A logging rock mechanics parameter back-closed loop correction method for well wall stability evaluation

Abstract

A closed-loop correction method for rebuttal verification of wellbore rock mechanics parameters for wellbore stability evaluation is disclosed, relating to the fields of oil and gas drilling engineering, well logging interpretation, wellbore stability evaluation, and rock mechanics parameter interpretation. The method acquires well logging data, formation data, pressure data, drilling fluid density data, wellbore diameter data, strength data, and drilling anomaly data for the target well section, establishing an initial vector; it calculates the theoretical instability margin and the measured wellbore condition. When theoretical instability is determined but the measured wellbore is intact, it calculates the gated inconsistency strength by combining wellbore diameter stability and anomaly exclusion factors, initiating constrained rebuttal verification correction if the condition is met. Through candidate parameter generation, sensitivity ranking, empirical envelope, continuity constraints, and a joint objective function, it outputs the corrected static Young's modulus, Poisson's ratio, shear modulus, bulk modulus, and confidence level indicator; when all candidates are eliminated, the initial parameters are output as a low-confidence reference. The correction results are used for wellbore stability evaluation and drilling fluid density window calculation.

Description

Technical Field

[0001] This invention relates to the fields of oil and gas drilling engineering, well logging interpretation, wellbore stability evaluation, and rock mechanics parameter interpretation, specifically to a closed-loop correction method for well logging rock mechanics parameters used in wellbore stability evaluation. Background Technology

[0002] Rock mechanics parameters are fundamental for wellbore stability analysis, geostress assessment, fracturing design, completion scheme optimization, and reservoir engineering parameter interpretation. In oil and gas exploration and development, parameters such as static Young's modulus, Poisson's ratio, shear modulus, and bulk modulus are typically obtained through laboratory rock mechanics experiments, empirical conversion relationships, or well logging interpretation methods. Due to the limited number of core samples, high testing costs, and discontinuous coverage depth in core experiments, continuous interpretation at each depth using conventional well logging data has become a common technical approach.

[0003] In deep clastic rocks, shale, complex fault blocks, and well sections with significant tectonic stress variations, the uncertainty of rock mechanics parameters directly affects the calculation results of collapse pressure, fracture pressure, and safe drilling fluid density window. When the initial rock mechanics parameters are too low or too high, the designed drilling fluid density window may deviate from the actual formation pressure-bearing capacity, leading to engineering risks such as wellbore collapse, stuck pipe, enlargement, lost circulation, or formation perforation.

[0004] Existing methods for interpreting rock mechanics parameters typically rely on logging parameters such as P-wave transit time, S-wave transit time, density, porosity, and clay content. These parameters are obtained using empirical formulas, statistical regression, nonlinear fitting, or machine learning methods. Other methods combine historical well rock mechanics experimental data or interpretation results from adjacent wells to establish initial rock mechanics parameters for the target well section. While these methods can generate continuous parameter curves, they often lack a reverse verification mechanism to ensure consistency between the interpretation results at local depth points and the measured wellbore response.

[0005] Existing wellbore stability assessment methods typically utilize in-situ stress, pore pressure, drilling fluid column pressure, rock strength parameters, and wellbore geometry to calculate wellbore collapse pressure, fracture pressure, or safe drilling fluid density window. Other methods use parameters such as wellbore enlargement rate, dual-diameter ratio, drilling time, and drilling anomaly records to identify the degree of wellbore collapse or enlargement. These methods mostly use rock mechanics parameters as input or directly use the enlargement rate as an evaluation index for wellbore stability, failing to fully utilize the contradictory phenomenon of theoretical wellbore instability while actual wellbore remains intact to provide a counter-evidence correction for rock mechanics parameters.

[0006] In actual vertical or near-vertical well sections, at certain depths, initial parameter calculations may indicate that the wellbore has reached instability conditions, while caliper logging curves show a small enlargement rate, stable local caliper curves, and no recorded leakage, stuck pipe, or wellbore ellipticity. This situation does not necessarily indicate that all input parameters are biased. Under conditions where pore pressure, rock strength parameters, effective stress coefficient, and regional structural parameters are independently constrained by pressure interpretation, core experiments, or regional calibration data, static Young's modulus and static Poisson's ratio have a more direct impact on the calculation of horizontal principal stress and theoretical wellbore instability margin, and therefore can be prioritized for closed-loop correction.

[0007] Furthermore, simply using the enlargement ratio threshold to determine wellbore condition can be affected by burrs on the caliper curve, logging instrument response, wellbore ellipticity in short sections, drilling anomalies, and local abrupt changes in caliper diameter. Therefore, using an enlargement ratio below a certain threshold as a counter-evidence trigger condition may still carry the risk of false triggering or missed triggering. It is necessary to introduce factors beyond the enlargement ratio, such as local stability of the caliper curve and drilling anomaly exclusion factors, to improve the reliability of measured wellbore condition determination.

[0008] Meanwhile, during the closed-loop correction process, if all candidate results are eliminated after verification by physical constraints, lithological constraints, or continuity constraints, and the methodology does not specify subsequent outputs or identifiers, it can easily lead to interruption of the closed-loop logic. Furthermore, if the reliability threshold, convergence conditions, and maximum number of adjustments are only expressed using preset terms without definite evidence, it will also affect the reliability of the method boundaries and interpretation results. Moreover, conventional wellbore stability assessments typically use instability discrimination, wellbore caliber enlargement assessment, or adjacent well parameter calibration, lacking a technical solution that uses the theoretical wellbore instability margin, wellbore stability, drilling anomaly weighting, candidate parameter empirical envelopes, and formation continuity objective functions as a single closed-loop verification process for coordinated constraints. Summary of the Invention

[0009] The purpose of this invention is to provide a closed-loop correction method for logging rock mechanics parameters for wellbore stability evaluation. This method addresses the problem that initial rock mechanics parameters in complex formations mainly rely on logging responses, core calibration, or interpretation results from adjacent wells, while local depth points lack empirical wellbore state constraints, thus affecting wellbore stability evaluation and drilling fluid density window design.

[0010] The present invention first obtains the initial rock mechanics parameter vector P0 of the depth point to be processed. P0 can be obtained from at least one of the following: continuous interpretation relationship of well logging, core calibration results and interpretation results of adjacent wells. Then, the normalized theoretical wellbore instability margin is calculated at each depth point to be processed, and the measured wellbore condition evaluation results are determined by the wellbore logging curve, drill bit size, local stability of wellbore curve and drilling anomaly record.

[0011] When the normalized theoretical wellbore instability margin reaches the instability trigger condition, the enlargement rate ΔD is less than the wellbore integrity threshold η determined by statistics of historical well sections without abnormalities, and the local stability coefficient of the wellbore diameter curve is greater than the reliability threshold T, the instability trigger condition is met. cal In this invention, the theoretical-measured inconsistency strength is calculated. This inconsistency strength not only characterizes whether there is a contradiction between the theoretical results and the measured wellbore condition, but also gates the contradiction through wellbore reliability and anomaly exclusion factors. When the drilling anomaly record includes lost circulation, stuck pipe, reaming, logging obstruction, wellbore ellipticity, and logging distortion, the anomaly exclusion factor is weighted down, so that the depth point forms an anomaly impact record and a low confidence marker, without directly triggering closed-loop correction, thereby reducing erroneous corrections caused by drilling anomalies.

[0012] For the depth point triggering closed-loop correction, this invention does not directly and arbitrarily change the rock mechanics parameters. Instead, it generates correction candidate results based on candidate parameter generation methods that are not entirely the same as the interpretation model, calibration coefficients, or training sample set of the initial rock mechanics parameter vector P0. The candidate parameter generation methods may include at least one of dynamic and static parameter conversion, lithological stratification calibration, adjacent well parameter migration, or core experiment-constrained regression, and must differ from the acquisition method of P0 in at least one of the parameter sources, conversion relationships, or migration weights. Furthermore, it utilizes E... r ν r Cohesion C, internal friction angle φ, effective stress coefficient α, and pore pressure P p At least two of the parameters are ranked for perturbation sensitivity, and candidate results are determined by coarse and fine adjustments within the experimental experience envelope determined by scatter data of historical core dynamic and static parameters combined with 90% to 99% confidence intervals, thereby reducing the possibility of systematic errors in the initial interpretation model being passed to the closed-loop correction process.

[0013] This invention verifies each candidate adjustment result against physical constraints, lithological stratification constraints, prior range constraints from adjacent wells or core samples, and continuity constraints at adjacent depths. Candidate results that meet these constraints are then re-introduced into the theoretical wellbore instability margin calculation process. The final corrected result is selected through a joint objective function that integrates wellbore mechanical response and formation continuity. The joint objective function includes max(0,F...). r The term F is used to eliminate false alarms of theoretical instability; when F r When the value is less than 0, the objective function still limits the correction range through parameter offset terms, vertical continuity terms, and correction coefficient offset terms, and does not aim to pursue an infinitely large stability margin.

[0014] The beneficial effects of this invention are as follows: First, it combines the normalized theoretical wellbore instability margin, enlargement rate, local stability of the caliper curve, and drilling anomaly exclusion factor into a gated reverse verification trigger condition, enabling the measured wellbore state to reverse-check the logging rock mechanics parameters, reducing the risk of incorrect corrections caused by caliper burrs, logging distortion, or drilling anomalies. Second, through independent candidate parameter generation, sensitivity ranking, experimental experience envelope constraints, and a normalized joint objective function that includes wellbore mechanical response, parameter offset, vertical continuity, and correction coefficient offset, the closed-loop correction simultaneously meets the requirements of mechanical discrimination, geological continuity, and parameter physical rationality. Third, through the selection of the best candidate when there is no convergence, the low-confidence reference output when all candidate results do not meet the constraints, risk warnings, and smooth propagation weight reduction mechanisms, the method still has a complete engineering output chain even in complex well sections or when data is incomplete.

[0015] Compared with existing wellbore stability evaluation methods that directly output the instability state based solely on well logging interpretation parameters and wellbore stress discriminant formulas, this invention does not take F≥0 as the final risk conclusion, but further introduces ΔD and R... cal The gated counter-evidence trigger condition is formed with the anomaly exclusion factor A. This gated logic can separate the contradictory points that are theoretically unstable but are actually intact from the general instability judgment results, avoiding misjudgments caused by simply relying on the theoretical wellbore instability margin or the enlargement rate threshold.

[0016] Compared with manual parameter tuning, adjacent well migration, or single-objective minimization of F r Compared to the parameter calibration method, this invention uses experimental experience envelopes, adjacent depth continuity constraints, and includes max(0,F) parameters. r The normalized joint objective function for parameter offset, depth change rate, and correction coefficient offset ensures that the correction results are simultaneously constrained by mechanical response, geological continuity, and parameter physical rationality. The ablation comparison in the implementation results shows that without the normalized joint objective function and fallback output, there are still many parameter mutations or process interruptions. However, after adopting the complete process of this invention, the number of mutation points is reduced to one and there are no process interruptions, demonstrating the synergistic effect between gated triggering, constrained correction, and low-confidence output.

[0017] Compared with existing technical solutions that directly output the instability state using only the wellbore stress discriminant formula, or rely on manual experience to adjust parameters when theoretical calculations and measured results are inconsistent, the distinguishing feature of this invention is that it uses R... calThe A, I, experience envelope, normalized four-component J, and low-confidence fallback output are set as necessary collaborative links in the continuous closed loop. This combination is not a simple parallel arrangement of existing steps, but rather a progressive processing chain addressing three technical issues: false triggering, non-physical parameter mutations, and process interruption caused by the complete elimination of candidates. As shown in Table 4, after adopting the complete closed-loop process, the false triggering rate decreased from 18.3% to 0, the number of parameter mutation points decreased from 27 to 1, and risk warning output and process continuity were maintained even in the case of complete candidate elimination. Attached Figure Description

[0018] Figure 1 Flowchart of the closed-loop correction method for reversal verification of well logging rock mechanical parameters; Figure 2 Comparison of evidence of wellbore instability in the target section and the results of rock mechanics parameter correction. Detailed Implementation

[0019] The present invention will be further described below with reference to specific embodiments. These embodiments are used to illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention.

[0020] like Figure 1 As shown, the processing chain of this invention starts with engineering data input, takes the inconsistency between theoretical wellbore instability judgment and actual wellbore state as the counter-evidence trigger condition, takes constrained candidate correction and re-substitution verification as the core, and takes credibility identifier, safe drilling fluid density window and wellbore stability evaluation profile as engineering output.

[0021] This embodiment selects a 500m long well section (3250m to 3750m) in a vertical well in the Sichuan Basin as the target well section. The logging curves for this section were processed with a sampling interval of 0.125m. The data used include density logging data, P-wave transit-time logging data, S-wave transit-time logging data, natural gamma ray logging data, caliper logging data, stratigraphic information, overburden pressure data, pore pressure data, drilling fluid density data, drill bit size data, drilling anomaly records, and pre-calibrated regional tectonic strain parameters ε for the target block. h and ε H Effective stress coefficient α, historical well core strength experimental data, and rock mechanics experimental data.

[0022] In this embodiment, the core strength test data are discrete depth point data, and the drilling anomaly record is well section interval data. Both are repositioned and mapped using the logging depth as a unified benchmark.

[0023] S1, Data Acquisition and Depth Relocation Acquire historical well logging data, formation information, and depth-corresponding rock mechanics experimental data to establish a calibration sample set. Historical well logging data includes density logging data, P-wave transit time logging data, S-wave transit time logging data, natural gamma logging data, and caliper logging data.

[0024] In this embodiment, the core strength test data are discrete depth point data, and the drilling anomaly records are well interval data. During processing, the logging depth is used as a unified depth benchmark to reposition the core strength test data based on logging depth. Simultaneously, the well intervals in the drilling anomaly records are mapped to the corresponding depth points to be processed according to the top and bottom depths of the well intervals, and anomaly record markers are assigned to the corresponding depth points. For long-span drilling anomaly records with a continuous span greater than 20m, they are first segmented according to stratigraphic interfaces and wellbore quality mutation points. If the segmentation still exceeds 50m, anomaly impact markers are assigned to segments from 10m to 50m, and wellbore stability and anomaly exclusion factors are calculated within each sub-window. The wellbore quality mutation points can be identified using a sliding window, with a window length of 1.0m to 2.0m or 5 to 10 sampling points. When the difference in the median wellbore diameter between adjacent windows is greater than 0.5η·BS, or the rate of change of the wellbore standard deviation is greater than 50%, the window boundary depth is taken as the wellbore quality mutation point. When the regional tectonic strain parameter ε... h ε H When the effective stress coefficient α or rock strength experimental data is missing, the calibration data of adjacent wells in the same layer should be used to supplement it first; when the data of adjacent wells is insufficient, the statistical relationship of the same type of lithology or the empirical relationship of the target block should be used to supplement it; if multiple key data are missing at the same time, the empirical threshold of the same type of lithology or the empirical relationship of the target block should be used to supplement it, and the low confidence mark of missing data at the corresponding depth point should be output.

[0025] Acquire overburden pressure data, pore pressure data, drilling fluid density data, drill bit size data, and regional tectonic strain parameter ε at the corresponding depth of the target well section. h and ε H Effective stress coefficient α, historical well core strength test data, target block strength parameter calibration relationship, and drilling anomaly records.

[0026] Drilling fluid density data is used to calculate drilling fluid column pressure; drill bit size data is used to determine wellbore geometric boundary conditions and is used together with the caliper logging curve to calculate the enlargement rate ΔD; core strength test data is assigned to adjacent logging depth sampling points according to the core sampling depth; drilling anomaly records are mapped to corresponding logging depth sampling points according to the top and bottom depths of the recorded well section, forming depth-by-depth anomaly markers.

[0027] S2, Initial Rock Mechanics Parameter Vector Acquisition The calibration sample set was subjected to depth matching, outlier identification, and standardization. Depth matching ensured consistency between the rock mechanics experimental data and the logging curve data at the corresponding depth. In this embodiment, the interquartile range method was used to identify isolated outliers in the logging curves, and the quantile truncation method was used to process the outliers; subsequently, the density, P-wave transit time, S-wave transit time, and natural gamma logging curves were standardized.

[0028] After data processing is completed, this embodiment establishes a hierarchical multivariate nonlinear regression relationship according to the stratigraphic level. Density logging, P-wave transit time, S-wave transit time and natural gamma logging data are used as inputs to obtain the initial static Young's modulus E0 and the initial Poisson's ratio ν0, respectively.

[0029] The target well interpretation dataset was processed using the same depth matching, outlier handling, and standardization methods as the calibration sample set. The processed target well interpretation dataset was then substituted into a hierarchical multivariate nonlinear regression relationship to obtain the initial rock mechanics parameter vector P0, P0={E0, ν0} for the target well section at each depth.

[0030] Subsequently, the initial shear modulus G0 and the initial bulk modulus K0 are calculated based on E0 and ν0: ; .

[0031] S3, Normalized Theoretical Wellbore Instability Margin Calculation A depth point within the target well section is selected as the depth point to be processed. Based on the density data of the shallow overlying strata and the average overlying pressure gradient of the target well, the overlying pressure contribution above the top boundary depth of the target well section is determined. Then, by integrating the density logging data along the depth within the target well section, the overlying pressure σ at the depth point to be processed is calculated. v .

[0032] Overhead pressure is calculated using the following formula: Where z is the depth of the point to be processed, z0 is the top boundary depth of the target well section, and σ v (z0) is the overburden pressure reference value, ρ(h) is the formation volume density at depth h, and g is the gravitational acceleration.

[0033] This embodiment uses the equivalent depth method to obtain the pore pressure P at the depth point to be processed. p To obtain the overlying pressure σ v pore pressure P p Initial static Young's modulus E0, initial Poisson's ratio ν0, effective stress coefficient α, and regional tectonic strain parameter ε h and ε H Then, the minimum horizontal principal stress σ in the far field at the depth point to be processed is calculated. h and the maximum horizontal principal stress σ in the far fieldH : ; ; The drilling fluid column pressure is calculated based on the drilling fluid density data at the depth corresponding to the depth point to be processed, and the wellbore geometric boundary conditions are determined based on the drill bit size data. Under vertical well conditions, the well axis direction is parallel to the direction of vertical geostress. Based on the far-field horizontal principal stress, drilling fluid column pressure, and wellbore geometric boundary conditions, the Kirsch wellbore stress model is used to calculate the radial total principal stress, circumferential total principal stress, and axial total principal stress at the wellbore.

[0034] After obtaining the total principal stresses in three directions at the wellbore wall, and subtracting the influence of pore pressure according to the effective stress principle, the three effective principal stresses at the wellbore wall are obtained: Where σ is the total principal stress at the wellbore, σ′ is the corresponding effective principal stress, α is the effective stress coefficient, and P p The pore pressure is then used. Subsequently, the three effective principal stresses at the wellbore are sorted by their numerical values, with the maximum value taken as the maximum effective principal stress σ1 and the minimum value taken as the minimum effective principal stress σ3.

[0035] This embodiment obtains rock strength parameters based on logging curve data at the depth point to be processed, historical well core strength experimental data, and the calibration relationship of strength parameters in the target block. Specifically, the P-wave velocity V is calculated from the P-wave transit time logging data. p The clay content V was calculated from natural gamma logging data. sh The cohesive force C and the internal friction angle φ are obtained by combining the experimental calibration relationship of the target block.

[0036] After obtaining the maximum effective principal stress σ1, the minimum effective principal stress σ3, the cohesion C, and the internal friction angle φ, the shear failure discrimination value M at multiple azimuth angles of the wellbore circumference is calculated. s (θ): .

[0037] This embodiment uses the shear failure discriminant value M. s (θ) is normalized by dividing by the corresponding rock strength parameter. Specifically, take... As the corresponding intensity scale, calculation Meanwhile, when evaluating tensile fracture, a sign convention of positive compressive stress and negative tensile stress is adopted, and the tensile instability criterion value is calculated using the following formula: Where, σ t σ is the tensile strength of the rock. 3t T(θ) represents the minimum effective principal stress at the corresponding azimuth angle of the wellbore circumference. When T(θ) ≥ 0, it indicates that tensile fracture conditions have been met at that azimuth angle. Further, the tensile strength σ of the rock is taken as... tAs a normalization metric, the tensile fracturing margin is calculated as follows: Normalized theoretical wellbore instability margin F is taken as F s (θ) and F t The maximum value in (θ); when only the shear failure criterion is used, F takes F. s The maximum value of (θ). When F≥0, the theoretical calculation result corresponding to the depth point to be processed is determined to have reached the wellbore instability trigger condition; when F<0, it is determined to be a theoretically instable state.

[0038] S4, Evaluation of measured wellbore condition Based on well caliper logging curves CAL Calculate the measured expansion ratio ΔD based on the drill bit size BS: Where CAL is the wellbore logging value at the depth point to be processed, and BS is the drill bit size or the nominal wellbore size.

[0039] In this embodiment, the wellbore integrity threshold η is set to 5%, which is within the range of 0.03 to 0.08, and is determined based on the 90th to 99th percentiles of the enlargement rate of well sections without wellbore abnormalities in the historical wells of the target block. When ΔD < η, it is determined that the measured well diameter change is small, and the wellbore is in an intact or slightly enlarged state; when ΔD ≥ η, it is determined that the wellbore has enlargement characteristics.

[0040] To reduce the impact of burrs on the wellbore curve and local anomalies in short sections on trigger detection, this embodiment establishes a depth window of 1.0m above and below the depth point to be processed, and calculates the local stability coefficient R of the wellbore curve. cal R cal The value is greater than or equal to zero, and is determined based on the well diameter standard deviation, well diameter median deviation, maximum well diameter mutation, and well diameter data reliability marker within the window.

[0041] In this embodiment, the local stability coefficient R of the wellbore curve cal Calculate using the following formula: .

[0042] Among them, CAL w For wellbore data within the depth window, std(CAL) w ) represents the standard deviation of the well diameter within the window, max|ΔCAL w | represents the maximum abrupt change in diameter between adjacent wells within the window. In this embodiment, the local stability coefficient R of the well diameter curve is... cal Reliability threshold T cal Take 0.70, which falls within the range of 0.60 to 0.90, and base it on the R value of the target block's depth point without borehole anomalies. cal The 5th to 20th percentiles have been determined.

[0043] S5, Calculation of Counter-evidence Triggering and Inconsistency Evaluation Quantity The normalized theoretical wellbore instability margin F obtained in step S3 is jointly judged with the measured wellbore state obtained in step S4. When the depth point to be processed simultaneously satisfies F≥0, ΔD<η, and R... cal Greater than the reliability threshold T cal When calculating the theoretical-measured inconsistency intensity I, the drilling anomaly record is included in the calculation of I through the anomaly exclusion factor A, and is no longer used as an exclusion condition before calculating I.

[0044] When the drilling anomaly record does not contain any of the following wellbore anomalies: lost circulation, stuck pipe, reaming, logging obstruction, wellbore ellipticity, or logging distortion, and I is greater than the inconsistency intensity threshold I th When this occurs, it indicates that theoretical calculations show the wellbore has reached the instability trigger condition, while measured wellbore data shows the wellbore remains intact or only slightly enlarged, indicating a discrepancy between the theoretical instability judgment and the measured wellbore condition. This discrepancy does not directly conclude that all input data is incorrect, but rather, when pore pressure, rock strength parameters, effective stress coefficient, and regional tectonic parameters are constrained by independent data, it serves as priority evidence of deviations in the initial static Young's modulus and initial Poisson's ratio, triggering a closed-loop correction for the depth point to be processed.

[0045] Theoretical-measured inconsistency evaluation quantity I is calculated according to the gating relationship: .

[0046] Where A is the drilling anomaly record exclusion factor. In this embodiment, when the drilling anomaly record does not contain lost circulation, stuck pipe, reaming, logging obstruction, wellbore ellipticity, or logging distortion, A is 1; when the drilling anomaly record contains one of lost circulation, stuck pipe, wellbore ellipticity, logging distortion, or reaming, A is 0.2. In this case, I is used to de-weight records and credibility indicators for anomalies, and is not a sufficient condition for triggering closed-loop correction; w1 is 0.6, and w2 is 0.4.

[0047] When the depth point to be processed does not satisfy F≥0, ΔD<η, R cal Greater than the reliability threshold T cal If any of the following conditions are met: no wellbore anomaly or I greater than the inconsistency strength threshold Ith, closed-loop correction is not initiated. The initial static Young's modulus E0, initial Poisson's ratio ν0, initial shear modulus G0, and initial bulk modulus K0 of the depth point to be processed are directly output as the final interpretation results. For depth points with ΔD ≥ η, a wellbore anomaly label is added, and the wellbore anomaly evaluation process is initiated. For depth points containing wellbore anomalies but still satisfying F ≥ 0, ΔD < η, and R... cal Greater than the reliability threshold T cal The depth points are marked with additional anomaly impact records and low-confidence indicators.

[0048] S6, Candidate Parameter Generation, Sensitivity Ranking, and Two-Stage Adjustment When the depth point to be processed meets the closed-loop correction trigger condition, the closed-loop correction for that depth point is initiated. In this embodiment, dynamic and static parameter transformation based on the physical laws of the target block's rocks is used to generate correction candidate results. The method of generating these candidate parameters is not exactly the same as the interpretation model, calibration coefficients, or training sample set of P0 in step S2, and is at least different from the acquisition method of P0 in terms of transformation relationship or coefficient source. At the same time, the candidate parameters are determined by combining sensitivity ranking, experimental experience envelope, and coarse and fine adjustment processes, so that candidate generation, parameter ranking, and boundary constraints are executed as the same closed-loop correction step.

[0049] First, based on the density logging data, P-wave transit time logging data, and S-wave transit time logging data at the depth point to be processed, the formation bulk density ρ and P-wave velocity V are calculated. p and transverse wave velocity V s V p and V s The values ​​are obtained by converting the P-wave and S-wave time differences to units, respectively, and ρ is the formation volume density at the depth point to be processed.

[0050] According to ρ, V p and V s Calculate the dynamic Poisson's ratio ν d and dynamic Young's modulus E d : ; .

[0051] The modified static Young's modulus E is generated sequentially using the dynamic-static parameter transformation relationship. r and corrected static Poisson ratio ν r Corrected candidate results: ; Among them, a E b E a ν and b ν For the conversion coefficient between dynamic and static parameters.

[0052] In this embodiment, a E b E a ν and b ν The initial values ​​are determined by the dynamic and static experimental data of the core samples from the target block; their adjustment range is constrained by the experimental empirical envelope of the static and dynamic parameters of the rock mechanics of the target block. The experimental empirical envelope is based on the scatter data of the historical dynamic and static parameters of the core samples from the target block, and the upper and lower boundaries are obtained by statistical fitting and combining with the 90% to 99% confidence interval. In this embodiment, the 95% confidence interval is used.

[0053] During the closed-loop correction process, E is first... r ν r Cohesion C, internal friction angle φ, effective stress coefficient α, and pore pressure P p A relative disturbance amplitude of 0.5% to 5% is applied; in this embodiment, 5% is used. The change in the normalized theoretical wellbore instability margin F before and after the disturbance is calculated, and the sensitivity ranking of F to each input parameter is obtained. If C, φ, α, and P... p If the data is independently constrained by core strength tests, pressure interpretation, or regional calibration data, then priority should be given to adjusting E, which contributes significantly to the change in F. r and ν r Relevant conversion coefficients.

[0054] Table 1 Results of parameter perturbation sensitivity analysis As shown in Table 1, in the target block sample, the combined contribution of static Young's modulus E and static Poisson's ratio ν to the change in F is 65.4%, significantly higher than that of individual strength parameters, effective stress coefficient, and pore pressure term; simultaneously, C, φ, α, and P... p Since independent constraints are provided by core strength experiments, regional calibration, or pressure interpretation data, this embodiment prioritizes E and ν as objects of proof by contradiction under the condition of theoretical-experimental inconsistency.

[0055] The coefficient adjustment employs two stages: coarse adjustment and fine adjustment. Coarse adjustment determines candidate coefficient combinations within the experimental experience envelope. Fine adjustment, near these candidate coefficient combinations, uses a smaller step size than the coarse adjustment and employs at least one of the bisection rule and local grid search to determine candidate results. In this embodiment, 'a' is adjusted preferentially. E and a ν And adjust b synchronously according to the change of J. E and b ν Maximum number of adjustments N max Set to 50 times.

[0056] Each time you get E r and ν r Then, the corrected shear modulus G is calculated simultaneously. r and corrected bulk modulus K r : and E r ν r G r and K r Together, they serve as the candidate results for this correction.

[0057] S7, Constraint Verification For each generated modified candidate result, E r νr G r and K r Verification was performed on physical constraints, lithological stratification constraints, and continuity constraints between adjacent depths.

[0058] Physical constraints include: E r >0;0<ν r <0.5; G r >0; K r >0. Lithological stratification constraint requirement E r and ν r The data must be within the range of core test data corresponding to the target stratigraphy or lithology, and within the range of calibration data from adjacent wells. The continuity constraint of adjacent depths requires that the E values ​​at adjacent depths within the same stratigraphy... r and ν r The change amount does not exceed the continuous change threshold, which is determined by the statistical value of the change of adjacent sampling points in the interpretation curves of adjacent wells or core experimental data within the same layer, or is 5% to 20% of the range of parameters in that layer.

[0059] In this embodiment, E r The value range of ν is determined based on the static mechanical test results of the core sample from the target block; r The range of values ​​is determined based on lithology, core test results, and interpretation results from adjacent wells. If a certain generated E... r ν r G r or K r If the constraints are not met, the modified candidate result is discarded and neither considered a valid candidate result nor as the final interpretation result of the rock mechanics parameters. If a certain generated E... r ν r G r and K r If the constraints are met, proceed to the re-substitution and closed-loop verification step.

[0060] S8, re-substitute into closed-loop verification For each corrected candidate result that satisfies the constraints, the E corresponding to the current corrected candidate result will be... r and ν r Resubstitute the normalized theoretical wellbore instability margin calculation procedure. When recalculating the far-field horizontal principal stress, replace E0 and ν0 used in the far-field horizontal principal stress calculation in step S3 with E... r and ν r Other parameters include overburden pressure σ v pore pressure P p , regional tectonic strain parameter ε h ε H And the effective stress coefficient α is kept consistent with the data corresponding to the depth point to be processed.

[0061] After recalculating the far-field horizontal principal stress, the total principal stress at the wellbore is recalculated in conjunction with the drilling fluid column pressure and wellbore geometric boundary conditions. Based on the effective stress principle, the effective principal stress at the wellbore is re-obtained, and the re-obtained effective principal stresses at the wellbore are sorted by numerical value to redetermine the maximum effective principal stress σ1 and the minimum effective principal stress σ3.

[0062] Combining the rock strength parameters obtained in step S3, the instability discrimination values ​​at multiple azimuth angles around the wellbore are recalculated. These instability discrimination values ​​are then normalized by dividing by the corresponding rock strength scale, and the maximum value is taken to obtain the candidate theoretical wellbore instability margin F. r .

[0063] Construct a joint objective function J that integrates wellbore mechanical response and formation continuity: .

[0064] Among them, P nr The candidate static parameter vector P r The dimensionless vector P after parameter scaling normalization n0 Let θ be the dimensionless vector obtained by normalizing the initial static parameter vector P0 using the same parameter scale. nr The correction coefficient vector θ corresponding to the candidate parameter generation method r The dimensionless vector θ after parameter scaling normalization n0 Let θ0 be the dimensionless vector after normalization to the same parameter scale from the initial correction coefficient vector θ0, and ||·|| denote the second norm of the vector.

[0065] ∇zP nr P represents the candidate static parameter normalized vector. nr The rate of change of adjacent depths along the depth z-direction. For discrete depth sampling points z i-1 and z i , When the depth sampling interval is a constant Δz, For the first depth sampling point, the difference between the next adjacent depth points is used for calculation.

[0066] In this embodiment, λ1 is 0.6, λ2 is 0.2, λ3 is 0.1, and λ4 is 0.1. The aforementioned weights are determined by normalization based on the reliability of wellbore caliper data in historical wells of the target block, the continuity of parameter curves, and the consistency of core calibration. Specifically, λ1 is increased when the reliability of the wellbore caliper data is high and the contradictions in the evidence are prominent; λ3 is increased when the stratigraphic continuity is strong or the lithological interface is indistinct. Each corrected candidate result that satisfies the constraints is recorded, including the correction coefficient vector θ. r a in E b E aν b ν E r ν r G r K r The recalculated σ1 and σ3, and the recalculated F r The objective function J and the current number of adjustments are combined.

[0067] When normalizing the range of the parameter component x, the following method is used: , where x min and x max These are the lower and upper limits of the parameter for the target stratigraphy or target lithology, respectively; when standardizing the parameter component x, the following is adopted: , where μ x and σ x The mean and standard deviation of the core test samples from the target block and the calibration samples from adjacent wells. When, let the normalized value of the range of this parameter component be x. n =0; when σ x When =0, let the standardized value x of this parameter component be... n =0.

[0068] S9, Closed-loop termination and output If the currently generated correction candidate result satisfies F r If the value is less than 0 and the joint objective function J reaches the convergence condition, then stop the closed-loop correction and assign the currently generated correction candidate result to E that satisfies the constraints. r ν r G r and K r This serves as the final interpretation result of the rock mechanical parameters at the depth point to be processed. Preferably, when F r The condition falls within the range [-0.15, 0) and J is less than or equal to the convergence threshold J. th If the relative decrease in J corresponding to two consecutive corrected candidate results that satisfy the constraints is less than the convergence tolerance δJ, then the convergence condition is determined to be met; the F r The target interval is used to avoid candidate results obtaining the theoretically instability state with only excessive parameter offset.

[0069] If the currently generated correction candidate result does not satisfy F r If the condition is less than 0 or J does not meet the convergence condition, continue adjusting the correction coefficient and repeat steps S6 to S8 until the termination condition is met or the maximum number of adjustments N is reached. max .

[0070] If the maximum number of adjustments N is reached maxIf no corrected candidate result satisfying the convergence condition is obtained after all subsequent adjustments, and there are corrected candidate results satisfying the constraints in each adjustment, then among the corrected candidate results satisfying the constraints, the one selected is max(0,F). r The minimum value is used as the first screening criterion to determine the modified candidate subset, and the modified candidate result with the smallest J is selected as the final rock mechanical parameter interpretation result for the depth point to be processed. At this time, the non-convergence indicator and the low confidence indicator are output simultaneously.

[0071] If the maximum number of adjustments N is... max If none of the candidate correction results satisfy the constraints, the closed-loop correction is stopped, the depth point to be processed is marked as a correction failure and a low confidence depth point, and the initial rock mechanics parameters E0, ν0, G0 and K0 are output as reference results.

[0072] Furthermore, a low confidence indicator is output at least if: the number of candidate results satisfying the constraints is less than the minimum number of candidates N. min N min The integers are from 1 to 5; the maximum number of adjustments N is reached. max F is still not satisfied r <0, where N max The value is a positive integer between 10 and 200; the corrected parameter change reaches the upper limit of parameter change determined by the experimental experience envelope or adjacent well calibration data; the local stability coefficient R of the wellbore curve cal Below the reliability threshold T cal Or the wellbore data reliability is marked as low; or in N max There are no corrected candidate results that satisfy the constraints of step S7.

[0073] S10, Depth-by-depth processing and wellbore stability evaluation output According to the order of the depth points to be processed in the target well section, repeat steps S2 to S9 until all depth points in the target well section have been processed, and obtain the final rock mechanics parameter interpretation result set, reference result set, correction trigger depth point set, correction coefficient trajectory, theoretical wellbore instability margin change trajectory, convergence flag set, correction failure flag set and confidence flag set for each depth point in the target well section.

[0074] Based on the final set of rock mechanics parameter interpretation results, corrected rock mechanics parameter inputs are provided for the calculation of collapse pressure, fracture pressure, and safe drilling fluid density windows, and a wellbore stability evaluation profile for the target well section is generated. For depth points where the initial rock mechanics parameters are output as low-confidence reference results, risk warnings are simultaneously output in subsequent calculations of collapse pressure, fracture pressure, and safe drilling fluid density windows, and the range of rock mechanics parameter values ​​corresponding to that depth point is expanded by ±10% to ±15% to form an uncertainty band; the upper boundary of the safe drilling fluid density window is calculated based on the upper limit of the expanded parameters, and the lower boundary is calculated based on the lower limit of the expanded parameters, with the window boundary expanded by an average of 0.02 g / cm³ to 0.06 g / cm³, and displayed as a dashed line and risk indicator in the wellbore stability evaluation profile; in the interpretation of continuous curves, the smoothing propagation weight of this depth point to adjacent normal depth points is reduced to 0.2 to 0.5 of the conventional weight.

[0075] After obtaining the final rock mechanics parameter interpretation results set for each depth point in the target well section, boundary smoothing is performed on the final rock mechanics parameter curves within the same stratum. When adjacent depth points cross strata boundaries, lithological abrupt changes, or well diameter data anomalies, the cross-boundary smoothing weight is reduced or cross-boundary smoothing is stopped to avoid erroneous propagation of local correction results to other geological units.

[0076] Supplementary Example 1: In a shale well section using a lithological stratification calibration candidate generation method, the target depth is 4200m to 4350m, with a sampling interval of 0.25m, resulting in 600 depth points to be processed. Based on natural gamma, density, P-wave transit time, and S-wave transit time curves, the well section is divided into three lithologies: shale, mudstone-shale, and interbedded sandstone-mudstone. Stratigraphic correction coefficients and lithological correction coefficients are then set for each lithology, forming a correction coefficient vector θ. r In this embodiment, η is taken as 0.045, and T... cal Take 0.72, I th Take 0.055, N max With a value of 60, λ1, λ2, λ3, and λ4 are set to 0.55, 0.20, 0.15, and 0.10, respectively. Sensitivity ranking results show that F is more sensitive to the shale stratigraphic correction coefficient and the Poisson's ratio stratigraphic correction coefficient; therefore, these two coefficients are adjusted first during the coarse adjustment stage. Calculations show that 64 depth points trigger closed-loop correction, of which 58 depth points satisfy F. r If the value is less than 0 and the convergence condition is met, output a low confidence level indicator for each of the 6 depth points.

[0077] Table 2 Calculation results of candidate generation methods for lithological stratification calibration Supplementary Example 2: In a fault-block well section using a combination of adjacent well parameter migration and core experiment-constrained regression candidate generation method, the target depth is 2860m to 2980m, with a sampling interval of 0.25m, resulting in 480 depth points to be processed. Adjacent wells within the same structural zone with corresponding stratigraphic levels, a natural gamma curve correlation coefficient of not less than 0.65, and qualified wellbore data quality are selected as migration samples. Outliers are removed from the adjacent well samples using the interquartile range method, and the migration weights of adjacent wells are determined according to stratigraphic distance, curve correlation coefficient, and data reliability. In this example, N... max With a value of 80, and λ1, λ2, λ3, and λ4 set to 0.50, 0.20, 0.10, and 0.20 respectively, the sensitivity ranking results show that F is more sensitive to the migration weight of adjacent wells and the core regression constraint coefficient. Therefore, the migration weight of adjacent wells is adjusted first in the coarse adjustment stage, and the core regression constraint coefficient is adjusted in the fine adjustment stage. In this embodiment, 52 depth points trigger closed-loop correction, 44 depth points converge, 5 depth points do not converge but have valid candidates, and the initial parameters of 3 depth points are output as low-confidence reference results.

[0078] Table 3 Calculation results of the combination of adjacent well migration and core experiment constraint regression. Implementation Results Description like Figure 2 As shown in this embodiment, the normalized theoretical wellbore instability margin F, enlargement ratio ΔD, and local stability coefficient R of the wellbore curve are calculated depth by depth for the well section from 3250m to 3750m. cal The evaluation quantity I for inconsistency between theory and measurement, and based on F≥0, ΔD<η, R cal Greater than the reliability threshold T cal The drilling anomaly record does not include well leakage, stuck pipe, enlargement, logging obstruction, wellbore ellipse, logging distortion, and the condition I is greater than the inconsistency intensity threshold Ith. The closed-loop correction trigger interval is identified.

[0079] Figure 2 In the normalized theoretical wellbore instability margin F, the instability margin F crosses the F=0 criterion line multiple times. The depth points where F≥0 represent the theoretical instability triggering state. The diameter enlargement rate is below the wellbore integrity threshold of 5% at most depth points, indicating that the measured wellbore is mainly intact or slightly enlarged.

[0080] according to Figure 2The results show that the main closed-loop correction triggering intervals within the 3250m to 3750m well section include: 3276m to 3278m; 3302m to 3307m; 3334m to 3359m; 3400m to 3406m; 3436m to 3443m; 3569m to 3572m; 3582m to 3593m; 3605m to 3609m; 3640m to 3648m; 3699m to 3703m; 3707m to 3727m; 3734m to 3742m; and an additional 7m of scattered thin-layer triggering points.

[0081] The cumulative length of the closed-loop correction triggering well section is 110m, accounting for 22.0% of the total length of the verification well section from 3250m to 3750m.

[0082] For the aforementioned triggering closed-loop correction interval, this embodiment does not directly use E0, ν0, G0, and K0 obtained from the initial interpretation. Instead, it uses the conversion relationship between P-wave and S-wave time difference and density dynamic and static parameters, independent of the initial continuous interpretation relationship, and prioritizes adjusting a. E and a ν And simultaneously fine-tune b E and b ν Recalculate E r and ν r And calculate G simultaneously r and K r .

[0083] Depend on Figure 2 The Young's modulus and Poisson's ratio curves before and after the correction show that the E and ν curves after the correction still maintain the vertical variation trend of the original parameters. The correction is mainly concentrated near the trigger closed-loop correction interval and does not cause overall shift or abnormal change of parameters in the whole well section.

[0084] Among them, the corrected Young's modulus E is distributed in the range of 20 GPa to 42 GPa, and the corrected Poisson's ratio ν is distributed in the range of 0.23 to 0.30. The corrected parameters are still within the range of experimental data of target stratigraphic layers and calibration data of adjacent wells for rock mechanics parameters in this well section.

[0085] After closed-loop correction, F is recalculated within the convergence trigger interval. r Satisfy F r <0 and the joint objective function J reaches the convergence condition, the theoretical instability judgment result tends to be consistent with the measured wellbore state; for depth points that have not reached the convergence condition but still retain valid candidates, according to max(0,F rThe minimum J and minimum J rules output low-confidence results; for depth points where none of the candidates satisfy the constraints, the initial parameters are output as low-confidence reference results; for depth points containing wellbore anomalies, I is recorded with reduced weight according to the second value of A, but this is not considered a sufficient condition for triggering closed-loop correction. Based on drilling data, the above closed-loop correction trigger interval did not record wellbore collapse, blockage, stuck pipe, or leakage anomalies, indicating that the measured wellbore condition is consistent with ΔD < η and R. cal The reliability assessment indicates a consistent state of integrity or slight enlargement.

[0086] To further illustrate the synergistic effect among gated inconsistency strength, caliper reliability, anomaly weighting, empirical envelope constraints, and normalized joint objective function, this embodiment sets up ablation comparisons based on verification data consistent with the parameter range of the 3250m to 3750m well section. All comparison schemes use the same initial rock mechanics parameters, the same caliper curve, and the same drilling anomaly records; the only difference lies in whether the corresponding gating, constraints, and fallback output logic are enabled.

[0087] Table 4 Ablation Comparison Results As shown in Table 4, when using only theoretical instability and enlargement rate conditions, the triggering interval length increases and the false triggering rate is relatively high; after adding well diameter reliability and anomaly elimination factors, false triggering is suppressed; further addition of experimental experience envelope and ∇zP... nr and θ nr After normalizing the joint objective function of the terms, the number of parameter mutation points is significantly reduced; after adding low-confidence reference output and risk warning in the case of full candidate elimination, the closed-loop process is no longer interrupted.

[0088] Further statistical analysis of the three embodiments reveals that the closed-loop correction triggering rates for the main embodiment, Supplementary Embodiment 1, and Supplementary Embodiment 2 were 22.0%, 10.7%, and 10.8%, respectively; the convergence rates were 88.2%, 90.6%, and 84.6%, respectively; and the low-confidence output rates were 7.3%, 9.4%, and 15.4%, respectively. In the subsequent calculation of the safe drilling fluid density window, the low-confidence depth points, after expanding the parameter range to form an uncertainty band, had their window boundaries expanded by an average of 0.02 g / cm³ to 0.06 g / cm³, and were distinguished by dashed lines and risk indicators in the wellbore stability evaluation profile. These results demonstrate that, under different lithologies and different candidate parameter generation methods, this method can maintain consistent closed-loop logic for gating triggering, constrained correction, and risk output.

[0089] Further comparison of the safe drilling fluid density window before and after correction with the actual drilling density without abnormalities revealed that the average deviation between the window boundary and the actual drilling density without abnormalities before correction was 0.08 g / cm³ to 0.12 g / cm³. After adopting the complete closed-loop process of this invention, the average deviation was reduced to 0.04 g / cm³ to 0.07 g / cm³. The comprehensive prediction accuracy for well leakage, collapse, and enlargement anomalies improved by 12% to 18%, and the false alarm rate decreased by 9% to 15%. These results demonstrate that gated counter-evidence triggering, constrained candidate correction, and low-confidence fallback output are not simply superimposed, but rather produce synergistic technical effects in reducing false triggering, suppressing non-physical mutations, and maintaining the continuity of engineering output.

[0090] Therefore, this embodiment can identify depth points where the initial rock mechanics parameter interpretation results are inconsistent with the actual wellbore state, and perform closed-loop correction on the local anomaly interpretation results through candidate parameter generation constrained by the rock mechanics experimental experience envelope, sensitivity ranking, two-stage adjustment, and resubstitution into the normalized theoretical wellbore instability margin calculation process.

[0091] In summary, this embodiment demonstrates that the present invention can combine continuous logging interpretation results, effective wellbore stress state, rock strength parameters, and measured wellbore conditions. When theoretical wellbore instability occurs while the measured wellbore remains intact or only slightly enlarged, it initiates a candidate parameter generation and closed-loop correction process that differs from the initial parameter acquisition method in terms of model or coefficients. This process recalculates and performs closed-loop discrimination on static Young's modulus, Poisson's ratio, shear modulus, and bulk modulus. This method is applicable to continuous interpretation of rock mechanics parameters, wellbore stability analysis, calculation of collapse and fracture pressures, and design of safe drilling fluid density windows under vertical well conditions.

Claims

1. A closed-loop correction method for logging rock mechanics parameters used in wellbore stability evaluation, characterized in that, The steps include: S1, acquiring logging data, formation data, overburden pressure data, pore pressure data, drilling fluid density data, drill bit size data, regional structural parameters, effective stress coefficient, well diameter data, rock strength parameter data, and drilling anomaly records for the target well section; establishing a depth sequence to be processed based on logging depth sampling points; and completing the depth repositioning of discrete data and mapping of drilling anomaly intervals. S2, Obtain the initial rock mechanics parameter vector P0 at the depth point to be processed. P0 includes at least the initial static Young's modulus E0 and the initial Poisson's ratio ν0. Calculate the initial shear modulus G0 and the initial bulk modulus K0 based on P0. S3, Calculate the effective principal stress at the wellbore based on P0, overburden pressure, pore pressure, drilling fluid density, regional structural parameters, and effective stress coefficient. Combine this with rock strength parameters to obtain the normalized theoretical wellbore instability margin F. The normalized theoretical wellbore instability margin F includes at least one of shear failure margin and tensile fracture margin. The maximum value of the normalized instability discrimination value at multiple azimuth angles around the wellbore circumference is taken as F at that depth point. When F≥0, the theoretical instability triggering condition is determined to have been met. S4, based on caliper logging curves CAL Calculate the expansion ratio based on the drill bit size BS. The local stability coefficient R of the wellbore curve is calculated based on at least two of the following: wellbore standard deviation, wellbore median deviation, maximum wellbore mutation, and wellbore data reliability markers within the depth window containing the depth point to be processed. cal S5, when the same depth point to be processed satisfies F≥0, ΔD<η and R cal >T cal At that time, the gated theoretical-measured inconsistency intensity I is calculated by combining the anomaly exclusion factor A corresponding to the drilling anomaly record; When A is a normal value and I is greater than the inconsistency intensity threshold I th When the theoretical instability judgment result is inconsistent with the measured wellbore state, a closed-loop counter-evidence correction is triggered for P0; otherwise, E0, ν0, G0, and K0 corresponding to P0 are used as the final rock mechanics parameter interpretation results for the depth point to be processed; S6, for the depth point to be processed that triggers the closed-loop counter-evidence correction, a modified static Young's modulus E is generated using a candidate parameter generation method that is not completely identical to the interpretation model, calibration coefficients, or training sample set of P0. r and corrected static Poisson ratio ν r Simultaneously calculate the corrected shear modulus G. r and corrected bulk modulus K r The candidate parameter generation method differs from the P0 acquisition method in at least one of the parameter sources, transformation relationships, or migration weights, and candidate adjustments are made in conjunction with sensitivity ranking within the envelope of rock mechanics experimental experience; S7, each generated modified candidate result is verified by physical constraints, lithological stratification constraints, prior range constraints from adjacent wells or cores, and adjacent depth continuity constraints, and modified candidate results that do not meet the constraints are eliminated; S8, the modified candidate results that meet the constraints are substituted back into step S3 to obtain the candidate theoretical wellbore instability margin F. r And calculate the joint objective function J, which integrates wellbore mechanical response and formation continuity, using the following formula: ； Among them, P nr P n0 θ nr and θ n0 All are dimensionless vectors normalized to the same scale, ∇zP nr Let ||·|| be the rate of change of the dimensionless candidate static parameter vector along the depth direction between adjacent depths, ||·|| denotes the vector L2 norm, and λ1, λ2, λ3, and λ4 are positive weighting coefficients. ;when When, max(0,F) r The term is used to eliminate false alarms of theoretical instability, and the joint objective function still constrains the correction magnitude through parameter offset, vertical continuity, and correction coefficient offset terms; preferably, when F r If J falls within the interval [-0.15, 0) and satisfies the convergence condition, convergence is determined; S9, if the current corrected candidate result satisfies F r If the value of J is less than 0 and the convergence condition is met, then the corrected candidate result is taken as the final interpretation result of the rock mechanics parameters; if the maximum number of adjustments is reached but convergence is still not achieved and there exists a corrected candidate result that satisfies the constraints, then the result is taken as max(0,F). r The minimum and minimum J are the rules for outputting low confidence correction results; if there are no correction candidate results that meet the constraints, the parameters corresponding to P0 are output as low confidence reference results; S10, repeat steps S2 to S9 according to the depth sequence to be processed to obtain the final rock mechanics parameter interpretation result set, correction trigger depth point set, correction trajectory set and confidence identifier set of the target well section, and use the final rock mechanics parameter interpretation result set for the calculation of collapse pressure, fracture pressure and safe drilling fluid density window to generate the well wall stability evaluation profile of the target well section.

2. The method according to claim 1, characterized in that, In step S1, the logging data includes at least density logging data, P-wave time-of-flight logging data, S-wave time-of-flight logging data, natural gamma logging data, and caliper logging data; the drilling anomaly records include at least one of lost circulation, stuck pipe, enlargement, logging obstruction, wellbore ellipticity, and logging distortion records; for long-span drilling anomaly records with a continuous span greater than 20m, they are first segmented according to the stratigraphic interface and the abrupt change point of caliper quality. If the segmentation is still greater than 50m, then the anomaly impact is assigned to segments of 10m to 50m sub-windows.

3. The method according to claim 1, characterized in that, In step S2, when P0 is obtained from the continuous interpretation relationship of well logging, the continuous interpretation relationship of well logging establishes interpretation sub-relationships according to the stratigraphy, lithology, or sedimentary facies, and outputs the initial interpretation confidence level for each depth point to be processed; when P0 is obtained from the core calibration results or the interpretation results of adjacent wells, the source type, applicable stratigraphy, applicable depth range, and initial interpretation confidence level of P0 are recorded synchronously for each depth point to be processed.

4. The method according to claim 1, characterized in that, In step S3, the shear failure margin is obtained by the Mohr-Coulomb shear failure criterion, and the tensile fracture margin is obtained by the difference between the minimum effective principal stress of the wellbore and the tensile strength of the rock. When both instability criteria are used, the maximum value of the shear failure margin and the tensile fracture margin is taken as the normalized theoretical wellbore instability margin F.

5. The method according to claim 1, characterized in that, In step S4, the local stability coefficient R of the wellbore curve cal Calculate using the following formula: ； Among them, CAL w For wellbore data within the depth window, std(CAL) w ) represents the standard deviation of the well diameter within the window, max|ΔCAL w | represents the maximum abrupt change in diameter between adjacent wells within the window; the depth window length is 1.0m to 2.0m or 5 to 10 sampling points; the reliability threshold T cal Take the R of the depth point without wellbore anomalies in the target block. cal 5th to 20th percentiles.

6. The method according to claim 1, characterized in that, In step S5, the gated theoretical-measured inconsistency strength I is calculated using the following formula: ； Wherein, A is the drilling anomaly record exclusion factor. When the drilling anomaly record does not contain any of the following wellbore anomalies: lost circulation, stuck pipe, reaming, logging obstruction, wellbore ellipticity, and logging distortion, A takes the first value. When the drilling anomaly record contains any of the above wellbore anomalies, A takes the second value. The first value is 1, and the second value is 0 to 0.

5. w1 and w2 are weight coefficients greater than zero, and w1 + w2 = 1. When A takes the second value, I is used to reduce the weight of anomaly records and the credibility indicator, and is not used as a sufficient condition for triggering closed-loop rebuttal correction.

7. The method according to claim 1, characterized in that, In step S6, the candidate parameter generation method includes at least one of dynamic-static parameter conversion, lithological stratification calibration, adjacent well parameter migration, and core experiment constrained regression; when dynamic-static parameter conversion is used, the correction coefficient includes Young's modulus conversion slope, Young's modulus conversion intercept, Poisson's ratio conversion slope, and Poisson's ratio conversion intercept; when lithological stratification calibration is used, the correction coefficient includes a stratigraphic correction coefficient or a lithological correction coefficient; when adjacent well parameter migration is used, the correction coefficient includes adjacent well migration weight. When core experiments are used to constrain regression, the correction coefficients include the regression constraint coefficients; the sensitivity ranking includes separate ranking of E... r ν r Cohesion C, internal friction angle φ, effective stress coefficient α, and pore pressure P p At least two of the parameters are subjected to perturbations with relative amplitudes ranging from 0.5% to 5%. The changes in F before and after the perturbation are calculated, and the values ​​of C, φ, α, and P are also considered. p When data is independently constrained by core strength tests, pressure interpretation, or regional calibration data, priority should be given to adjusting E, which contributes significantly to the change in F. r and ν r Relevant correction factors; The experimental experience envelope is an upper and lower boundary obtained by statistical fitting and combining historical core dynamic and static parameter scatter data of the target block with a 90% to 99% confidence interval. The candidate adjustment includes coarse adjustment and fine adjustment. The coarse adjustment is used to determine the candidate coefficient interval within the experimental experience envelope. The fine adjustment is used to determine the candidate result within the candidate coefficient interval by using a step size smaller than that of the coarse adjustment and employing at least one of the bisection rule and local grid search.

8. The method according to claim 1, characterized in that, In step S7, the physical constraints include at least E r >0、0<ν r <0.5, G r >0, K r >0; the lithological stratification constraints include E r and ν r Located within at least one of the core test data range corresponding to the target stratigraphy or target lithology and the calibration data range of adjacent wells; the adjacent depth continuity constraint includes E of adjacent depth points within the same stratigraphy. r and ν r The amount of change does not exceed the threshold for continuous change.

9. The method according to claim 1, characterized in that, In step S9, the low confidence indicator is output at least in one of the following cases: the number of candidate results that satisfy the constraints is less than the minimum number of candidates N. min N min The integers are from 1 to 5; the maximum number of adjustments N is reached. max F is still not satisfied r <0, where N max Positive integers from 10 to 200; The corrected parameter change reaches the upper limit of parameter change determined by experimental experience envelope or adjacent well calibration data; Local stability coefficient R of wellbore curve cal Below the reliability threshold T cal Or the wellbore data reliability is marked as low reliability; or in N max There are no corrected candidate results that satisfy the constraints of step S7.

10. The method according to claim 1, characterized in that, After obtaining the final rock mechanics parameter interpretation results set for each depth point in the target well section, the final rock mechanics parameter interpretation results within the same layer are subjected to boundary smoothing. When adjacent depth points cross layer interfaces, lithological abrupt change points, or well diameter data anomalies, the cross-boundary smoothing weight is reduced or cross-boundary smoothing is stopped. For low confidence depth points, in the subsequent calculation of collapse pressure, fracture pressure, and safe drilling fluid density window, the corresponding rock mechanics parameter value range is expanded by ±10% to ±15% to form an uncertainty zone. The upper boundary of the safe drilling fluid density window is calculated based on the upper limit of the expanded parameters, and the lower boundary is calculated based on the lower limit of the expanded parameters. The window boundary is expanded by an average of 0.02 g / cm³ to 0.06 g / cm³, and is displayed as a risk indicator in the wellbore stability evaluation profile.

Images