Reservoir damage prediction method for infill well underbalanced drilling coupled with geomechanical model

Abstract

The present invention discloses a method for predicting reservoir damage in underbalanced drilling of infill wells coupled with a geomechanical model, and relates to the technical field of oil and gas drilling engineering. The present invention comprises the following steps: step (1), collecting well logging, mud recording and seismic data, and establishing a three-dimensional fine geological model in combination with the finite element method; step (2), using the well seismic data in combination with the modeling technology of step (1) to model the original properties of the reservoir, and establish a three-dimensional geomechanical model with reservoir and mechanical properties. The present invention describes in detail the characteristics of energy-depleted formations due to long-term mining by combining the geomechanical model with experimental results specifically conducted on old well areas. This method enables the model to accurately reflect actual geological data, ensuring the high accuracy and reliability of the analysis results.

Description

Technical Field

[0001] The present invention belongs to the field of oil and gas drilling engineering, and in particular relates to a method for predicting reservoir damage in underbalanced drilling of infill wells coupled with a geomechanical model. Background Art

[0002] As oilfields age, many enter the late stages of operation, experiencing subsurface energy depletion and declining production capacity. By intensifying the well pattern and employing volumetric fracturing, the interwell stimulation of the entire reservoir can be effectively increased, potentially unlocking remaining reserves between wells in older well areas. Given the lower formation energy in older well areas, underbalanced drilling (UBD) is a promising option. However, maintaining bottomhole fluid pressure below formation pressure is challenging during UBD operations, leading to transient overbalance. Without filter cake protection around the wellbore, large amounts of drilling fluid filtrate and solids can directly invade the reservoir, potentially causing more severe reservoir damage than overbalanced drilling with conventional water-based drilling fluids. Furthermore, in ultra-low water saturation and hydrophilic formations, the capillary pressure in the reservoir is extremely high, and underpressure alone is insufficient to overcome this pressure, leading to reverse imbibition and potentially damaging the reservoir. Furthermore, physical and chemical reactions between the drilling fluid and the formation generate chemical driving forces that induce water to invade. Liquid intrusion can cause a water lock effect, reduce the permeability of oil and gas reservoirs, cause reservoir damage, and affect subsequent reservoir reconstruction and production capacity assurance.

[0003] When drilling infill wells to open up the reservoir in the old well area, the reservoir cannot adapt to the changes in external conditions, causing the reservoir permeability to decrease, which directly leads to reservoir damage. Taking measures such as predicting reservoir damage in advance as the main method and eliminating it later as the auxiliary method can effectively reduce reservoir damage. The existing methods for protecting underbalanced reservoirs in old well areas are mainly through numerical simulation and experimental testing. However, the numerical simulation method of reservoir damage only considers a limited stratigraphic interval and cannot reflect the full picture of reservoir damage. At the same time, on-site coring is difficult and the data measured in the experiment is limited, resulting in deviations between on-site construction and experiments. In summary, there is currently a lack of an effective prediction method for underbalanced reservoir damage that takes into account geological conditions. Through data collection and numerical model analysis, the underbalanced reservoir damage of infill wells in old well areas can be predicted in advance, which will effectively guide the adjustment of drilling fluid properties, efficient drilling and subsequent production capacity improvement.

[0004] In view of this, the present invention is proposed. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide a method for predicting reservoir damage in underbalanced drilling of infill wells coupled with a geomechanical model, thereby solving the problems raised in the above-mentioned background technology.

[0006] In order to solve the above technical problems, the basic concept of the technical solution adopted by the present invention is:

[0007] A method for predicting reservoir damage in underbalanced drilling of infill wells coupled with a geomechanical model includes the following steps:

[0008] Step (1), collecting well logging, mud logging and seismic data, and establishing a three-dimensional fine geological model by combining the finite element method;

[0009] Step (2), using well seismic data combined with the modeling technology of step (1) to model the original properties of the reservoir, and establish a three-dimensional geomechanical model with reservoir mechanical properties;

[0010] Step (3), collecting cores of the target reservoir, analyzing the potential damage mechanism of the reservoir from macroscopic and microscopic perspectives, and performing sensitivity parameter analysis on the cores through sensitivity experiments;

[0011] Step (4), combining the neural network method to fit the original ecological formation sensitivity attributes, and loading the preliminary well trajectory into the three-dimensional geomechanical model of the reservoir to output the original ecological physical and chemical attribute parameters of the wellbore in the reservoir prediction and evaluation;

[0012] Step (5), establishing the water invasion control equation for underbalanced infill wells, updating the rock attribute parameters under the drilling dynamic disturbance conditions, and outputting the key reservoir parameters after reservoir contamination;

[0013] Step (6) compares and analyzes the degree of reservoir damage before and after reservoir contamination, establishes a dynamic reservoir damage prediction index based on machine learning, and provides targeted guidance on drilling fluid performance control and drilling construction based on the selected main controlling factors.

[0014] Optionally, the modeling method of the geomechanical model includes structural modeling, attribute modeling, rock mechanics modeling and geostress modeling, and step (1) is to establish an accurate block structural model and a small layer development structural framework based on the voxel structural framework modeling technology.

[0015] Optionally, the well seismic data in step (2) include but are not limited to seismic data and drilling cores. Before modeling in step (2), a comprehensive analysis of the well seismic data and the data source is required to determine the spatial distribution of different lithologies and the variation patterns of rock physical parameters.

[0016] Optionally, when obtaining the parameter distribution characteristics of key reservoir properties in the geomechanical model of step (2), data such as well logging, core data, and indoor tests are collected, and the parameter distribution characteristics of key reservoir properties are obtained through numerical simulation methods using lithologic analysis as the constraint results.

[0017] Optionally, the experiments used to analyze the potential damage mechanism of the reservoir from the macroscopic and microscopic perspectives in step (3) include but are not limited to XRD, SEM, and mercury injection experiments, and the potential damage mechanism of the reservoir includes but is not limited to: geological stratification, reservoir rock composition analysis, pore structure characteristics, interface type, permeability value, relationship between porosity and permeability, fluid characteristics of the reservoir, potential damage factors of clay minerals, water lock damage, stress sensitivity and solid phase intrusion damage.

[0018] Optionally, the sensitivity experiments in step (3) include but are not limited to velocity sensitivity, water sensitivity, salt sensitivity, acid sensitivity, alkali sensitivity and stress sensitivity of the reservoir core. The data obtained from the sensitivity experiments are analyzed to derive the main damage factors and damage degree of the reservoir, providing a theoretical basis for reservoir prediction.

[0019] Optionally, in step (4), after the preliminary well trajectory is loaded into the three-dimensional geomechanical model of the reservoir, the spatial distribution and variation laws of geological attributes and physical and chemical attributes are learned and the spatial variability of lithology is simulated by Kriging interpolation to generate a lithology distribution that conforms to the actual situation, and the original physical and chemical attribute parameters of the wellbore in reservoir damage prediction are inverted and supplemented.

[0020] Optionally, the geostress modeling method includes the following steps when constructing a geomechanical model:

[0021] The underground structure and geological structure are imported into the finite element model, and the formation properties are synchronously imported into the finite element calculation platform using the grid interaction algorithm. Finite element numerical calculations are performed using the particle swarm optimization algorithm, and the three-dimensional geostress model is obtained by iterative solution. At the same time, the stress magnitude is constrained by collecting experimental results such as geostress experiments and on-site small pressure analysis, and the stress direction is constrained by paleomagnetic and wave anisotropy results, as well as imaging logging interpretation results.

[0022] Optionally, the water intrusion control equation in step (5) is based on the law of conservation of mass, where the oil and water continuity equations are:

[0023]

[0024] According to Darcy's law, the motion equations of oil and water are obtained:

[0025]

[0026] Where: B o 、B w are the volume coefficients of formation crude oil and formation water, S o 、S w are oil and water saturation, ρ0, ρ w are the densities of oil and water, respectively. is porosity, t is drilling time, K is absolute permeability, μ o、μ w are the viscosities of oil and water, respectively, P o 、P w The pressures K of oil and water are respectively ro , K rw are the relative permeabilities of oil and water, respectively.

[0027] Optionally, in step (6), different comprehensive factors of geological structures and mineral components of different strata are taken into consideration, weights are introduced, and a dynamic reservoir damage prediction index is established based on machine learning.

[0028] After adopting the above technical solution, the present invention has the following beneficial effects compared with the prior art. Of course, any product implementing the present invention does not necessarily need to achieve all of the advantages described below at the same time:

[0029] By combining geomechanical models with experimental results specifically conducted in older well areas, we detailed the characteristics of energy-depleted formations resulting from long-term mining. This approach enables the model to accurately reflect actual geological data, ensuring the high accuracy and reliability of the analysis results.

[0030] 2. This method not only considers the single impact of reservoir damage factors on the formation, but also combines machine learning to consider the correlation between reservoir damage factors, thereby improving the prediction accuracy of the main controlling factors of reservoir damage and breaking the barriers that are difficult to achieve through conventional experiments.

[0031] 3. This method also takes into account the impact of water invasion on reservoir damage under the interaction between formation physical properties and drilling fluid. Conventional simulation method model parameters are difficult to apply to actual formation conditions, and at the same time cannot effectively communicate the reservoir damage influencing factors determined by the experiment.

[0032] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The drawings described below are only some embodiments. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.

[0034] In the picture:

[0035] Figure 1 This is a flow chart of the drilling reservoir damage prediction method.

[0036] Figure 2 Porosity distribution cloud map of the three-dimensional geostress model constructed by the present invention

[0037] Figure 3 The gas saturation distribution cloud map of the three-dimensional ground stress model constructed by the present invention

[0038] Figure 4 Young's modulus distribution cloud map of the three-dimensional geostress model constructed by the present invention

[0039] Figure 5 The brittleness index distribution cloud diagram of the three-dimensional ground stress model constructed by the present invention

[0040] Figure 6 Porosity and permeability distribution map;

[0041] Figure 7 Core damage comparison chart;

[0042] Figure 8 Schematic diagram of crack damage surface morphology.

[0043] It should be noted that these drawings and textual descriptions are not intended to limit the conceptual scope of the present invention in any way, but rather to illustrate the concept of the present invention for those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0044] The present invention will now be described in further detail with reference to the accompanying drawings.

[0045] See also Figure 1-8 As shown, in this embodiment, a method for predicting reservoir damage in underbalanced drilling of infill wells coupled with a geomechanical model is provided, comprising the following steps:

[0046] Step (1) is to establish an accurate block structural model and sub-layer development structural framework based on the comparison of target reservoir drilling data, fault interpretation results and adjacent layer data based on the voxel structural framework modeling technology. The thickness of each sub-layer is calibrated by measuring, logging and seismic data, combined with the stratified data of the pilot well, and the three-dimensional fine geological modeling of the block containing the sub-layer is achieved by correction and verification using inter-well data;

[0047] Step (2) first performs pre-processing operations such as correction and interpolation on the seismic data and drill cores to ensure the consistency and accuracy of the data. The processed seismic data, drill cores and other data sources are comprehensively analyzed to determine the spatial distribution of different lithologies and the variation patterns of rock physical parameters. Combined with the modeling technology of step (1), the seismic data and drill cores are used to model the original properties of the reservoir. A three-dimensional geomechanical model with reservoir and mechanical properties is established;

[0048] The geomechanical model in step (2) includes structural modeling, property modeling, rock mechanics modeling and geostress modeling;

[0049] The geostress modeling method includes the following steps when constructing a geomechanical model:

[0050] When obtaining the parameter distribution characteristics of key reservoir properties in the geomechanical model in step (2), data such as well logging, core data, and indoor tests are collected, and the parameter distribution characteristics of key reservoir properties are obtained through numerical simulation methods using lithologic analysis as a constraint result. The parameter distribution characteristics include but are not limited to polished porosity, permeability, gas saturation, etc.

[0051] Step (3), collecting cores of the target reservoir, analyzing the potential damage mechanism of the reservoir from macroscopic and microscopic perspectives, and performing sensitivity parameter analysis on the cores through sensitivity experiments;

[0052] The experiments used in the macroscopic and microscopic analysis of the potential damage mechanism of the reservoir in step (3) include but are not limited to XRD, SEM, and mercury injection experiments. The potential damage mechanism of the reservoir includes but is not limited to: geological stratification, reservoir rock composition analysis, pore structure characteristics, interface type, permeability value, relationship between porosity and permeability, fluid characteristics of the reservoir, potential damage factors of clay minerals, water lock damage, stress sensitivity, and solid phase intrusion damage;

[0053] The sensitivity tests in step (3) include, but are not limited to, velocity sensitivity, water sensitivity, salt sensitivity, acid sensitivity, alkali sensitivity, and stress sensitivity of the reservoir core. The data obtained from the sensitivity tests are analyzed to determine the main damage factors and damage levels of the reservoir, providing a theoretical basis for reservoir prediction. The sensitivity tests are conducted in accordance with the "SY5358-2010 Reservoir Sensitivity Flow Test Evaluation Method."

[0054] Step (4), combining the neural network method to fit the original ecological formation sensitivity attributes, and loading the preliminary well trajectory into the three-dimensional geomechanical model of the reservoir, outputting the wellbore original ecological physical and chemical attribute parameter R0 in the reservoir prediction and evaluation;

[0055] Combining seismic interpretation results with the finite element method, the underground structure and geological structure are imported into the finite element model. The grid interaction algorithm is used to synchronously import the formation properties into the finite element calculation platform. Finite element numerical calculations are performed using the particle swarm optimization algorithm, and the three-dimensional geostress model is obtained through iterative solution. At the same time, the stress magnitude is constrained by collecting experimental results such as geostress experiments and field small pressure analysis. The stress direction is constrained by paleomagnetic and wave anisotropy results, as well as imaging logging interpretation results.

[0056] When performing finite element model analysis, mesh adjustment and optimization are necessary. The size of the mesh significantly impacts both the results and efficiency of finite element calculations. Adaptive meshing technology based on error control effectively controls mesh size by evaluating the error of each mesh cell and refining or coarsening the mesh in different areas. Furthermore, a triangulation-based mesh generation algorithm generates the mesh by dividing the model into multiple triangles, effectively improving the efficiency and accuracy of simulation calculations.

[0057] In step (4), after the preliminary well trajectory is loaded into the three-dimensional geomechanical model of the reservoir, the spatial distribution and variation laws of geological attributes and physical and chemical attributes are learned and the spatial variability of the lithology is simulated by Kriging interpolation to generate a lithology distribution that conforms to the actual situation, and the original physical and chemical attribute parameters R0 of the wellbore in the reservoir damage prediction are inverted and supplemented. The parameter set R0 mainly includes but is not limited to: porosity, permeability, connectivity, etc.

[0058] To ensure the accuracy and reliability of the inversion results for chemical property parameters, a variety of information, including well logging, core, and seismic data, was first collected, including reservoir transverse and longitudinal wave velocities, density logging data, and seismic shear impedance, Lamé impedance, and shear impedance. Next, this data was cleaned, outliers were processed, and missing values ​​were filled to meet modeling standards. By analyzing the processed data, correlations between reservoir physical properties (such as density, Young's modulus, and Poisson's ratio) and rock mechanical properties were established. A comprehensive three-dimensional rock mechanical model was constructed, incorporating the reservoir model, rock mechanical properties, and seismic response. The modeling process took into account the reservoir's spatial structure, rock mechanical property distribution, and seismic wave propagation characteristics. Using this model, a three-dimensional inversion was performed, accurately calculating key reservoir parameters such as density, Young's modulus, and Poisson's ratio. Finally, the model was environmentally constrained and quality-controlled using field logging data and laboratory rock mechanical test results, ensuring the high accuracy and reliability of the inversion results.

[0059] Through mechanical and water invasion coupled finite element calculations, the rock property parameters under drilling dynamic disturbance conditions are updated, and the key reservoir parameter set R1 after reservoir contamination is output. The parameter set mainly includes but is not limited to: porosity, permeability, connectivity, etc.

[0060] The sets R1 and R0 obtained from the experiments and simulations were compared and analyzed. Porosity, permeability, and connectivity were combined and different weights were set according to actual conditions to establish a comprehensive evaluation index for reservoir damage. The degree of reservoir damage before and after reservoir pollution was compared and analyzed.

[0061] Based on the screened main controlling factors, targeted guidance on drilling fluid performance regulation and drilling construction can effectively reduce reservoir damage in infill wells in old well areas and improve the oil and gas production capacity of infill wells in old well areas.

[0062] Using a neural network algorithm, based on the characteristic types of experimental data, we first calculate the characteristics of the influencing factors according to their correlation, then sort the characteristics of the influencing factors, and finally comprehensively consider the factors affecting reservoir damage, set a threshold to obtain a set of influencing factors with a high degree of correlation, and thus screen out the main controlling factors of reservoir damage under real formation conditions. The steps are as follows:

[0063] Assume that the capacity of the sample data set D is n, the feature dimension is m, and the sample data adopts a ij , i∈1:n, j∈1:m, each data feature uses c1, c2, ..., c k ,…,c m Indicates that c is the value range of the data feature, and its correlation degree I(a ij ,c k ) is calculated as:

[0064]

[0065] Where F is the fitting function of the neural network algorithm; I(a ij ,c k ) is the data feature correlation, I(a ij ,c k ) value is larger, indicating that the correlation between the data type and the feature is higher. The results calculated according to formula (1) are sorted from high to low, and the sub-datasets are further screened. The threshold Q′ is:

[0066] Q′=0.65min(I(a ij ,c k ))+0.65min(I(a ij ,c k )) (8)

[0067] Will I(a ij ,c k ) and Q′, if I(a ij ,c k )<Q′, it means that this feature has a high correlation with the data type, and the data is added to the sub-dataset; if I(a ij ,c k )≥Q′, it means that this feature has a large dispersion with the data type and it is eliminated.

[0068] Step (5), establishing the water invasion control equation for underbalanced infill wells, updating the rock attribute parameters under the drilling dynamic disturbance conditions, and outputting the key reservoir parameters after reservoir contamination;

[0069] During normal overbalanced drilling, the water phase of the drilling fluid intrudes into the formation under the combined effects of positive pressure differential and capillary forces, while the oil phase near the wellbore is expelled to deeper formations away from the wellbore. This is a process known as downstream imbibition. Countercurrent imbibition refers to the phenomenon where the wetting and non-wetting phases flow in opposite directions. During underbalanced drilling, the saturated oil phase in a water-wet formation enters the wellbore under the influence of an underpressure differential, while the water phase of the drilling fluid intrudes into the formation under capillary forces directed toward the interior of the formation. Macroscopically, the non-wetting oil phase and the wetting water phase flow in opposite directions.

[0070] As the water phase of the drilling fluid enters the formation, it occupies the pore space previously occupied by oil, forcing some of the oil from the formation into the wellbore. Simultaneously, because the bottomhole pressure is lower than the formation pressure, the crude oil in the formation also enters the wellbore due to the underpressure differential. Throughout this process, the wetting phase (water phase) enters the formation, while the non-wetting phase (oil phase) enters the wellbore, with the wetting and non-wetting phases flowing in opposite directions. Because the imbibition process occurs spontaneously due to capillary forces, this type of imbibition is called countercurrent imbibition.

[0071] Considering factors such as the small pore throats and strong hydrophilicity of shale reservoirs, which have extremely strong capillary self-imbibition energy; the chemical potential difference between the pore fluid in the shale formation and the drilling fluid; and the semi-permeable membrane properties of mud shale due to its special hydration properties, a water invasion control equation was established and introduced into the geomechanical model. Through the coupling of mechanics and water invasion, the rock property parameters under the dynamic disturbance conditions of drilling were updated, and the key reservoir parameters after reservoir contamination were output.

[0072] Capillary water intrusion control equation

[0073] The water intrusion control equation is based on the law of conservation of mass, where the oil and water continuity equations are:

[0074]

[0075] According to Darcy's law, the motion equations of oil and water are obtained:

[0076]

[0077] Where: B o 、B w are the volume coefficients of formation crude oil and formation water, S o 、S w are oil and water saturation, ρ0, ρ w are the densities of oil and water, respectively. is porosity, t is drilling time, K is absolute permeability, μ o 、μ w are the viscosities of oil and water, respectively, P o 、P w The pressures K of oil and water are respectively ro , K rw are the relative permeabilities of oil and water, respectively.

[0078] When drilling with oil-based drilling fluid, "activity" is used to express the chemical potential between the drilling fluid phase and the fluid in the shale. Maintaining the activity of the oil-based drilling fluid within a certain range prevents water from migrating between the formation and the drilling fluid throughout the drilling process. When water droplets emulsified with oil in the oil-based drilling fluid come into contact with the shale, they form a thin film on the wellbore wall, which can be considered a semipermeable membrane. If the chemical potential of the water phase in the formation fluid is higher than that of the water phase in the drilling fluid, water will migrate from the formation into the drilling fluid; conversely, water will migrate from the aqueous phase of the drilling fluid into the formation.

[0079] Under water-based drilling fluid conditions, shale can have semipermeable membrane properties due to its special hydration properties. However, during the osmotic process, solutes can also pass through the shale semipermeable membrane, so it is not an ideal semipermeable membrane.

[0080] Chemical potential governing equation

[0081]

[0082] Where: D0 is the diffusion rate, I m is the membrane efficiency, c is the solute concentration, μ is the fluid viscosity, v is the dissociation constant, R is the gas constant, and β is the compressibility coefficient of the fluid.

[0083] Step (6) comprehensively considers the potential damage mechanism, main damage factors and damage degree of the reservoir from both macroscopic and microscopic perspectives, compares and analyzes the damage degree before and after reservoir pollution, establishes dynamic reservoir damage prediction indicators based on machine learning, and determines the main controlling factors of reservoir damage under real formation conditions. Based on the selected main controlling factors, targeted guidance is provided for drilling fluid performance regulation and drilling construction, which can effectively reduce reservoir damage in infill wells in old well areas and improve the oil and gas production capacity of infill wells in old well areas. In addition, considering the different comprehensive factors of geological structure and mineral composition of different formations, weights are introduced and dynamic reservoir damage prediction indicators are established based on machine learning.

[0084] The present invention is not limited to the above-described embodiments. Any structural changes made under the guidance of the present invention, which have the same or similar technical solutions as the present invention, should be understood to fall within the scope of protection of the present invention. The technologies, shapes, and structural parts not described in detail in the present invention are all well-known technologies.

Claims

1. A method for predicting reservoir damage in infill well underbalanced drilling coupled with a geomechanical model, characterized in that: The following steps are involved: Step (1), collecting well logging, mud logging and seismic data, and combining them with finite element method to establish a three-dimensional fine geological model; Step (2), using the well seismic data combined with the geological modeling technology of step (1) to model the original properties of the reservoir, and establish a three-dimensional geomechanical model with reservoir mechanical properties; Step (3) is to collect cores from the target reservoir, analyze the potential damage mechanism of the reservoir from macroscopic and microscopic perspectives, and conduct core sensitivity parameter analysis through sensitivity experiments; The experiments used in the macroscopic and microscopic analysis of the potential damage mechanism of the reservoir in step (3) include XRD, SEM, and mercury injection experiments. The potential damage mechanism of the reservoir includes geological stratification, reservoir rock composition analysis, pore structure characteristics, interface type, permeability value, relationship between porosity and permeability, fluid characteristics of the reservoir, potential damage factors of clay minerals, water lock damage, stress sensitivity, and solid phase intrusion damage; Step (4), combining the neural network method to fit the original ecological formation sensitivity attributes, and loading the preliminary well trajectory into the three-dimensional geomechanical model of the reservoir to output the original ecological physical and chemical attribute parameters of the wellbore in the reservoir prediction and evaluation; In step (4), after loading the preliminary well trajectory into the three-dimensional geomechanical model of the reservoir, the spatial distribution and variation of geological attributes and physical and chemical attributes are learned and the spatial variability of lithology is simulated by Kriging interpolation to generate a lithology distribution that conforms to the actual situation, and the original physical and chemical attribute parameters of the wellbore in reservoir damage prediction are inverted and supplemented. , parameter set Including porosity, permeability, and connectivity; To ensure the accuracy and reliability of the inversion results of chemical property parameters, we first collected a variety of information including well logging, core and seismic data, including reservoir transverse and longitudinal wave velocities, density logging data, and seismic shear wave impedance, Lamé impedance and shear impedance. Next, we cleaned these data, processed outliers and filled missing values ​​to meet the modeling standards. By analyzing the processed data, we established the correlation between reservoir physical properties and rock mechanical properties, and constructed a comprehensive three-dimensional rock mechanical model that includes reservoir model, rock mechanical properties and seismic response. During the modeling process, the spatial structure of the reservoir, the distribution of rock mechanical properties and the propagation characteristics of seismic waves were taken into account. Using this model, we performed a three-dimensional inversion and accurately calculated the key parameters of the reservoir density, Young's modulus and Poisson's ratio. Finally, we used field logging data and indoor rock mechanical test results to environmentally constrain and quality control the model, ensuring the high accuracy and reliability of the inversion results. Through mechanical and water invasion coupled finite element calculation, the rock property parameters under drilling dynamic disturbance conditions are updated, and the key reservoir parameter set after reservoir contamination is output. ; Comparative analysis of experimental and simulation sets and collection , combining porosity, permeability, and connectivity and setting different weights according to actual conditions to establish a comprehensive evaluation index for reservoir damage, and compare and analyze the degree of reservoir damage before and after reservoir pollution; Using a neural network algorithm, based on the characteristic types of experimental data, we first calculated the characteristics of the influencing factors according to their correlation, then sorted the characteristics of the influencing factors, and finally comprehensively considered the factors affecting reservoir damage, set a threshold to obtain a set of influencing factors with a high degree of correlation, and thus screened out the main controlling factors of reservoir damage under real formation conditions. The steps are as follows: Assume sample data set Capacity , the feature dimension is , the sample data uses , , , each data feature uses , ,…, ,…, express, is the value range of the data feature, and its correlation The calculation formula is: in, is the fitting function of the neural network algorithm; is the data feature correlation, The larger the value, the higher the correlation between the data type and the feature; Step (5) establishes the water invasion control equation for underbalanced infill wells, updates the rock property parameters under the drilling dynamic disturbance conditions through the coupling of mechanics and water invasion, and outputs the key reservoir parameters after reservoir contamination; Step (6) compares and analyzes the degree of reservoir damage before and after reservoir contamination, establishes a dynamic reservoir damage prediction index based on machine learning, and provides targeted guidance on drilling fluid performance control and drilling construction based on the selected main controlling factors.

2. The reservoir damage prediction method for underbalanced drilling of infill wells coupled with a geomechanical model according to claim 1, characterized in that: The modeling methods of geomechanical models include structural modeling, attribute modeling, rock mechanics modeling and geostress modeling. Step (1) is to establish an accurate block structural model and small layer development structural framework based on the volume element structural framework modeling technology.

3. The reservoir damage prediction method for underbalanced drilling of infill wells coupled with a geomechanical model according to claim 1, characterized in that: The well seismic data in step (2) include seismic data and drilling cores. Before modeling in step (2), a comprehensive analysis should be performed using the well seismic data and the data source to determine the spatial distribution of different lithologies and the variation patterns of rock physical parameters.

4. The method for predicting reservoir damage in infill well underbalanced drilling coupled with a geomechanical model according to claim 1, characterized in that: When obtaining the parameter distribution characteristics of the key reservoir properties in the geomechanical model of step (2), well logging, core, and indoor test data are collected, and the parameter distribution characteristics of the key reservoir properties are obtained through numerical simulation methods using lithologic analysis as the constraint results.

5. The method for predicting reservoir damage in infill well underbalanced drilling coupled with a geomechanical model according to claim 1, characterized in that: The sensitivity experiments in step (3) include velocity sensitivity, water sensitivity, salt sensitivity, acid sensitivity, alkali sensitivity and stress sensitivity of the reservoir core. The data obtained from the sensitivity experiments are analyzed to obtain the main damage factors and damage degree of the reservoir, providing a theoretical basis for reservoir prediction.

6. The method for predicting reservoir damage in infill well underbalanced drilling coupled with a geomechanical model according to claim 1, characterized in that: In step (4), after the preliminary well trajectory is loaded into the three-dimensional geomechanical model of the reservoir, the spatial distribution and variation law of geological attributes and physical and chemical attributes are learned and the spatial variability of lithology is simulated by Kriging interpolation to generate a lithology distribution that conforms to the actual situation, and the original physical and chemical attribute parameters of the wellbore in reservoir damage prediction are inverted and supplemented.

7. The method for predicting reservoir damage in infill well underbalanced drilling coupled with a geomechanical model according to claim 1, characterized in that: The geostress modeling method includes the following steps when constructing a geomechanical model: The underground structure and geological structure are imported into the finite element model, and the formation properties are synchronously imported into the finite element calculation platform using the grid interaction algorithm. Finite element numerical calculations are performed using the particle swarm optimization algorithm, and the three-dimensional geostress model is obtained by iterative solution. At the same time, the stress magnitude is constrained by collecting the results of geostress experiments and on-site small pressure analysis experiments, and the stress direction is constrained by paleomagnetic and wave anisotropy results, as well as imaging logging interpretation results.

8. The method for predicting reservoir damage in infill well underbalanced drilling coupled with a geomechanical model according to claim 1, characterized in that: The water intrusion control equation in step (5) is based on the law of conservation of mass, where the oil and water continuity equations are: ； （1） ； （2） According to Darcy's law, the motion equations of oil and water are obtained: ； （3） ； （4） Where: 、 are the volume coefficients of formation crude oil and formation water, respectively. 、 are oil and water saturation, respectively. 、 are the densities of oil and water, respectively. is the porosity, is the drilling time, is the absolute permeability, 、 are the viscosities of oil and water, 、 The pressure of oil and water respectively 、 are the relative permeabilities of oil and water, respectively.

9. The method for predicting reservoir damage in infill well underbalanced drilling coupled with a geomechanical model according to claim 1, characterized in that: In step (6), the different comprehensive factors of geological structure and mineral composition of different strata are taken into consideration, weights are introduced, and a dynamic reservoir damage prediction index is established based on machine learning.

Images