir quality coupling prediction model is input, and the calculated porosity value and reservoir quality grade prediction result of the section to be predicted are output. The reservoir quality coupling prediction model includes a multiple regression model relating geochemical parameters, physical property parameters and porosity, and a reservoir quality grade classification model. The multiple regression model relating geochemical parameters, physical property parameters and porosity is a multiple regression model constructed by the diagenetic facies type, geochemical parameters, physical property parameters and porosity of the section. The reservoir quality classification model is obtained by machine learning using several training samples. The training samples include the geochemical and physical property parameters of sections with different reservoir quality grades and the corresponding reservoir quality grades. The section to be predicted is located in the same block as the section.

[0008] The following are further optimizations and / or improvements to one of the above-mentioned technical solutions: Furthermore, the aforementioned geochemical parameters include the content of major elements, trace elements, rare earth elements, isotope content, and elemental characteristic ratios; the physical property parameters include porosity, cement content, and the proportion of dissolution pores.

[0009] Furthermore, the content of the aforementioned major elements includes one or more of silicon dioxide, calcium oxide, magnesium oxide, potassium oxide, and sodium oxide; the content of trace elements includes one or more of Mn, Sr, Fe, Ti, Ba, V, and Ni; the content of rare earth elements includes the sum of all rare earth elements ΣREE, δCe, δEu, and (La / Yb)N; and the content of isotopes includes δ 18 O content, 87 Sr / 86 Sr, Δ 47 The content of one or more of the cluster isotopes; the characteristic element ratios include one or more of the following: iron-titanium ratio, manganese-titanium ratio, and thorium-uranium ratio.

[0010] Furthermore, reservoir quality grades are classified based on diagenetic facies types and the distribution range of their corresponding geochemical and physical property parameters.

[0011] Furthermore, the above-mentioned reservoir quality grades are divided into four categories based on the diagenetic facies type and the corresponding distribution range of geochemical and physical parameters: High-quality reservoir: Type I diagenetic facies: strongly dissolved to weakly cemented facies; geochemical and physical property parameters: Mn content 300 ppm to 800 ppm, Sr content < 150 ppm, δEu 1.05 to 1.30, Fe / Ti ratio < 5. 87 Sr / 86 Sr is 0.7080 to 0.7092, Δ 47 Cluster isotope temperature ranges from 80℃ to 120℃, with porosity >6%; Medium-grade reservoir: Type II diagenetic facies: moderate dissolution-moderate cementation facies; geochemical and physical property parameters range from 150 ppm to 400 ppm Mn, 150 ppm to 350 ppm Sr, δEu from 0.95 to 1.05, and iron-titanium ratio from 5 to 12. 87 Sr / 86 Sr ranges from 0.7088 to 0.7100, δ 18 The oxygen content is -6‰ to 2‰, and the porosity is 2% to 6%. Tight reservoir or non-reservoir: Type III diagenetic facies: weakly dissolved to strongly cemented facies; geochemical and physical property distribution range: CaO content > 50%, ΣREE < 5 ppm, Mn content 400 ppm to 1200 ppm, Sr content 400 ppm to 700 ppm, δEu < 0.95. 87 Sr / 86 Sr is 0.7075 to 0.7085, and porosity is <2%; High-quality hydrothermal reservoir: Class IV diagenetic facies: hydrothermal alteration-semi-filling facies; geochemical and physical property parameters: Mn content > 500 ppm, δ 18 O ranges from -8‰ to -4‰. 87 Sr / 86 Sr>0.710, δEu>1.30, Δ 47 The cluster isotope temperature ranges from 140°C to 180°C, and the porosity ranges from 5% to 12%.

[0012] Furthermore, the multiple regression model relating the aforementioned geochemical parameters, physical property parameters, and porosity was constructed using the following method: In constructing the aforementioned multiple regression model, geochemical parameters include manganese content, europium anomaly, and iron-titanium ratio, while physical property parameters include cement content. Multiple linear regression analysis is performed on the geochemical parameters, physical property parameters, and porosity to construct a multiple regression model relating these parameters to porosity, as shown in the following expression: Φ = a × Mn + b × δEu + c × Fe / Ti – d × CC + e In the formula, Φ is the calculated porosity, %; Mn is the manganese content, %; δEu is the europium anomaly, %; Fe / Ti is the iron-titanium ratio, dimensionless; CC is the cement content, %; and a, b, c, d, and e are the fitting coefficients, dimensionless.

[0013] Furthermore, the above reservoir quality classification model is constructed using the following method: Obtain a sample set, which includes the diagenetic facies type, geochemical parameters, and physical property parameters of several segments with different reservoir quality grades, as well as the corresponding reservoir quality grades; The sample set is divided into a training sample set and a test sample set; A machine learning model was trained using a training sample set to obtain an initial reservoir quality coupling prediction model; The initial reservoir quality coupling prediction model is tested using a test sample set. If the prediction accuracy is met, the initial reservoir quality coupling prediction model is output as the optimal reservoir quality coupling prediction model; otherwise, the machine learning model is retrained using a training sample set.

[0014] This invention eliminates the ambiguity of traditional reservoir quality prediction methods from a geological mechanism perspective by strongly coupling diagenetic facies and geochemical parameters in modeling, significantly improving the accuracy of deep and ultra-deep reservoir quality prediction. Compared with existing technologies, this invention can significantly improve the drilling rate of high-quality reservoirs. Furthermore, this method can achieve high-precision reservoir evaluation under conditions of no coring, minimal coring, and limited data, greatly reducing the exploration cost and drilling risk of ultra-deep wells. It is applicable to deep and ultra-deep complex reservoirs in major oil and gas areas of Xinjiang, possessing strong industrial application value and providing key technical support for the national deep-earth engineering strategy and oil and gas reserve and production increase in Xinjiang. Attached Figure Description

[0015] Appendix Figure 1 This is a flowchart of a method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling; Appendix Figure 2 Diagenetic facies-geochemical coupling calibration map for the 7617-7643 meter interval of Well X; Appendix Figure 3 This is a prediction diagram of the reservoir quality of well X. Appendix Figure 4 This is a map showing the distribution of high-quality reservoir sweet spots.

[0016] Figures 2 to 3 In this context, Mn or manganese content represents the Mn element content; Sr or strontium content represents the Sr element content; Δ 47 Temperature refers to the temperature of cluster isotopes; reservoir I, reservoir II, reservoir III, and reservoir IV correspond to high-quality reservoir, medium-quality reservoir, tight reservoir or non-reservoir, and hydrothermal high-quality reservoir, respectively. Detailed Implementation

[0017] The present invention is not limited to the following embodiments, and the specific implementation can be determined according to the technical solution of the present invention and the actual situation.

[0018] Dessert Area: An area with multiple dessert sections distributed in a continuous pattern on a flat surface, possessing overall development value.

[0019] The method described in this invention is applicable to deep / ultra-deep carbonate reservoirs, clastic reservoirs, interbedded volcanic and sedimentary reservoirs, fracture-vuggy reservoirs, and complex reservoirs subjected to long-term hydrothermal fluid alteration at depths exceeding 5000 meters. It is particularly suitable for ultra-deep oil and gas exploration areas with extremely high heterogeneity, inconspicuous seismic response characteristics, low logging interpretation accuracy, and where traditional exploration methods struggle to effectively identify the location of high-quality reservoirs.

[0020] The machine learning model can be a random forest algorithm. ppm refers to mass percentage, 1 ppm being one part per million; percentages in this invention, unless otherwise specified, refer to mass percentages. Cement refers to chemically precipitated substances other than detrital particles and matrix. The content of cement can be determined using existing known methods, such as thin section identification of rocks and minerals, and instrumental quantitative analysis methods (including X-ray diffraction spectroscopy and thermogravimetric analysis). Thin section identification of rocks and minerals: By statistically analyzing the field of view of rock thin sections under a polarizing microscope, the proportion of cement within the field of view is directly calculated. This is the most commonly used classic method in geology, and it can intuitively distinguish the content of different types of cement. X-ray diffraction spectroscopy: Based on the Bragg diffraction principle, by comparing standard mineral spectra, the mineral phases of cement are quantitatively analyzed, and their proportions are accurately calculated. Thermogravimetric analysis: By analyzing the mass change of samples under programmed temperature control, the content of cements such as carbonates is quantitatively analyzed.

[0021] This invention provides a method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling, comprising: Obtain geochemical and physical parameters of the segment to be predicted; Using the geochemical and physical property parameters of the section to be predicted as input, the reservoir quality coupling prediction model is input, and the calculated porosity value and reservoir quality grade prediction result of the section to be predicted are output. The reservoir quality coupling prediction model includes a multiple regression model relating geochemical parameters, physical property parameters and porosity, and a reservoir quality grade classification model. The multiple regression model relating geochemical parameters, physical property parameters and porosity is a multiple regression model constructed from the geochemical parameters, physical property parameters and porosity of the section. The reservoir quality classification model is obtained by machine learning using several training samples. The training samples include the diagenetic facies type, geochemical parameters and physical property parameters of sections with different reservoir quality grades, as well as the corresponding reservoir quality grades. The section to be predicted is located in the same block as the section.

[0022] The geochemical parameters include the content of major elements, trace elements, rare earth elements, isotopes, and elemental characteristic ratios; the physical property parameters include porosity and cement content.

[0023] The major element content includes one or more of silicon dioxide (SiO2), calcium oxide (CaO), magnesium oxide (MgO), potassium oxide (K2O), and sodium oxide (Na2O); the trace element content includes one or more of Mn, Sr, Fe, Ti, Ba, V, and Ni; the rare earth element content includes the sum of all rare earth elements ΣREE, δCe, δEu, and (La / Yb)N; the isotope content includes δ 18 O content,87 Sr / 86 Sr, Δ 47 The content of one or more of the cluster isotopes; the characteristic element ratios include one or more of the following: iron / titanium ratio (Fe / Ti), manganese / titanium ratio (Mn / Ti), and thorium / uranium ratio (Th / U).

[0024] Reservoir quality grades are determined based on diagenetic facies types and the distribution range of their corresponding geochemical and physical properties.

[0025] For carbonate reservoirs, reservoir quality grades are classified into four categories based on diagenetic facies type and the corresponding geochemical and physical property parameter distribution ranges: High-quality reservoir: Type I diagenetic facies: strongly dissolved to weakly cemented facies; geochemical and physical property parameters: Mn content 300 ppm to 800 ppm, Sr content < 150 ppm, δEu 1.05 to 1.30 (weak to moderately positive anomaly), Fe / Ti ratio < 5. 87 Sr / 86 Sr is 0.7080 to 0.7092, Δ 47 Cluster isotope temperature ranges from 80℃ to 120℃, with porosity >6%; Medium-grade reservoir: Type II diagenetic facies: moderate dissolution-moderate cementation facies; geochemical and physical property parameters range from 150 ppm to 400 ppm, Sr from 150 ppm to 350 ppm, δEu from 0.95 to 1.05 (no obvious anomaly), and Fe / Ti ratio from 5 to 12. 87 Sr / 86 Sr ranges from 0.7088 to 0.7100, δ 18 The oxygen content is -6‰ to 2‰, and the porosity is 2% to 6%. Tight reservoir or non-reservoir: Type III diagenetic facies: weakly dissolved to strongly cemented facies; geochemical and physical property parameters: CaO content > 50%, ΣREE < 5 ppm, Mn content 400 ppm to 1200 ppm, Sr content 400 ppm to 700 ppm, δEu < 0.95 (negative anomaly). 87 Sr / 86 Sr is 0.7075 to 0.7085, and porosity is <2%; High-quality hydrothermal reservoir: Class IV diagenetic facies: hydrothermal alteration-semi-filling facies; geochemical and physical property parameters: Mn content > 500 ppm, δ 18 O ranges from -8‰ to -4‰ (significantly negative). 87 Sr / 86 Sr > 0.710, δEu > 1.30 (strong positive anomaly), Δ47 The cluster isotope temperature ranges from 140°C to 180°C, and the porosity ranges from 5% to 12%.

[0026] The multiple regression model relating the geochemical parameters, physical properties, and porosity was constructed using the following method: In constructing the aforementioned multiple regression model, geochemical parameters include manganese content, europium anomaly, and iron-titanium ratio, while physical property parameters include cement content. Multiple linear regression analysis is performed on the geochemical parameters, physical property parameters, and porosity to construct a multiple regression model relating these parameters to porosity, as shown in the following expression: Φ = a × Mn + b × δEu + c × Fe / Ti – d × CC + e In the formula, Φ is the calculated porosity (%), Mn is the manganese content (%), δEu is the europium anomaly (dimensionless) (%), Fe / Ti is the iron-titanium ratio (dimensionless), CC is the cement content (%), and a, b, c, d, and e are the fitting coefficients (dimensionless). The correlation coefficient Rfit is... 2 >0.95.

[0027] When evaluating the reservoir quality grade of a predicted segment, the porosity calculated using this multiple regression model can be compared with the porosity in the reservoir quality grade for identification, thus eliminating the need for measured porosity. For the segment to be predicted, the porosity calculated using this multiple regression model is first obtained, and then classified, since porosity is involved in the reservoir quality grade. The calculated porosity is then compared with the porosity in the reservoir quality grade, and simultaneously, other input geochemical and physical property parameters are simultaneously benchmarked against the geochemical and physical property parameters in the reservoir quality grade, ultimately yielding the reservoir quality grade result.

[0028] The reservoir quality classification model is constructed using the following method: Obtain a sample set, which includes the diagenetic facies type, geochemical parameters, and physical property parameters of several segments with different reservoir quality grades, as well as the corresponding reservoir quality grades; The sample set is divided into a training sample set and a test sample set; A machine learning model was trained using a training sample set to obtain an initial reservoir quality coupling prediction model; The initial reservoir quality coupling prediction model is tested using a test sample set. If the prediction accuracy is met, the initial reservoir quality coupling prediction model is output as the optimal reservoir quality coupling prediction model; otherwise, the machine learning model is retrained using a training sample set.

[0029] The present invention will be further described below with reference to embodiments: Example 1: As Figure 1As shown, a method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling includes: S01: Basic data collection and standardization.

[0030] Collect and organize all data from the work area to form a unified database. The basic data includes: ① Core data: lithological identification, thin sections of rocks, thin sections of cast bodies, scanning electron microscopy, cathodoluminescence; ②Physical property data: porosity, pore throat structure parameters; ③ Geochemical parameter data: Major element content (SiO2, CaO, MgO, K2O, Na2O), trace element content (Mn, Sr, Fe, Ti, Ba, V, Ni), rare earth element content (ΣREE, δCe, δEu, (La / Yb)N), isotope content (δ 18 O content, 87 Sr / 86 Sr, Δ 47 Cluster isotope content); ④ Well logging data: GR, RT, CNL, DEN, AC, imaging logging; ⑤ Geological background: burial depth, stratum temperature, stratum pressure, tectonic location, sedimentary facies, and fault distribution.

[0031] Standardize and organize physical property data, geochemical parameter data, etc.

[0032] S02: Identification of diagenetic types and division of diagenetic stages.

[0033] After data standardization and processing, all rocks (rock thin sections, thin section castings, scanning electron microscopy (stained), cathodoluminescence, etc.) were observed and described piece by piece to identify and quantify the development intensity and distribution characteristics of various diagenetic processes. The identified diagenetic types include: ① Compaction effect: Determined based on contact relationship (point contact, line contact, concave-convex contact, seam contact); ② Cementing effect: cementation by calcite, dolomite, silica, gypsum, pyrite, chlorite, etc.; ③ Dissolution: particle dissolution, intergranular dissolution, mold holes, dissolution fissures, caverns, etc.; ④ Replacement processes: replacement of dolomite, silicification, sulfides, etc.; ⑤ Hydrothermal processes: saddle-shaped dolomite, hydrothermal quartz, sphalerite, fluorite, and other hydrothermal products; ⑥ Fracture action: manifestations of fracture action such as structural fractures, dissolution fractures, semi-filled fractures, and unfilled fractures.

[0034] Based on the identification of diagenetic types, the diagenetic evolution stages (i.e., diagenetic stages) are determined according to oil and gas reservoir standards, including: early diagenetic stage A, early diagenetic stage B, intermediate diagenetic stage A, intermediate diagenetic stage B, and late diagenetic stage (among which the late diagenetic stage is further distinguished into two types: normal deep burial evolution and hydrothermal fluid superposition and alteration), providing a geological basis for subsequent diagenetic facies classification.

[0035] S03: Quantitative classification of diagenetic facies.

[0036] Based on the analysis of diagenetic types and stages, and according to the principle of "diagenetic combination type + final reservoir quality," the reservoirs are classified into four diagenetic facies and reservoir types: ①Type I: Strong dissolution-weak cementation facies (corresponding to high-quality reservoirs). This type of diagenetic facies is characterized by strong dissolution, weak cementation and compaction, and the development of numerous primary intergranular pores and secondary dissolution pores, with large fracture apertures and low filling degree; ②Type II: Moderate dissolution-medium cementation phase (moderate reservoir). Dissolution and cementation occur simultaneously; reservoir pore spaces are partially preserved and partially filled; physical properties are moderate. ③Class III: Weakly dissolving to strongly cemented phase (poor reservoir / non-reservoir). Cement is abundant, dissolution is weak, primary porosity is basically eliminated, secondary porosity is underdeveloped, and the overall structure is dense. ④Class IV: Hydrothermal alteration-semi-filling phase (high-quality special reservoir). Significantly affected by deep hydrothermal fluids, it develops hydrothermal minerals such as saddle-shaped dolomite and hydrothermal quartz. The dissolution effect is strong, and the fractures and pores are partially filled, but it still retains high reservoir performance.

[0037] The classification of diagenetic facies and reservoir types is a conventional method in this field. It uses a comprehensive approach, employing indicators such as the proportion of dissolution porosity, the volume fraction of cement, the surface density of fractures, the content of hydrothermal minerals, as well as porosity and permeability, to make quantitative determinations, ensuring that the classification of diagenetic facies types is objective and reliable.

[0038] S04: Screening of geochemical indicators strongly coupled with diagenesis.

[0039] To achieve quantitative characterization of diagenetic facies and reservoir quality, indicators most sensitive to changes in diagenetic environment, fluid properties, and diagenetic intensity (i.e., geochemical indicators strongly coupled with diagenesis) were selected from commonly used geochemical parameters to construct a diagenetic facies-geochemical coupled indicator system. The specific indicators in this system are divided into five categories: ①Major element content: SiO2, CaO, MgO, K2O, Na2O, etc.; ② Trace element content: Mn, Sr, Fe, Ti, Ba, V, Ni, etc.; ③ Rare earth element content: ΣREE, δCe, δEu, (La / Yb)N, etc.; ④ Isotope content: δ 18 O、 87 Sr / 86 Sr, Δ 47 wait; ⑤ Elemental characteristic ratios: Fe / Ti, Mn / Ti, Th / U, etc.

[0040] The aforementioned indicators can effectively reflect the temperature, pressure, fluid source, redox conditions, water-rock reaction intensity, and hydrothermal activity during reservoir formation. They have a clear physicochemical correspondence with the diagenetic process and can be used as quantitative input parameters for reservoir quality coupling prediction models. Specific parameter selection can be determined based on the actual reservoir conditions in the study area.

[0041] S05: Diagenetic facies-geochemical fingerprinting.

[0042] After determining the diagenetic facies type and index system, geochemical fingerprinting is performed on each diagenetic facies type to form the distribution range of geochemical parameters and physical property parameters corresponding to different diagenetic facies types, and the following reservoir quality grades are constructed.

[0043] The reservoir quality of the carbonate rocks is classified into four categories based on the diagenetic facies type and the corresponding geochemical and physical property parameters: High-quality reservoir: Type I diagenetic facies: strongly dissoluted to weakly cemented facies. This type of diagenetic facies is characterized by strong dissolution, weak cementation, and well-preserved primary and secondary dissolution porosity. It corresponds to the high-quality reservoir development section. The distribution range of geochemical and physical parameters is as follows: Mn content is 300 ppm to 800 ppm, Sr content is <150 ppm, δEu is 1.05 to 1.30 (weak to moderately positive anomaly), and the iron-titanium ratio (Fe / Ti) is <5. 87 Sr / 86 Sr is 0.7080 to 0.7092, Δ 47 Cluster isotope temperature ranges from 80℃ to 120℃, with porosity >6%; Medium-grade reservoir: Type II diagenetic facies: moderately dissoluting to moderately cementing facies. This type of diagenesis is characterized by simultaneous development of dissolution and cementation, partial retention and filling of pores, and moderate physical properties. Geochemical and physical property parameters range from 150 ppm to 400 ppm, Sr content (strontium content) from 150 ppm to 350 ppm, δEu from 0.95 to 1.05 (no obvious anomaly), and Fe / Ti ratio from 5 to 12. 87 Sr / 86 Sr ranges from 0.7088 to 0.7100, δ18 The oxygen content is -6‰ to 2‰, and the porosity is 2% to 6%. Tight reservoirs or non-reservoirs: Type III diagenetic facies: weakly dissolved to strongly cemented facies. This type of diagenetic facies is characterized by extremely strong cementation, weak dissolution, strong compaction, and basically undeveloped porosity. The distribution range of geochemical and physical parameters is as follows: CaO content > 50%, ΣREE < 5 ppm, Mn content 400 ppm to 1200 ppm, Sr content 400 ppm to 700 ppm, δEu < 0.95 (negative anomaly). 87 Sr / 86 Sr is 0.7075 to 0.7085, and porosity is <2%; High-quality hydrothermal reservoir: Class IV diagenetic facies: hydrothermal alteration-semi-filling facies. This type of diagenesis is strongly influenced by deep hydrothermal fluids, developing hydrothermal minerals such as saddle-shaped dolomite and hydrothermal quartz. While exhibiting strong dissolution, it also shows semi-filling characteristics and still possesses high reservoir capacity. Geochemical and physical property parameters: Mn content > 500 ppm, δ... 18 O ranges from -8‰ to -4‰ (significantly negative). 87 Sr / 86 Sr > 0.710, δEu > 1.30 (strong positive anomaly), Δ 47 The cluster isotope temperature ranges from 140°C to 180°C, and the porosity ranges from 5% to 12%.

[0044] By calibration, each type of diagenetic facies acquires quantifiable, identifiable, and repeatable geochemical characteristics, providing stable geological constraints for subsequent porosity prediction.

[0045] S06: Construction of a prediction model for "diagenetic facies-geochemistry-reservoir physical property coupling".

[0046] After the diagenetic facies-geochemical fingerprinting was completed, a reservoir quality coupling prediction model (i.e., the "diagenetic facies-geochemical-reservoir physical property coupling" prediction model) was constructed by combining multiple linear regression and random forest algorithms. The reservoir quality coupling prediction model includes a multiple regression model relating geochemical parameters, physical property parameters and porosity, as well as a reservoir quality grade classification model. The multiple regression model relating geochemical parameters, physical property parameters and porosity is a multiple regression model constructed from the geochemical parameters, physical property parameters and porosity of the strata.

[0047] (1) The multiple regression model relating the geochemical parameters, physical property parameters and porosity is constructed as follows: In constructing the aforementioned multiple regression model, geochemical parameters include manganese content, europium anomaly, and iron-titanium ratio, while physical property parameters include cement content. Multiple linear regression analysis is performed on the geochemical parameters, physical property parameters, and porosity to construct a multiple regression model relating these parameters to porosity, as shown in the following expression: Φ = a × Mn + b × δEu + c × Fe / Ti – d × CC + e In the formula, Φ is the calculated porosity (%), Mn is the manganese content (%), δEu is the europium anomaly (%), Fe / Ti is the iron-titanium ratio (dimensionless), CC is the cement content (%), and a, b, c, d, and e are the fitting coefficients (dimensionless). The correlation coefficient R of the fitting is... 2 >0.95. When constructing the aforementioned multiple regression model, porosity, manganese content, europium anomaly, iron-titanium ratio, and cement content collected from the work area were used. Manganese content, europium anomaly, iron-titanium ratio, and cement content were used as independent variables, and porosity as the dependent variable for multiple regression. During multiple regression, the correlation coefficient R0 must be satisfied. 2 With a value greater than 0.95, a multiple regression model was obtained, from which the fitting coefficients a, b, c, d, and e were determined. When predicting the reservoir quality grade of the section to be predicted, the manganese content, europium anomaly, iron-titanium ratio, and cement content of the section to be predicted were substituted into the multiple regression model with determined fitting coefficients to obtain the calculated porosity value. Thus, there is no need to use measured porosity, eliminating the need for the step of obtaining porosity measurements.

[0048] (2) The reservoir quality classification model is constructed using the following method: Obtain a sample set, which includes the diagenetic type, geochemical parameters, and physical properties of several layers with different reservoir quality grades, as well as the corresponding reservoir quality grades. The sample set is divided into a training sample set and a test sample set; A machine learning model was trained using a training sample set to obtain an initial reservoir quality coupling prediction model; The initial reservoir quality coupling prediction model is tested using a test sample set. If the prediction accuracy is met, the initial reservoir quality coupling prediction model is output as the optimal reservoir quality coupling prediction model; otherwise, the machine learning model is retrained using a training sample set.

[0049] S07: Continuous prediction accuracy of reservoir quality in single-well profiles.

[0050] Obtain geochemical and physical parameters of the segment to be predicted; Using the geochemical and physical property parameters of the section to be predicted as input, the reservoir quality coupling prediction model is input, and the calculated porosity value and reservoir quality grade prediction result of the section to be predicted are output. The reservoir quality coupling prediction model includes a multiple regression model relating geochemical parameters, physical property parameters and porosity, and a reservoir quality grade classification model. The multiple regression model relating geochemical parameters, physical property parameters and porosity is a multiple regression model constructed by the diagenetic facies type, geochemical parameters, physical property parameters and porosity of the section. The reservoir quality classification model is obtained by machine learning using several training samples. The training samples include the diagenetic type, geochemical parameters and physical property parameters of the section with different reservoir quality grades, as well as the corresponding reservoir quality grades.

[0051] The geochemical parameters include the content of major elements, trace elements, rare earth elements, isotopes, and elemental characteristic ratios; the physical property parameters include porosity and cement content.

[0052] The reservoir quality classification model is applied to a single well profile, and reservoir quality prediction is performed meter by meter according to depth. First, logging data, geochemical parameters, and physical properties (cement content) are matched and categorized by depth. Then, the reservoir quality classification model is sequentially substituted into consecutive depth points, outputting the diagenetic facies type, calculated porosity, and reservoir quality grade for each depth segment. This step allows for the visual identification of high-quality reservoir development sections, tight sections (i.e., tight reservoirs), hydrothermal stimulation sections (i.e., hydrothermal high-quality reservoirs), and fractured sections on the entire well profile, achieving high-precision prediction of reservoir quality in sections with little or no coring, providing a basis for well location evaluation and stratification testing.

[0053] S08: Multi-well comparison and planar reservoir quality prediction and sweet spot delineation.

[0054] After the reservoir quality prediction of a single well profile is completed, further multi-well correlation and high-quality reservoir prediction will be carried out, specifically including: ① By comparing multiple wells, a regional diagenetic facies comparison profile is established to analyze the lateral variation law of diagenetic facies; ② Using structural location, sedimentary facies zone, fault distribution, and paleogeomorphological features as constraints, planar interpolation is performed on diagenetic facies, porosity, and the probability of high-quality reservoirs to generate regional reservoir porosity contour maps, reservoir quality classification maps, and distribution maps of high-quality reservoir sweet spots. ③ The sweet spot area is delineated based on the degree of planar superposition between Class I and Class IV diagenetic facies, providing a direct target for exploration deployment; S09: Actual drilling verification and model iterative optimization.

[0055] The reservoir quality prediction results are compared and verified with core data, test production, and well logging interpretation results from subsequent drilling wells. The high-quality reservoir encounter rate and the error between the measured porosity and the porosity calculated by the multivariate regression model are statistically analyzed. If the error exceeds the allowable range, the process returns to the reservoir quality classification model training stage, supplements sample data, adjusts parameter weights, and retrains the machine learning model until the prediction accuracy meets the exploration and production requirements (e.g., the prediction accuracy can be that the mean absolute percentage error (MAPE) is lower than the mean absolute percentage error threshold (e.g., 5%)), forming a standardized prediction method that can be stably promoted in the work area.

[0056] Example 2: This embodiment takes a certain ultra-deep carbonate reservoir as the specific implementation object. The burial depth in this area is between 7,500 meters and 8,200 meters. The reservoir is extremely heterogeneous and has frequent hydrothermal activity. The drilling rate of high-quality reservoirs using traditional prediction methods is less than 40%.

[0057] First, 46 core samples, 92 thin rock sections, and 46 sets of measured porosity and permeability data from different reservoirs in the area were collected, along with the corresponding major element, trace element, rare earth element, and isotope contents (δ¹²⁺). 18 O、 87 Sr / 86 Sr, Δ 47 Cluster isotopes and elemental characteristic ratios (Fe / Ti, Mn / Ti, Th / U). Through thin section observation, strong compaction, multi-stage calcite cementation, development of hydrothermal dolomite and hydrothermal quartz, and the prevalence of dissolution cavities and tectonic fractures were identified. The overall diagenetic stage was in the late diagenetic stage, accompanied by hydrothermal superposition and modification.

[0058] The reservoir was then classified into four diagenetic facies: strongly dissolved-weakly cemented facies (Type I diagenetic facies, high-quality reservoir), moderately dissolved-moderately cemented facies (Type II diagenetic facies, medium-quality reservoir), weakly dissolved-strongly cemented facies (Type III diagenetic facies, tight reservoir / non-reservoir), and hydrothermal alteration-semi-filling facies (Type IV diagenetic facies, hydrothermal high-quality reservoir). For example, geochemical fingerprinting yielded a field-specific fingerprint: High-quality reservoir (Type I diagenetic facies): manganese content between 420 ppm and 780 ppm, europium anomaly between 1.12 and 1.28, iron-titanium ratio between 3.2 and 4.8, porosity between 6.2% and 12.5%, Sr content less than 50 ppm, and strontium isotope ratio (…). 87 Sr / 86 Sr) ranges from 0.7082 to 0.7085, Δ 47Cluster isotope temperatures range from 156℃ to 178℃; tight reservoirs (Type III diagenetic facies): calcium oxide content ranges from 52% to 56%, ΣREE content ranges from 2.1 ppm to 4.3 ppm, Mn content ranges from 650 ppm to 950 ppm, Sr content ranges from 550 ppm to 670 ppm, and δEu is less than 0.83. 87 Sr / 86 The Sr content is 0.7076 to 0.7077, and the porosity is less than 1.8%.

[0059] Here, following the method shown in Example 1, the distribution ranges of geochemical and physical property parameters for the four reservoir quality grades are determined as follows: The diagenetic facies type is determined through diagenetic observation; based on this, physical property parameters (such as porosity) are directly measured experimentally. The distribution range of these physical property parameters is the distribution range of the actual measured data. Similarly, the distribution range of the geochemical parameters is also the distribution range of the measured data.

[0060] Based on the above calibration results, a reservoir quality classification model specific to the work area was constructed. A multiple regression model relating geochemical parameters, physical properties, and porosity used manganese content, europium anomaly, iron-titanium ratio, and cement content as inputs, and output calculated porosity values. Reservoir prediction was performed on the 7617-7643 meter interval of Well X, and the results showed that this interval belongs to a high-quality reservoir (e.g., ...). Figure 2 The calculated porosity was 8.6%. Actual drilling results showed that the measured porosity of this section was 8.2%, with oil production reaching 86.5 cubic meters per day and natural gas production reaching 126,000 cubic meters per day, and the prediction error was only 4.8%.

[0061] The reservoir quality prediction for well X is as follows: Figure 3 The distribution map of high-quality reservoir sweet spots in this area is as follows: Figure 4 .

[0062] Through its widespread application across the region, this method has increased the drilling rate of high-quality reservoirs in the work area from 42% to 80%, and controlled the average prediction error of porosity calculation values ​​within 3.7%, significantly improving the success rate of ultra-deep exploration.

[0063] The method described in this invention establishes an intrinsic correspondence between diagenesis and geochemical parameters, enabling high-precision quantitative prediction of reservoir quality, thereby improving the drilling rate of high-quality reservoirs and reducing the risks of ultra-deep well exploration and development.

[0064] The above technical features constitute various embodiments of the present invention, which have strong adaptability and implementation effect. Unnecessary technical features can be added or removed according to actual needs to meet the needs of different situations.

Claims

1. A method for predicting the quality of deep reservoirs based on diagenetic facies geochemical coupling, characterized in that, include: Obtain geochemical and physical parameters of the segment to be predicted; Using the geochemical and physical property parameters of the section to be predicted as input, the reservoir quality coupling prediction model is input, and the calculated porosity value and reservoir quality grade prediction result of the section to be predicted are output. The reservoir quality coupling prediction model includes a multiple regression model relating geochemical parameters, physical property parameters and porosity, and a reservoir quality grade classification model. The multiple regression model relating geochemical parameters, physical property parameters and porosity is a multiple regression model constructed from the geochemical parameters, physical property parameters and porosity of the section. The reservoir quality classification model is obtained by machine learning using several training samples. The training samples include the diagenetic facies type, geochemical parameters and physical property parameters of sections with different reservoir quality grades, as well as the corresponding reservoir quality grades. The section to be predicted is located in the same block as the section.

2. The method for predicting the quality of deep reservoirs based on diagenetic facies geochemical coupling according to claim 1, characterized in that, Geochemical parameters include the content of major elements, trace elements, rare earth elements, isotope content, and elemental characteristic ratios; physical property parameters include porosity and cement content.

3. The method for predicting the quality of deep reservoirs based on diagenetic facies geochemical coupling according to claim 2, characterized in that, The major element content includes one or more of silicon dioxide, calcium oxide, magnesium oxide, potassium oxide, and sodium oxide; and / or the trace element content includes one or more of Mn, Sr, Fe, Ti, Ba, V, and Ni; and / or the rare earth element content includes the sum of all rare earth elements ΣREE, δCe, δEu, and (La / Yb)N; and / or the isotope content includes δ 18 O content, 87 Sr / 86 Sr, Δ 47 One or more of the cluster isotopes; or / and, one or more of the elemental characteristic ratios, including the iron-titanium ratio, the manganese-titanium ratio, and the thorium-uranium ratio.

4. The method for predicting the quality of deep reservoirs based on diagenetic facies geochemical coupling according to claim 1, 2, or 3, is characterized in that, Reservoir quality grades are determined based on diagenetic facies types and the distribution range of their corresponding geochemical and physical properties.

5. The method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling according to claim 4, characterized in that, For carbonate reservoirs, reservoir quality grades are classified into four categories based on diagenetic facies type and the corresponding distribution range of geochemical and physical property parameters: High-quality reservoir: Type I diagenetic facies: strongly dissolved to weakly cemented facies; geochemical and physical property parameters: Mn content 300 ppm to 800 ppm, Sr content < 150 ppm, δEu 1.05 to 1.30, Fe / Ti ratio < 5. 87 Sr / 86 Sr ranges from 0.7080 to 0.7092, Δ 47 Cluster isotope temperature ranges from 80℃ to 120℃, with porosity >6%; Medium-grade reservoir: Type II diagenetic facies: moderate dissolution-moderate cementation facies; geochemical and physical property parameters range from 150 ppm to 400 ppm Mn, 150 ppm to 350 ppm Sr, δEu from 0.95 to 1.05, and iron-titanium ratio from 5 to 12. 87 Sr / 86 Sr ranges from 0.7088 to 0.7100, δ 18 The oxygen content is -6‰ to 2‰, and the porosity is 2% to 6%. Tight reservoir or non-reservoir: Type III diagenetic facies: weakly dissolved to strongly cemented facies; geochemical and physical property distribution range: CaO content > 50%, ΣREE < 5 ppm, Mn content 400 ppm to 1200 ppm, Sr content 400 ppm to 700 ppm, δEu < 0.

95. 87 Sr / 86 Sr ranges from 0.7075 to 0.7085, with porosity <2%; High-quality hydrothermal reservoir: Class IV diagenetic facies: hydrothermal alteration-semi-filling facies; geochemical and physical property parameters: Mn content > 500 ppm, δ 18 O ranges from -8‰ to -4‰. 87 Sr / 86 Sr>0.710, δEu>1.30, Δ 47 The cluster isotope temperature ranges from 140°C to 180°C, and the porosity ranges from 5% to 12%.

6. The method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling according to claim 1, 2, 3, or 5, characterized in that, The multiple regression model relating geochemical parameters, physical properties, and porosity was constructed using the following method: In constructing the aforementioned multiple regression model, geochemical parameters include manganese content, europium anomaly, and iron-titanium ratio, while physical property parameters include cement content. Multiple linear regression analysis is performed on the geochemical parameters, physical property parameters, and porosity to construct a multiple regression model relating these parameters to porosity, as shown in the following expression: Φ = a × Mn + b × δEu + c × Fe / Ti – d × CC + e In the formula, Φ is the calculated porosity, %; Mn is the manganese content, %; δEu is the europium anomaly, dimensionless; Fe / Ti is the iron-titanium ratio, dimensionless; CC is the cement content, %; and a, b, c, d, and e are fitting coefficients, dimensionless.

7. The method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling according to claim 4, characterized in that, The multiple regression model relating geochemical parameters, physical properties, and porosity was constructed using the following method: In constructing the aforementioned multiple regression model, geochemical parameters include manganese content, europium anomaly, and iron-titanium ratio, while physical property parameters include cement content. Multiple linear regression analysis is performed on the geochemical parameters, physical property parameters, and porosity to construct a multiple regression model relating these parameters to porosity, as shown in the following expression: Φ = a × Mn + b × δEu + c × Fe / Ti – d × CC + e In the formula, Φ is the calculated porosity, %; Mn is the manganese content, %; δEu is the europium anomaly, dimensionless; Fe / Ti is the iron-titanium ratio, dimensionless; CC is the cement content, %; and a, b, c, d, and e are fitting coefficients, dimensionless.

8. The method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling according to claim 1, 2, 3, 5, or 7, is characterized in that, The reservoir quality classification model is constructed using the following method: Obtain a sample set, which includes the diagenetic facies type, geochemical parameters, and physical property parameters of several segments with different reservoir quality grades, as well as the corresponding reservoir quality grades; The sample set is divided into a training sample set and a test sample set; A machine learning model was trained using a training sample set to obtain an initial reservoir quality coupling prediction model; The initial reservoir quality coupling prediction model is tested using a test sample set. If the prediction accuracy is met, the initial reservoir quality coupling prediction model is output as the optimal reservoir quality coupling prediction model; otherwise, the machine learning model is retrained using a training sample set.

9. The method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling according to claim 4, characterized in that, The reservoir quality classification model is constructed using the following method: Obtain a sample set, which includes the diagenetic facies type, geochemical parameters, and physical property parameters of several segments with different reservoir quality grades, as well as the corresponding reservoir quality grades; The sample set is divided into a training sample set and a test sample set; A machine learning model was trained using a training sample set to obtain an initial reservoir quality coupling prediction model; The initial reservoir quality coupling prediction model is tested using a test sample set. If the prediction accuracy is met, the initial reservoir quality coupling prediction model is output as the optimal reservoir quality coupling prediction model; otherwise, the machine learning model is retrained using a training sample set.

10. The method for predicting the quality of deep reservoirs based on diagenetic geochemical coupling according to claim 6, characterized in that, The reservoir quality classification model is constructed using the following method: Obtain a sample set, which includes the diagenetic facies type, geochemical parameters, and physical property parameters of several segments with different reservoir quality grades, as well as the corresponding reservoir quality grades; The sample set is divided into a training sample set and a test sample set; A machine learning model was trained using a training sample set to obtain an initial reservoir quality coupling prediction model; The initial reservoir quality coupling prediction model is tested using a test sample set. If the prediction accuracy is met, the initial reservoir quality coupling prediction model is output as the optimal reservoir quality coupling prediction model; otherwise, the machine learning model is retrained using a training sample set.

Images