Volume fractured horizontal well segmentation clustering method based on mechanical specific energy and genetic algorithm

Abstract

The invention discloses a volume fractured horizontal well segmentation clustering method based on mechanical specific energy and a genetic algorithm, and relates to the field of well drilling and completion engineering in petroleum and natural gas. A traditional geometric division method is difficult to meet the refining requirement on a segmentation scheme in actual engineering; drilling dynamic data are collected, a drilling dynamic data set table under the equal well depth interval is formed according to the well depth step length, and drill torque and bit pressure are calculated; mechanical specific energy of the horizontal well section is calculated according to the drilling dynamic data set table, and a mechanical specific energy value set along a borehole is obtained; generating an initial population based on a set constraint condition, and obtaining the fitness of each individual in the initial population; randomly extracting a set number of individuals from the initial population to form a candidate set, and performing fitness calculation, selection, crossover and mutation operations until the maximum number of iterations is reached to complete iterative optimization; and outputting an optimal segmentation scheme and perforation cluster distribution, and performing drilling construction according to the optimal segmentation scheme and the perforation cluster distribution.

Description

Technical Field

[0001] The present invention relates to the field of drilling and completion engineering in oil and gas, and specifically to a method for segmenting and clustering horizontal wells in volume fracturing based on mechanical specific energy and genetic algorithm. Background Art

[0002] With the rapid development of social economy, the demand for the exploitation of unconventional oil and gas resources (such as shale gas and tight oil) is becoming increasingly urgent. In the process of exploiting these resources, volume fracturing technology has become the key technology to improve the oil and gas production per well. The volume fracturing technology can significantly enhance the fluidity of oil and gas around the wellbore by creating a complex fracture network in the reservoir, effectively improving the productivity per well. The horizontal well drilling technology, as an optimal supporting technology for volume fracturing, can make the wellbore drill along with the reservoir, thereby increasing the contact area between the wellbore and the reservoir, providing strong support for the efficient exploitation of oil and gas resources. However, to achieve good results of the horizontal well volume fracturing technology, it is necessary to segment and cluster the horizontal section based on the physical properties of the reservoir. The segmentation and clustering need to be based on various attributes of the drilled reservoir. First, the reservoir is divided into different segments according to the attributes, and then each segment is further subdivided into clusters according to the distribution characteristics of the attributes. The segmentation and clustering play an important role in effectively tapping the production of oil and gas wells. However, in actual operation, due to the complex variety of parameters characterizing the physical properties of the reservoir, it is also necessary to select appropriate parameters to carry out the horizontal section division. Mechanical specific energy, as a physical quantity calculated through various engineering parameters (such as drilling pressure, torque, drilling speed, etc.) during the actual drilling process, can more accurately reflect the difficulty of rock fragmentation and can be used as a comprehensive index reflecting the reservoir heterogeneity. In addition to the selection of division parameters, when dealing with complex multi-constraint conditions and dynamic adjustment requirements, traditional geometric division methods have deficiencies in dealing with multi-constraints (such as section length, non-fracturing well section) and dynamic adjustment requirements, and it is difficult to meet the refined requirements for the segmentation scheme in actual engineering. As an optimization algorithm, the genetic algorithm simulates natural selection and genetic mechanisms, and can quickly search for the global optimal solution or approximate optimal solution in complex multi-dimensional optimization problems. In view of this, the purpose of the present invention is to make full use of the advantages of mechanical specific energy and genetic algorithm to develop a method for segmenting and clustering horizontal wells in volume fracturing. This method calculates the mechanical specific energy of the drilled reservoir through drilling data, and combines the genetic algorithm to globally optimize the segmentation scheme, which can effectively handle complex constraints and dynamic adjustment requirements, and generate a better segmentation scheme. This method can provide effective technical support for the precise control and optimization of horizontal well volume fracturing operations, thereby effectively improving the production and economic benefits of oil and gas wells, promoting the efficient development of unconventional oil and gas resources, and having significant social and economic benefits and broad application prospects. Summary of the Invention

[0003] The purpose of the present invention is to overcome the deficiencies of the prior art and provide a method for segmenting and clustering horizontal wells in volume fracturing based on mechanical specific energy and genetic algorithm, including the following steps: Step 1: Collect drilling dynamic data, form a drilling dynamic data set table at equal well depth intervals according to the well depth step length, and calculate the bit torque and weight on bit. Step 2: Calculate the mechanical specific energy of the horizontal section based on the drilling dynamic data set table to obtain a set of mechanical specific energy values along the wellbore. Step 3: Generate an initial population based on the set constraints and obtain the fitness of each individual in the initial population. Step 4: Randomly select a set number of individuals from the initial population to form a candidate set. Use the tournament selection method to screen out the individuals with high fitness in the candidate set; generate offspring through the segment boundary replacement strategy for crossover operation; randomly adjust the segment boundary depth and reorder for mutation operation; repeat the fitness calculation, selection, crossover, and mutation operations until the maximum number of iterations is reached to complete the iterative optimization. Step 5: Output the optimal segmentation scheme and perforation cluster distribution, and perform drilling construction according to the optimal segmentation scheme and perforation cluster distribution.

[0004] Further, the forming of the drilling dynamic data set table at equal well depth intervals according to the well depth step length and the calculation of the bit torque and weight on bit include: Based on the interval from the starting well depth to the ending well depth of the horizontal section, take values of the drilling dynamic data at the set well depth step length within the interval to form a drilling dynamic data set table at equal well depth intervals; calculate the bit torque and weight on bit corresponding to each record based on the drilling dynamic data set table: Calculate the bit torque and weight on bit for the sections with and without the use of a positive displacement motor (PDM) respectively. For the section with a PDM, the bit torque is TBit = Kt*ΔP, where Kt is the torque output per unit pressure difference of the PDM, and ΔP is the pressure difference of the PDM in the drilling state; where ΔP is estimated using the standpipe pressure, which is the standpipe pressure when the bit is drilling at the bottom of the well minus the standpipe pressure when the bit is lifted off the bottom of the well; the weight on bit is estimated through the measured hook load, which is the hook load when the bit is lifted off the bottom of the well minus the hook load when the bit is drilling at the bottom of the well. For the section without the use of a PDM, the bit torque is predicted using the friction torque model. The rotary table torque is used as the input value of the friction torque model to calculate the bit torque from the wellhead to the bottom of the well, and the estimation method of the weight on bit is the same as when using a PDM.

[0005] Further, the calculation of the mechanical specific energy of the horizontal section based on the drilling dynamic data set table to obtain a set of mechanical specific energy values along the wellbore includes: Based on the drilling dynamic data, calculate the mechanical specific energy at the set well depth step length within the horizontal section interval to form a mechanical specific energy data set. The mechanical specific energy is calculated using the following formula: Wherein, Dbit is the bit diameter, ROP is the drilling rate, Kn is the output rotation speed of the screw under the unit mud flow rate, Q is the mud flow rate, TBit is the bit torque, N is the rotary table speed. If there is no screw, Kn is taken as 0.

[0006] Further, generating an initial population based on the set constraint conditions includes: The constraint conditions include the number of fracturing stages, the range of stage lengths, the range of the number of clusters per stage, the length of each cluster, and the non-fracturing well section. Generating the initial population includes: Step 31: Set the total number of individuals in the population. Each individual represents a fracturing segmentation scheme and consists of fracturing stage number - 1 fracturing boundary depth values. The boundary values are arranged in ascending order. Step 32: Randomly generate individuals. Starting from the starting depth, generate fracturing boundaries with random step lengths. The step length range is within the range of stage lengths. Step 33: Repeat Step 31 and Step 32 until the number of individuals in the initial population reaches the total number of individuals set in the population.

[0007] Further, obtaining the fitness of each individual in the initial population includes: For the legality verification of the fracturing boundary of an individual, first perform the fracturing stage number check to verify whether the number of fracturing boundaries of the individual is fracturing stage number - 1. If not, directly return -Infinity. Then perform the stage length constraint check: traverse all fracturing stages and calculate whether the length of each stage satisfies minStageLength ≤ stage length ≤ maxStageLength. Record the number of violations stageViolations. The calculation method of the number of violations is: For the perforation cluster division and verification of an individual, first perform perforation cluster generation. For each fracturing stage, divide the perforation clusters according to the length of each cluster; and the length of each cluster is greater than half of the length of each cluster. Then perform the cluster number constraint check, count the number of perforation clusters in each stage. If it exceeds the range of the number of clusters per stage, record the number of violations. Perform fitness calculation. First, perform data mapping, map the well depth range of the perforation clusters to the mechanical specific energy dataset. For each perforation cluster, extract the mechanical specific energy values within its corresponding well depth range and calculate the cluster variance. Cluster variance = Variance(MSEs, MSEs + 1, …, MSEe) Where s and e are the data indices corresponding to the starting and ending depths of the cluster respectively. If the cluster length is invalid, the variance is set to 0. Finally, sum up the cluster variances of all perforation clusters and take the negative value as the fitness. If there are any stage or cluster violations, return -Infinity to impose a penalty. The final fitness formula is: .

[0008] Further, randomly extracting a set number of individuals from the initial population to form a candidate set, and screening out the individuals with high fitness in the candidate set by using the tournament selection method, including: Step 41: Randomly extract a set number of individuals from the population to form a candidate set; compare the fitness values of the candidate individuals, and select the individual with the highest fitness as the winner; Step 42: Repeat Step 41 until the size of the new population reaches the total number of individuals in the set population.

[0009] Further, generating offspring through a segmented boundary replacement strategy and performing crossover operations, including: randomly selecting two parent individuals parent1 and parent2 from the population, for their segmented boundary lists, traversing each segmented boundary position, and generating an offspring child by selecting the corresponding boundary value of parent1 or parent2 with a set probability; reordering the boundary values of the offspring to ensure monotonic increase, and constraining the boundary values of child within the valid interval.

[0010] Further, randomly adjusting the depth of the segmented boundary and reordering it to perform mutation operations, including: each individual in the population triggers the mutation operation with a set probability, randomly selects a position in the individual's segmented boundary list, applies a random perturbation within a set range to the boundary value corresponding to this position to generate a new value, and constrains it to the valid interval; replaces the original boundary value and reorders the entire list, if the perforation clusters generated after mutation conflict with non-fractured well sections, a penalty is imposed in the fitness calculation.

[0011] Further, outputting the optimal segmentation scheme and the distribution of perforation clusters, including: dividing the perforation clusters from the optimal solution of the genetic algorithm according to the length of each cluster, ensuring that the number of clusters is within the range of the number of clusters in each section and avoiding non-fractured well sections; performing depth correction, including comparing the depth deviation between the perforation clusters and the formation characteristic interface by combining gamma logging data, if there is a deviation, translating the position of the cluster, or re-anchoring the depth range of the cluster based on the absolute depth markers of the casing collars to eliminate errors; outputting the distribution of the depth-corrected perforation clusters, including the segmentation number, start and end depths, and cluster details, to ensure accurate positioning of the target reservoir for the fracturing operation.

[0012] The beneficial effects of the present invention are: dynamically characterizing the reservoir heterogeneity based on the mechanical specific energy; optimizing the variance of the mechanical specific energy of the perforation clusters within the fracturing section through the genetic algorithm, and combining constraints such as the section length range, the number of clusters limit, and non-fractured well sections to achieve uniform segmentation of the reservoir physical properties. The present invention is of great significance for improving the fracturing effect of horizontal wells and the oil and gas production of single wells. Description of the Drawings

[0013] Figure 1Schematic flow chart of the method for segmenting and clustering horizontal wells in volume fracturing based on mechanical specific energy and genetic algorithm; Figure 2 Implementation flow chart of the method for segmenting and clustering horizontal wells in volume fracturing based on mechanical specific energy and genetic algorithm. Specific implementation mode

[0014] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings, but the protection scope of the present invention is not limited to the following description.

[0015] The features and performance of the present invention will be further described in detail below with reference to the embodiments.

[0016] As Figure 1 shown, the method for segmenting and clustering horizontal wells in volume fracturing based on mechanical specific energy and genetic algorithm includes the following steps: Step 1, collect drilling dynamic data, form a drilling dynamic data set table at equal well depth intervals according to the well depth step length, and calculate the bit torque and weight on bit; Step 2, calculate the mechanical specific energy of the horizontal well section according to the drilling dynamic data set table, and obtain a set of mechanical specific energy values along the wellbore; Step 3, generate an initial population based on the set constraints, and obtain the fitness of each individual in the initial population; Step 4, randomly select a set number of individuals from the initial population to form a candidate set, and use the tournament selection method to screen the individuals with high fitness in the candidate set; generate offspring through the segment boundary replacement strategy and perform crossover operations; randomly adjust the segment boundary depth and reorder, and perform mutation operations; repeat the fitness calculation, selection, crossover, and mutation operations until the maximum number of iterations is reached to complete the iterative optimization; Step 5, output the optimal segmentation scheme and perforation cluster distribution, and perform drilling construction according to the optimal segmentation scheme and perforation cluster distribution.

[0017] The formation of the drilling dynamic data set table at equal well depth intervals according to the well depth step length and the calculation of the bit torque and weight on bit include: Based on the interval from the starting well depth to the ending well depth of the horizontal well section, take values of the drilling dynamic data at equal well depth intervals within the interval to form a drilling dynamic data set table at equal well depth intervals; calculate the bit torque and weight on bit corresponding to each record based on the drilling dynamic data set table: Calculate the bit torque and weight on bit for the well sections with and without the use of the screw rod respectively; When using the screw section, the bit torque is TBit = Kt*ΔP, where Kt is the torque output per unit differential pressure of the screw and ΔP is the differential pressure of the screw in the drilling state; where ΔP is estimated using the standpipe pressure, which is the standpipe pressure when the bit is drilling at the bottom of the well minus the standpipe pressure when the bit is lifted off the bottom of the well; the weight on bit is estimated by measuring the hook load, which is the hook load when the bit is lifted off the bottom of the well minus the hook load when the bit is drilling at the bottom of the well. When not using the screw section, the bit torque is predicted using the friction torque model. The rotary table torque is used as the input value of the friction torque model to calculate from the wellhead to the bottom of the well to predict the bit torque. The method for estimating the weight on bit is the same as when using the screw.

[0018] Calculating the mechanical specific energy of the horizontal section according to the drilling dynamic data set table to obtain the set of mechanical specific energy values along the wellbore, including: Based on the drilling dynamic data, calculate the mechanical specific energy at a set well depth step within the horizontal section interval to form a mechanical specific energy data set. The mechanical specific energy is calculated using the following formula: where Dbit is the bit diameter, ROP is the rate of penetration, Kn is the output rotational speed per unit mud flow rate of the screw, Q is the mud flow rate, TBit is the bit torque, N is the rotary table speed. If there is no screw, Kn is taken as 0.

[0019] Generating an initial population based on the set constraint conditions, including: The constraint conditions include the number of fracturing stages, the range of stage lengths, the range of the number of clusters per stage, the length of each cluster, and the non-fracturing section; generating the initial population includes: Step 31: Set the total number of individuals in the population. Each individual represents a fracturing segmentation scheme, which consists of fracturing stage number - 1 fracturing boundary depth values, and the boundary values are arranged in ascending order. Step 32: Randomly generate individuals. Starting from the starting depth, generate fracturing boundaries at random steps, and the step range is within the stage length range. Step 33: Repeat Step 31 and Step 32 until the number of individuals in the initial population is the total number of individuals set in the population.

[0020] Obtaining the fitness of each individual in the initial population, including: Validity check of the fracturing boundary of the individual. First, check the number of segments to verify whether the number of fracturing boundaries of the individual is fracturing stage number - 1; if not, directly return -Infinity. Then, check the stage length constraint: traverse all fracturing stages and calculate whether the length of each stage satisfies minStageLength ≤ stage length ≤ maxStageLength; record the number of violations stageViolations, and the calculation method of the number of violations is: ; For the division and verification of individual perforation clusters, first, perforation clusters are generated. For each fracturing stage, perforation clusters are divided according to the length of each cluster; and the length of each cluster is greater than half of the length of each cluster; then, the cluster number constraint check is carried out. The number of perforation clusters in each stage is counted. If it exceeds the range of the number of clusters in each stage, the number of violations is recorded; Fitness calculation is performed. First, data mapping is carried out. The well depth range of the perforation clusters is mapped to the mechanical specific energy dataset. For each perforation cluster, the mechanical specific energy values within its corresponding well depth range are extracted, and the cluster variance is calculated; Cluster variance = Variance(MSEs, MSEs + 1, …, MSEe) Where s and e are the data indices corresponding to the starting and ending depths of the cluster respectively. If the cluster length is invalid, the variance is set to 0. Finally, the cluster variances of all perforation clusters are accumulated and negated as the fitness; if there are any stage or cluster violations, -Infinity is returned to impose a penalty. The final fitness formula is: .

[0021] Randomly selecting a set number of individuals from the initial population to form a candidate set and using the tournament selection method to screen out the individuals with high fitness in the candidate set includes: Step 41: Randomly select a set number of individuals from the population to form a candidate set; compare the fitness values of the candidate individuals and select the individual with the highest fitness as the winner; Step 42: Repeat Step 41 until the size of the new population reaches the total number of individuals within the set population.

[0022] Generating offspring through the segmented boundary replacement strategy and performing crossover operations includes: Randomly select two parent individuals parent1 and parent2 from the population. For their segmented boundary lists, traverse each segmented boundary position and generate an offspring child by selecting the corresponding boundary value of parent1 or parent2 with a set probability; Reorder the boundary values of the offspring to ensure monotonic increase and constrain the boundary values of child within the valid interval.

[0023] Randomly adjusting the segmented boundary depth and reordering to perform mutation operations includes: Each individual in the population triggers the mutation operation with a set probability. Randomly select a certain position in the individual's segmented boundary list, apply a random perturbation within a certain range to the boundary value corresponding to this position to generate a new value, and constrain it to the valid interval; Replace the original boundary values and reorder the entire list. If the perforation clusters generated after mutation conflict with non-fractured well sections, a penalty is imposed in the fitness calculation.

[0024] The output of the optimal segmentation scheme and perforation cluster distribution includes: Divide the perforation clusters from the optimal solution of the genetic algorithm according to the length of each cluster, ensure that the number of clusters is within the range of the number of clusters in each section, and avoid non-fractured well sections. Perform depth calibration, including comparing the depth deviation between the perforation clusters and the formation characteristic interface by combining gamma logging data. If there is a deviation, translate the cluster position, or re-anchor the well depth range of the cluster based on the absolute depth marker of the casing collar data to eliminate errors. Output the perforation cluster distribution after depth calibration, including the segmentation number, start and end depths, and cluster details, to ensure accurate positioning of the target reservoir for the fracturing operation.

[0025] As Figure 2 shown, the technical solution adopted by the present invention is: a volumetric fracturing horizontal well segmentation method based on mechanical specific energy and genetic algorithm, and the specific steps are as follows: First, collect drilling dynamic data to form a drilling dynamic data set table at equal well depth intervals, and calculate the bit torque and weight on bit. Next, calculate the mechanical specific energy of the horizontal well section to obtain a set of mechanical specific energy values along the wellbore. Then, define the constraint conditions such as the number of fracturing sections, section length range, number of clusters in each section range, and non-fractured well sections of the horizontal well section. Subsequently, initialize the population to generate an initial candidate solution that satisfies the section length constraint. After that, perform fitness calculation, use the variance of the mechanical specific energy corresponding to the perforation clusters in each section as the fitness function, and impose a penalty on the solutions that violate the constraints. Next, perform the selection operation, use the tournament selection method to screen individuals with high fitness; generate offspring through the segmentation boundary replacement strategy and perform the crossover operation; randomly adjust the segmentation boundary depth and reorder, and perform the mutation operation; repeat the fitness calculation, selection, crossover, and mutation operations until the maximum number of iterations is reached to complete the iterative optimization. Finally, output the optimal segmentation scheme and perforation cluster distribution, and perform depth calibration to dynamically adjust the well section range deviation. The analysis process is as Figure 1 shown.

[0026] The present invention needs to record the relevant drilling dynamic data of the construction well, including but not limited to the hook load (HKLoad), rate of penetration (ROP), mud flow rate (Q), rotary table speed (N), rotary table torque (TSurface), bit diameter (DBit), and standpipe pressure (SPP).

[0027] Organize the drilling dynamic data of the horizontal well section. Based on the interval from the starting well depth (totalStart) to the ending well depth (totalEnd) of the horizontal well section, take values of the drilling dynamic data at a certain well depth step within the interval to form a table of drilling dynamic data set at equal well depth intervals.

[0028] Calculate the bit torque (TBit) and weight on bit (WOB) corresponding to each record based on the table of drilling dynamic data set: Analyze whether a positive displacement motor (PDM) is used during the drilling of the construction well in the entire horizontal well section. Calculate the bit torque (TBit) and weight on bit (WOB) respectively for the well sections with and without the use of a PDM.

[0029] For the well section with a PDM, the bit torque is TBit = Kt*ΔP, where Kt is the output torque per unit differential pressure of the PDM, ΔP is the differential pressure of the PDM in the drilling state, and ΔP is estimated using the standpipe pressure (SPP), which is the standpipe pressure when the bit is drilling at the bottom minus the standpipe pressure when the bit is lifted off the bottom. The weight on bit (WOB) is estimated through the measured hook load (HKLoad), which is the hook load when the bit is lifted off the bottom minus the hook load when the bit is drilling at the bottom.

[0030] For the well section without a PDM, the bit torque (TBit) is predicted using a friction torque model. The rotary table torque (TSurface) is used as the input value of the friction torque model to calculate the bit torque from the wellhead to the bottom. The estimation method of the weight on bit (WOB) is the same as that when using a PDM.

[0031] Based on the drilling dynamic data, calculate the mechanical specific energy (MSE) at a certain well depth step within the horizontal well section interval to form a mechanical specific energy data set (mseData). The mechanical specific energy is calculated using the following formula: where Dbit is the bit diameter, ROP is the rate of penetration, Kn is the output rotational speed per unit mud flow rate of the PDM, Q is the mud flow rate, TBit is the bit torque, N is the rotary table speed, and if there is no PDM, Kn is taken as 0.

[0032] Set the constraint conditions for the genetic algorithm, including the number of fracturing stages (numStages), the range of stage lengths (minimum stage length, maximum stage length) (minStageLength, maxStageLength), the range of the number of clusters per stage (minimum number of clusters, maximum number of clusters) (minClustersPerStage, maxClustersPerStage), the length of each cluster (clusterLength), and the non-fracturing well sections (avoidZones).

[0033] Generate an initial population that satisfies the stage length constraints: Step 1: Set the total number of individuals in the population (populationSize). Each individual represents a segmentation scheme, which consists of numStages - 1 segment boundary depth values, and the boundary values are arranged in ascending order.

[0034] Step 2: Randomly generate individuals. Starting from the starting depth, generate segment boundaries with random step lengths. The step length range is within [minStageLength, maxStageLength]; ensure that the last segment boundary does not exceed totalEnd - minStageLength to prevent the length of the last segment from being insufficient.

[0035] Step 3: Repeat Step 1 and Step 2 until the number of individuals in the initial population is populationSize.

[0036] Calculate the fitness of the individuals in the population. Fitness is used to evaluate the quality of individual segmentation. In the present invention, it is achieved by minimizing the variance of the mechanical specific energy (MSE) corresponding to the perforation clusters within each segment to ensure the uniformity of reservoir characteristics within the same fracturing segment. The specific steps are as follows: Verify the legality of the segment boundaries of the individual. First, check the number of segments to verify whether the number of segment boundaries of the individual is numStages - 1. If not, directly return -Infinity (indicating that the scheme is completely infeasible); then perform a segment length constraint check: traverse all fracturing segments and calculate whether the length of each segment satisfies minStageLength ≤ segment length ≤ maxStageLength. Record the number of violations stageViolations. The calculation method of the number of violations is: Divide and verify the perforation clusters of the individual. First, generate perforation clusters. For each fracturing segment, divide the perforation clusters according to the clusterLength. And ensure that the clusters do not overlap with non-fracturing well sections (avoidZones), and the length of each cluster is at least clusterLength / 2 to avoid invalid short clusters; then perform a cluster number constraint check, count the number of perforation clusters in each segment, and if it exceeds the range of [minClustersPerStage, maxClustersPerStage], record the number of violations (clusterViolations).

[0037] Perform fitness calculation. First, perform data mapping to map the well depth range of the perforation clusters to the mechanical specific energy dataset mseData. For each perforation cluster, extract the mechanical specific energy values within its corresponding well depth range and calculate the variance. The cluster variance = Variance(MSEs, MSEs+1, …, MSEe), where s and e are the data indices corresponding to the starting and ending depths of the cluster respectively. If the cluster length is invalid (e.g., completely located in a non-fracturing well section), the variance is set to 0. Finally, sum up the cluster variances of all perforation clusters and take the negative value as the fitness (the optimization objective is to minimize the variance). If there are any stage or cluster violations (stageViolations>0 or clusterViolations>0), return -Infinity to impose a penalty. The final fitness formula is as follows: Perform the selection operation. The present invention is implemented based on the tournament selection method to screen out individuals with high fitness from the current population and enter the next generation, driving the population to evolve towards a better solution: First, randomly select a certain number (tournamentSize) of individuals from the population to form a candidate set; then compare the fitness values of the candidate individuals and select the individual with the highest fitness as the winner. For example, if the fitness values of the individuals in the candidate set are [-10, -Infinity, -5, -20, -Infinity], then select the individual with a fitness of -5.

[0038] Repeat the execution until the size of the new population reaches the original population size (populationSize). During this process, individuals with a fitness of -Infinity cannot win and are naturally eliminated.

[0039] Perform the crossover operation. Generate offspring individuals by fusing the section boundary information of two parent individuals. The present invention adopts a section boundary replacement strategy. The crossover operation needs to ensure that the generated offspring individuals meet the section length constraints (minStageLength and maxStageLength) and the boundary monotonically increasing property, and at the same time be compatible with the correction ability of subsequent mutation operations. The specific process is as follows: Randomly select two parent individuals parent1 and parent2 from the population. For their section boundary lists (e.g., parent1 = [1550, 1650, 1750, 1850], parent2 = [1570, 1670, 1770, 1870]), traverse each section boundary position and select the corresponding boundary value of parent1 or parent2 with a certain probability (e.g., 50%) to generate the offspring child (e.g., it may generate child = [1550, 1670, 1750, 1870]); Subsequently, reorder the offspring boundary values to ensure monotonic increase, and constrain the boundary values of the child within the valid interval [totalStart + minStageLength, totalEnd - minStageLength] (for example, when totalEnd = 1900 and minStageLength = 50, if the boundary of the last segment exceeds 1850, it is forced to be corrected to 1850) to avoid exceeding the segment length limit.

[0040] Perform the mutation operation. On the premise of ensuring the legality of the individual, fine-tune the segmented boundaries to explore the neighboring solution space. The present invention adopts a random boundary offset strategy, combined with the boundary sorting and constraint correction mechanism, to ensure that the mutated individual still meets the requirements of the segment length range, non-fracturing well segments, and boundary monotonicity. The specific steps are as follows: Each individual in the population triggers the mutation operation with a certain probability (mutationRate), randomly selects a position (pos) in the individual's segmented boundary list, applies a random perturbation within a certain range to the boundary value corresponding to this position to generate a new value, and constrains it to the valid interval [totalStart + minStageLength, totalEnd - minStageLength]; Subsequently, replace the original boundary value and reorder the entire list to ensure strict increase of the boundaries. If the perforation cluster generated after mutation conflicts with the avoidZones, a penalty is imposed in the fitness calculation (return -Infinity) to prompt the subsequent evolution to eliminate this individual.

[0041] Perform iterative optimization (that is, repeat the previous fitness calculation, selection, crossover, and mutation operations until the maximum number of iterations is reached or the convergence condition is met) to obtain the final segmented and clustered result: Calculate the fitness values for all individuals in the current population, and perform selection, crossover, and mutation operations until the maximum number of iterations (maxGenerations) is reached or the convergence condition is met. Update the population after each iteration, record the historical optimal solution (bestSolution), and enter the next generation of iteration.

[0042] Finally, output the segmented boundaries corresponding to bestSolution and the division of perforation clusters in each segment, ensuring that the scheme meets the segment length range, cluster number limit, and the constraint of avoiding non-fracturing well segments (avoidZones).

[0043] Depth correction of the optimal segmented and clustered scheme obtained by the genetic algorithm is performed based on the actual engineering measurement data to ensure the accuracy of the fracturing construction and obtain the final scheme: Divide the perforation clusters from the optimal solution bestSolution of the genetic algorithm by the cluster length, ensuring that the number of clusters is within the range of [minClustersPerStage, maxClustersPerStage], and avoiding the non-fracturing well sections (avoidZones).

[0044] Perform depth calibration. For example, compare the depth deviation between the perforation clusters and the formation characteristic interface by combining gamma logging data. If there is a deviation, translate the cluster position, or re-anchor the cluster well depth range based on the absolute depth markers of the casing collars to eliminate the error.

[0045] Output the distribution of the perforation clusters after depth calibration, including the section number, start and end depths, and cluster details, ensuring accurate positioning of the target reservoir for the fracturing operation.

[0046] The volume fracturing horizontal well sectioning method based on mechanical specific energy and genetic algorithm of the present invention comprises the following specific steps: First, collect drilling dynamic data to form a table of drilling dynamic data sets at equal well depth intervals, and calculate the bit torque and weight on bit; Next, calculate the mechanical specific energy of the horizontal well section to obtain a set of mechanical specific energy values along the wellbore; Then, define the constraint conditions such as the number of fracturing sections, section length range, number of clusters per section range, and non-fracturing well sections of the horizontal well section; Subsequently, generate an initial candidate solution that meets the section length constraint and initialize the population; After that, calculate the variance of the mechanical specific energy corresponding to the perforation clusters in each section as the fitness function, and impose penalties on the solutions that violate the constraints; Then, use the tournament selection method to screen out individuals with high fitness and perform the selection operation; Generate offspring through the section boundary replacement strategy and perform the crossover operation; Randomly adjust the section boundary depth and reorder it to perform the mutation operation; Repeat the fitness calculation, selection, crossover, and mutation operations until the maximum number of iterations is reached to complete the iterative optimization; Finally, output the optimal sectioning plan and the distribution of perforation clusters, and perform depth calibration to dynamically adjust the well section range.

[0047] The embodiment provides a volume fracturing horizontal well sectioning method based on mechanical specific energy and genetic algorithm, and the specific steps are as follows: Step 71: Record the drilling dynamic data of the construction well, including but not limited to the hook load (HKLoad), rate of penetration (ROP), mud flow rate (Q), rotary table speed (N), rotary table torque (TSurface), bit diameter (DBit), and standpipe pressure (SPP).

[0048] Step 72: Based on the interval from the starting well depth (totalStart) to the ending well depth (totalEnd) of the horizontal well section, take values of the drilling dynamic data at a certain well depth step within the interval to form a table of drilling dynamic data sets at equal well depth intervals.

[0049] Step 73: Calculate the bit torque (TBit) and weight on bit (WOB) corresponding to each record in the set table.

[0050] (1) For the positive displacement motor section, the bit torque TBit = Kt * ΔP, where Kt is the torque output per unit differential pressure of the positive displacement motor, and ΔP is the differential pressure of the positive displacement motor in the drilling state. ΔP can be estimated using the standpipe pressure (SPP), which is the standpipe pressure when the bit is drilling at the bottom hole minus the standpipe pressure when the bit is lifted off the bottom hole. The weight on bit (WOB) can be estimated by measuring the hook load (HKLoad), which is the hook load when the bit is lifted off the bottom hole minus the hook load when the bit is drilling at the bottom hole.

[0051] (2) For the section without using the positive displacement motor, the bit torque (TBit) is predicted using the friction torque model. The rotary table torque (TSurface) is used as the input value of the friction torque model to calculate from the wellhead to the bottom hole to predict the bit torque. The method for estimating the weight on bit (WOB) is the same as that when using the positive displacement motor.

[0052] Step 74: Based on the drilling dynamic data, calculate the mechanical specific energy (MSE) at a certain well depth step within the horizontal well section interval to form a mechanical specific energy data set (mseData).

[0053] Step 75: Set the constraint conditions for the genetic algorithm, including the number of fracturing stages (numStages), the range of stage lengths (minimum stage length, maximum stage length) (minStageLength, maxStageLength), the range of the number of clusters per stage (minimum number of clusters, maximum number of clusters) (minClustersPerStage, maxClustersPerStage), the length of each cluster (clusterLength), and the non-fracturing well sections (avoidZones).

[0054] Step 76: Generate an initial population that meets the stage length constraints.

[0055] (1) Set the total number of individuals in the population (populationSize) and the individual generation rule. Each individual represents a segmentation scheme, which consists of numStages - 1 segment boundary depth values, and the boundary values are arranged in ascending order.

[0056] (2) Randomly generate individuals. Starting from the starting depth, generate segment boundaries with random steps, and the step range is within [minStageLength, maxStageLength]; ensure that the last segment boundary does not exceed totalEnd - minStageLength to prevent the length of the last segment from being insufficient; if the number of segments requirement cannot be met during the generation process, supplement the segment boundaries to numStages - 1 to ensure the legality of the individuals.

[0057] (3) Repeat steps 71 and 72 until the number of individuals in the initial population is populationSize.

[0058] Step 77: Calculate the fitness of the individuals in the population.

[0059] Step 78: Perform a selection operation on the population.

[0060] Step 79: Perform a crossover operation on the population.

[0061] Step 710: Perform a mutation operation on the population.

[0062] Step 711: Repeat steps 77 to 710 until the number of repetitions reaches the set limit. Record the historical best solution (bestSolution) in each iteration and then enter the next generation iteration. After the iteration ends, output the final best solution.

[0063] Step 712: Depth correction of the optimal segmentation and clustering scheme obtained by the genetic algorithm based on the actual measured data of the project to ensure the accuracy of the fracturing construction and obtain the final scheme.

[0064] The following further elaborates on the embodiments in combination with specific examples.

[0065] First step: Record the drilling dynamic data of the wells that have been constructed, including the hook load (HKLoad), rate of penetration (ROP), mud flow rate (Q), rotary table speed (N), rotary table torque (TSurface), bit diameter (DBit), and standpipe pressure (SPP).

[0066] Second step, organize the drilling dynamic data into equal intervals. As shown in Table 1, it is an example display of the drilling dynamic data corresponding to each meter of well depth interval from 3000m to 4000m.

[0067] Table 1 Parameters of the wells that have been constructed Third step: Calculate the bit torque (TBit) and weight on bit (WOB) according to the data in Table 1.

[0068] (1) Assume that the section without a positive displacement motor (3000 - 3300 meters) is not used.

[0069] Taking a well depth of 3000m as an example: 1) Calculation of TBit. Assume that the calculation result of the friction torque model is 85% of TSurface, then TBit = 15500 × 0.85 = 13175 N·m; 2) WOB calculation: WOB = 1100 - 1050 = 50 kN (assuming HKLoad = 1100 kN when the bit is lifted off the bottom).

[0070] (2) Assume using the positive displacement motor section (3300 - 4000 m).

[0071] Taking a well depth of 3300 m as an example: 1) ΔP calculation: ΔP = SPP during drilling - SPP when lifted off = 25 - 20 = 5 MPa (assuming SPP = 20 MPa when lifted off); 2) TBit calculation (positive displacement motor unit pressure difference torque coefficient Kt = 500 N·m / MPa): TBit = Kt × ΔP = 500 × 5 = 2500 N·m; 3) WOB calculation: WOB = 1300 - 1250 = 50 kN (assuming HKLoad = 1300 kN when lifted off).

[0072] Fourth step, calculate the mechanical specific energy, which is calculated for the non - positive displacement motor section and the positive displacement motor section to calculate the mechanical specific energy (MSE), and the formula used is: Assume the obtained results are as shown in the following table: Table 2 Example table of mechanical specific energy data set Fifth step: Set the constraint conditions of the genetic algorithm. To simplify the calculation process in the example, the number of clusters in the section length constraint condition is not considered: numStages = 5; minStageLength = 150; maxStageLength = 300; clusterLength = 10; avoidZones = {{3050, 3060}}; totalStart = 3000; totalEnd = 4000.

[0073] Sixth step: Generate the initial population. Assume the initial population individuals (populationSize) are 4, including the following individuals: (1) Individual A: {3150, 3450, 3750, 3850} (2) Individual B: {3300, 3550, 3800, 3900} (3) Individual C: {3050, 3200, 3400, 3600} (4) Individual D: {3100, 3150, 3500, 3800} Seventh step, fitness calculation: (1)Calculation of the fitness of individual A: 1) Segmented intervals: Segment 1: 3000 - 3150 meters (150 meters, legal), Segment 2: 3150 - 3450 meters (300 meters, legal), Segment 3: 3450 - 3750 meters (300 meters, legal), Segment 4: 3750 - 3850 meters (100 meters, legal), Segment 5: 3850 - 4000 meters (150 meters, legal).

[0074] 2) Generation of perforation clusters, generated according to the cluster length. For example, for the interval of Segment 2 (3150 - 3450 meters), the clusters are {{3150, 3160}, {3160, 3170},..., {3440, 3450}}. Cluster legality check: Avoid non-fracturing segments {3050, 3060}, legal.

[0075] 3) Variance calculation: Assume the cluster is 3150 - 3160 meters. According to the mechanical specific energy in Table 2, the variance is obtained as 24.8. For the cluster 3450 - 3460 meters, the variance is 18.5. Calculate in sequence, then the total fitness: -Σ(24.8 + 18.5 +...) = -320.5.

[0076] (2)Calculation of the fitness of individual B: 1) Segmented intervals: Segment 1: 3000 - 3300 meters (300 meters, legal), Segment 2: 3300 - 3550 meters (250 meters, legal), Segment 3: 3550 - 3800 meters (250 meters, legal), Segment 4: 3800 - 3900 meters (100 meters, legal), Segment 5: 3900 - 4000 meters (100 meters, legal).

[0077] 2) Generation of perforation clusters, generated according to the cluster length. For example, for the interval of Segment 3 (3550 - 3800 meters), the clusters are {{3550, 3560}, {3560, 3570},..., {3790, 3800}}. Cluster legality check: Avoid non-fracturing segments {3050, 3060}, legal.

[0078] 3) Variance calculation: Assume the cluster is 3550 - 3560 meters. According to the mechanical specific energy in Table 2, the variance is 20.1. Calculate in sequence, then the total fitness: -Σ(20.1 +...) = -280.6.

[0079] (3)Calculation of the fitness of individual C Segment 1: 3000 - 3050 meters (50 meters, illegal), fitness - Infinity.

[0080] (4)Calculation of the fitness of individual D Segment 2: 3100 - 3150 m (50 m, violation), fitness - Infinity.

[0081] Step 8, selection operation, tournament rule, generate 4 new individuals: (1) First tournament: select A and B → B wins (higher fitness) (2) Second tournament: select C and D → randomly select D (both are - Infinity) (3) Third tournament: select B and D → B wins (4) Fourth tournament: select A and C → A wins The new population obtained is: B, D, B, A.

[0082] Step 9, crossover operation, perform parental selection in the population from the previous step. For example, select B{3300, 3550, 3800, 3900} and A{3150, 3450, 3750, 3850}, and adopt a crossover strategy of 50% probability for each bit to generate offspring E: (1) Bit 1: select 3300 of B; bit 2: select 3450 of A; bit 3: select 3800 of B; bit 4: select 3850 of A. Finally, obtain offspring E: {3300, 3450, 3800, 3850}.

[0083] (2) Segment length check: Segment 3: 3450 - 3800 m (350 m, violation) → needs mutation correction.

[0084] Step 10, mutation operation, adjust the generated offspring. For example, adjust offspring E as follows: (1) Perturbation boundary: The 3rd boundary 3800 → 3750 (segment length = 300 m, legal).

[0085] (2) The corrected offspring E: {3300, 3450, 3750, 3850}.

[0086] Step 11, record the optimal result after the mutation operation, then start iterative optimization, record the optimal result after each iteration. Assume that it is set to iterate 15 times. The optimal result selected from all the results after the iteration is: {3300, 3550, 3750, 3900}, and the clusters are arranged according to a segment length of 10 m.

[0087] Step 12, depth correction. According to the logging data, the boundary at 3550 m is corrected to 3552 m. Then the final segmentation result is {3300, 3552, 3800, 3900}.

Claims

1. A method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm, characterized in that: The steps include: Step 1: Collect drilling dynamic data, form a set table of drilling dynamic data at equal well depth intervals according to the well depth step length, and calculate the drill bit torque and drilling pressure; Step 2: Calculate the mechanical specific energy of the horizontal well section according to the drilling dynamic data set table to obtain a set of mechanical specific energy values ​​along the wellbore; Step 3: Based on the set constraints, generate the initial population and obtain the fitness of each individual in the initial population; Step 4: randomly select a set number of individuals from the initial population to form a candidate set, and use the tournament selection method to select individuals with high fitness in the candidate set; Generate offspring through segment boundary replacement strategy and perform crossover operation; randomly adjust segment boundary depth and reorder to perform mutation operation; repeat fitness calculation, selection, crossover and mutation operation until the maximum number of iterations is reached to complete iterative optimization; Step 5: Output the optimal segmentation scheme and perforation cluster distribution, and perform drilling construction according to the optimal segmentation scheme and perforation cluster distribution.

2. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The method of forming a drilling dynamic data set table at equal well depth intervals according to the well depth step length and calculating the drill bit torque and drilling pressure includes: Based on the interval from the starting well depth to the ending well depth of the horizontal well section, the drilling dynamic data is taken according to the set well depth step in the interval to form a drilling dynamic data set table at equal well depth intervals; based on the drilling dynamic data set table, the drill bit torque and drilling pressure corresponding to each record are calculated: The drill bit torque and bit pressure are calculated for the well sections with and without screw rods. When using the screw section, the drill bit torque is TBit = Kt*ΔP, where Kt is the screw output torque per unit pressure difference, and ΔP is the screw pressure difference when drilling. ΔP is estimated using the riser pressure, which is the riser pressure when the drill bit is drilling at the bottom of the well minus the riser pressure when the drill bit is lifted off the bottom of the well. The drilling pressure is estimated by measuring the hook load, which is the hook load when the drill bit is lifted off the bottom of the well minus the hook load when the drill bit is drilling at the bottom of the well. In the well section where the screw rod is not used, the drill bit torque is predicted using the friction torque model. The turntable torque is used as the input value of the friction torque model and is calculated from the wellhead to the bottom of the well to predict the drill bit torque. The drilling pressure estimation method is the same as when the screw rod is used.

3. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The method of calculating the mechanical specific energy of the horizontal well section according to the drilling dynamic data set table to obtain a set of mechanical specific energy values ​​along the wellbore includes: Based on the drilling dynamic data, the mechanical specific energy is calculated according to the set well depth step in the horizontal well section to form a mechanical specific energy data set. The mechanical specific energy is calculated using the following formula: Where Dbit is the drill bit diameter, ROP is the drilling speed, Kn is the output speed of the screw under unit mud flow, Q is the mud flow, TBit is the drill bit torque, N is the rotary table speed, and if there is no screw, Kn is 0.

4. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The generating of the initial population based on the set constraints includes: The constraints include the number of fracturing stages, the range of stage lengths, the range of the number of clusters per stage, the length of each cluster, and the non-fracturing well sections; the generation of the initial population includes: Step 31: Set the total number of individuals in the population, each individual represents a segmentation scheme, which consists of the number of fracturing stages minus 1 segment boundary depth value, and the boundary values ​​are arranged in ascending order; Step 32: Randomly generate individuals, starting from the starting depth, and generate segment boundaries according to random step lengths, and the step length range is within the segment length range; Step 33: Repeat steps 31 and 32 until the number of individuals in the initial population is equal to the total number of individuals in the set population.

5. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The step of obtaining the fitness of each individual in the initial population includes: The legitimacy check of the individual segment boundary is first performed to check the number of segments to verify whether the number of the individual segment boundary is the number of fracturing stages - 1; if not, directly return - Infinity; then perform the segment length constraint check: traverse all fracturing stages, calculate whether the length of each stage satisfies minStageLength ≤ stage length ≤ maxStageLength; record the number of violations stageViolations, the number of violations is calculated as follows: Individual perforation cluster division and verification: First, perforation cluster generation is performed. For each fracturing stage, perforation clusters are divided according to the length of each cluster; and the length of each cluster is greater than the length of each cluster / 2; then cluster number constraint check is performed to count the number of perforation clusters in each stage. If the number of clusters in each stage exceeds the range, the number of violations is recorded; To calculate the fitness, firstly, data mapping is performed to map the well depth range of the perforation cluster to the mechanical specific energy data set. For each perforation cluster, the mechanical specific energy value within the corresponding well depth range is extracted and the cluster variance is calculated. Cluster variance = Variance(MSEs,MSEs+1,…,MSEe) Among them, s and e are the data indexes corresponding to the starting and ending depths of the cluster, respectively. If the cluster length is invalid, the variance is set to 0. Finally, the cluster variances of all perforation clusters are accumulated and the negative is taken as the fitness. If there is any segment or cluster violation, -Infinity is returned to impose a penalty. The final fitness formula is: 。 6. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The method of randomly extracting a set number of individuals from the initial population to form a candidate set, and using the tournament selection method to screen individuals with high fitness in the candidate set includes: Step 41: randomly extract a set number of individuals from the population to form a candidate set; compare the fitness values ​​of the candidate individuals, and select the individual with the highest fitness as the winner; Step 42: Repeat step 41 until the size of the new population reaches the total number of individuals in the set population.

7. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The generation of offspring by segment boundary replacement strategy and the crossover operation include: Randomly select two parent individuals parent1 and parent2 from the population, segment their boundary lists, traverse each segment boundary position, and select the corresponding boundary value of parent1 or parent2 with a set probability to generate the child child; reorder the child boundary values ​​to ensure monotonically increasing, and constrain the boundary value of child to be within a valid interval.

8. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The random adjustment of the segment boundary depth and reordering to perform a mutation operation includes: each individual in the population triggers a mutation operation with a set probability, randomly selects a position in the individual segment boundary list, applies a random perturbation within a set range to the boundary value corresponding to the position to generate a new value, and constrains it to a valid interval; replaces the original boundary value and reorders the entire list, and if the perforation cluster generated after the mutation conflicts with the non-fracturing well section, a penalty is imposed in the fitness calculation.

9. The method for segmenting and clustering volume fracturing horizontal wells based on mechanical specific energy and genetic algorithm according to claim 1, characterized in that: The output optimal segmentation scheme and perforation cluster distribution include: dividing the perforation clusters according to the length of each cluster from the optimal solution of the genetic algorithm, ensuring that the number of clusters is within the range of the number of clusters in each section and avoiding non-fractured well sections; performing depth calibration, including comparing the depth deviation between the perforation cluster and the formation characteristic interface in combination with the gamma logging data, and if there is a deviation, translating the cluster position, or re-anchoring the cluster well depth range based on the absolute depth mark of the casing coupling data to eliminate the error; outputting the perforation cluster distribution after depth calibration, including the segment number, the start and end depths, and the cluster details, to ensure that the fracturing operation accurately locates the target reservoir.

Images