Deep drilling mudstone expansion jam prediction and through-hole operation scheduling optimization method

Abstract

This invention relates to a method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling. The method divides boreholes into sections based on depth and assigns a risk budget item to each section, including at least one of the following: the number of precursor windows, the duration of a phase, and the number of invertible events. Within a sliding time window, at least two of the following parameters are collected: torque, pump pressure, returned material parameters, and mechanical drilling rate. These parameters are compared with a baseline to form primary evidence, which is then combined with exclusionary evidence to identify precursor windows. When a precursor window occurs consecutively for a set number of times, the blockage phase is advanced and a predetermined exploratory action is performed. If the pump pressure or torque does not recover to the baseline's allowable range within a set time limit, it is determined to be an invertible event. The corresponding budget is deducted based on the number of precursor windows, phase duration, and number of invertible events. When any budget is exhausted, a borehole operation deadline and urgency level for that section are generated. The borehole operation needs of multiple sections are prioritized by deadline. When the difference in deadlines is not greater than a threshold, a scheduling plan is generated by combining urgency level and scheduling constraints.

Description

Technical Field

[0001] This invention relates to the field of drilling engineering technology, specifically to a method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling. Background Technology

[0002] Current technologies, such as the one published in CN113496302A ("A Method and System for Intelligent Identification and Early Warning of Drilling Risks"), have improved the accuracy of risk identification to some extent. However, they remain at the level of identification and early warning, failing to form a closed-loop control mechanism deeply coupled with subsequent engineering actions. Their limitation lies in the disconnect between risk modeling and the actual downhole physical evolution process. This method primarily relies on historical well logging data to construct a risk case database and uses a random forest machine learning model to train and identify data of different time periods to achieve real-time early warning of target well risks. However, this method is essentially a data-driven classification framework, its core being the transformation of the risk identification problem into a model classification problem. It does not characterize the stage evolution, cumulative effects, and reversibility characteristics of risks, nor does it provide a continuous quantitative expression of the risk development path. In actual mudstone expansion and blockage processes, risks typically accumulate gradually and alternate between locally recoverable and irrecoverable states. However, this method only outputs the existence and type of risk within a certain time window, making it difficult to reflect the dynamic progression of risk from precursors to critical blockage. As a result, the early warning results remain more of a static judgment of whether it has occurred, and cannot support refined intervention decisions.

[0003] While this technology incorporates multi-timescale data input and risk verification mechanisms—for example, constructing multiple identification models using logging data of different time lengths and verifying them in conjunction with trends or rates of change—its verification logic still focuses on reducing false alarms and false negatives, without extending further to assessing risk recoverability or evaluating intervention effectiveness. After identifying a risk, the method only confirms its existence through parametric trend analysis and issues a warning to on-site personnel, but lacks a tiered response mechanism for different risk states, and doesn't address determining whether the well section's condition is reversible or trending towards blockage through exploratory operations. This means that in practical applications, engineers still need to rely on experience to decide whether to perform perforation, circulation, or other treatments; the system itself cannot provide structured operational suggestions, and there remains a strong reliance on human intervention. Furthermore, this method does not consider the continuity and stability of risk identification results over time; for example, it lacks mechanisms such as evidence latching or cooling to suppress misjudgments caused by short-term fluctuations. Therefore, in complex downhole environments, frequent fluctuations in identification results may still occur, affecting the reliability of decision-making.

[0004] While the process mentions issuing warnings to on-site personnel and updating the risk case database after risk confirmation, the procedure essentially ends after the warning is issued, failing to address how to prioritize and coordinate borehole, circulation, or drilling operations based on the risk status of multiple well sections. In deep boreholes, especially when multiple well sections simultaneously present varying degrees of blockage risk, simple risk warnings cannot resolve critical issues such as which section to address first, when to address it, and whether buffer operations are necessary. Existing methods lack risk budgeting or risk consumption mechanisms, making it impossible to uniformly quantify and compare the risks of different well sections. Consequently, they cannot generate time-constrained work deadlines, nor can they optimize scheduling based on urgency across multiple well sections. This deficiency makes it difficult to translate risk identification results into executable scheduling instructions, forcing on-site personnel to rely on manual experience for prioritization and decision-making, which can easily lead to processing delays or unreasonable resource allocation.

[0005] This technology fails to consider the differences in downhole disturbance caused by various operational behaviors and their cumulative effects. In practical engineering, high-intensity through-hole operations and enhanced circulation operations can significantly impact wellbore stability and mudstone expansion. Continuous high-disturbance operations can easily trigger secondary risks. However, existing early warning methods only focus on risk identification itself, without classifying subsequent operation types. Furthermore, they lack strategies such as distinguishing between high-disturbance and low-disturbance operations, prohibiting adjacent disturbances, or implementing buffer step insertion control strategies. They also lack a closed-loop judgment mechanism for the degree of post-operation state recovery. This makes it difficult to avoid risk amplification due to unreasonable operation sequences, even if risks are identified early, limiting their effectiveness in complex well conditions. Summary of the Invention

[0006] The purpose of this invention is to provide a method for predicting mudstone expansion and blockage in deep boreholes and optimizing the scheduling of borehole operations, thereby addressing some of the drawbacks and shortcomings pointed out in the background art.

[0007] The present invention adopts the following technical solution to solve the above-mentioned technical problems: a method for predicting mudstone expansion and blockage in deep boreholes and optimizing the scheduling of borehole operations, comprising: dividing the borehole into well sections according to depth, and setting risk budget items for each well section, wherein the risk budget items include at least one of the following: a budget for the number of precursor windows, a budget for the duration of a stage, and a budget for the number of low reversible events;

[0008] Within the sliding time window, at least two of the following parameters are collected: torque, pump pressure, return material parameters, and mechanical drilling speed. These parameters are compared with the baseline to form primary evidence. The precursor window is determined by combining the excluded evidence. When the precursor window appears consecutively for a set number of times, the blockage phase is initiated, followed by a predetermined probing action. If the pump pressure or torque does not recover to the baseline allowable range within the set time limit, it is determined to be a low reversibility event.

[0009] The corresponding risk budget is deducted based on the number of warning windows identified, the duration of the phase, and the number of low reversibility events. When any risk budget is exhausted, the deadline and urgency of the borehole operation for the corresponding well section are generated. The borehole operation requirements for multiple well sections are sorted according to the priority of the deadline, and a scheduling plan is generated by combining the urgency and scheduling constraints when the difference between the deadlines is not greater than the threshold.

[0010] Furthermore, the identification of the precursor window adopts an evidence latching and cooling-off determination mechanism; after the precursor window is identified, the latching is only released when a first set number of non-precursor windows appear consecutively; after excluding evidence triggering, a cooling-off period is entered, during which the precursor window is identified again only when the main evidence meets the second set number of times.

[0011] Furthermore, the predetermined test action is a graded progressive test, including a first test action and a second test action; the first test action is executed first, and if the pump pressure or torque does not recover to the baseline allowable range within a first time limit, the second test action is executed; and the well section status is divided into reversible, low reversible or near-blockage status according to the recovery time limit, so as to determine the low reversibility event and adjust the risk budget deduction range and urgency.

[0012] Furthermore, when generating the scheduling plan, a non-adjacent disturbance constraint is introduced to divide the through-hole operation into low-disturbance through-hole operation and high-disturbance through-hole operation; when an adjacent high-disturbance through-hole operation occurs, or when a high-disturbance through-hole operation is immediately followed by a drilling operation or cyclic operation in the same well section, a stable cyclic buffer step is forcibly inserted.

[0013] Furthermore, the first probing action is a low-disturbance probing action, which includes at least one of short-term displacement adjustment and short-range reciprocating motion; the second probing action is a high-disturbance probing action, which includes at least one of enhanced cleaning cycle and short-range mechanical through hole; the second probing action is performed only if the pump pressure or torque fails to recover to the baseline allowable range within the first time limit after the first probing action is performed.

[0014] Furthermore, the low-disturbance through-hole operation and the high-disturbance through-hole operation are classified according to the disturbance amplitude of the operation parameters. The disturbance amplitude is characterized by at least two of the pump pressure change, torque change, and circulation displacement change. When the disturbance amplitude is not less than the first threshold, it is determined to be a high-disturbance through-hole operation; otherwise, it is determined to be a low-disturbance through-hole operation.

[0015] Furthermore, the stabilization cycle buffering step continues until at least two of the following closure conditions are met: the pump pressure drops back to the baseline allowable range and remains there, the torque drops back to the baseline allowable range and remains there, and the return parameters recover to a stable range. If the closure conditions are not met within the second time limit, the operation is marked as not closed, and the urgency of subsequent operations is increased or the deadline for the second through-hole operation is moved forward.

[0016] Furthermore, when the difference in the cutoff time of the through-hole operation of multiple well sections is not greater than the threshold, the number of equipment switching times of the candidate scheduling sequence is calculated, and the scheduling sequence that satisfies the following constraints is selected from the candidate scheduling sequence with the minimum number of equipment switching times: adjacent through-hole operations are preferentially corresponding to different well sections, and two high-disturbance through-hole operations are not arranged consecutively in the same well section.

[0017] Furthermore, the candidate scheduling sequence is generated by a stepwise expansion method from a set of well segments whose deadline difference is not greater than the threshold. In each expansion step, an unscheduled well segment through-hole operation is added to the end of the sequence. When adding, priority is given to satisfying that adjacent through-hole operations correspond to different well segments. If there is no addition option that satisfies this condition, it is allowed to add through-hole operations in the same well segment, and the corresponding position is marked as a constraint conflict point.

[0018] Furthermore, when it is necessary to insert an operation between two consecutive high-disturbance through-hole operations in the same well section, if the closure condition is met after the previous high-disturbance through-hole operation, then a low-disturbance through-hole operation is inserted; if the closure condition is not met after the previous high-disturbance through-hole operation, then the stable cycle buffer step is inserted; after insertion, the number of equipment switching is recalculated. If the recalculated number of equipment switching exceeds the sum of the minimum number of switching and the allowable increment, then the next candidate scheduling sequence is selected.

[0019] The beneficial effects of this invention are as follows: By dividing the borehole into sections according to depth and introducing a multi-dimensional risk budgeting mechanism that includes budgeting the number of precursor windows, the duration of each stage, and the number of low-reversibility events, the previously difficult-to-quantify risk evolution process of mudstone expansion and blockage is transformed into a measurable and trackable dynamic consumption process. Simultaneously, based on a sliding time window, key parameters such as torque, pump pressure, returned material parameters, and mechanical drilling rate are continuously monitored. Combined with the identification of precursor windows using primary evidence and exclusionary evidence, and with the evidence latching and cooling judgment mechanism, short-term disturbances and misjudgments are effectively suppressed, improving the accuracy and stability of blockage precursor identification. Furthermore, by using graded progressive exploration to determine the recoverability of the well section, the identification of low-reversibility events becomes more objective, enabling earlier and more reliable detection of high-risk well sections and improving the foresight of blockage prediction.

[0020] Based on the risk budget exhaustion, the deadline and urgency of through-hole operations are generated and incorporated into the multi-well section collaborative scheduling process, achieving risk-driven operation sequencing and decision optimization. During scheduling, by introducing non-adjacent disturbance constraints, stable cycle buffer steps, and high- and low-disturbance operation classifications, the continuous superposition of high-disturbance operations is effectively avoided, reducing the risk of drastic fluctuations in downhole conditions. Simultaneously, by combining the principle of minimizing equipment switching times with a gradual expansion mechanism of candidate scheduling sequences, operational efficiency is considered while meeting safety constraints. Attached Figure Description

[0021] Figure 1 This is the logic diagram for determining and scheduling the risk of mudstone expansion and blockage in deep boreholes in this invention.

[0022] Figure 2 This is a comparison chart of baseline parameters and initial risk budgets for well sections A, B, and C in Embodiment 1 of the present invention.

[0023] Figure 3 This is a diagram showing the evolution of the precursor deviation and latching cooling identification of well section C during the sliding time window in Embodiment 1 of the present invention.

[0024] Figure 4 This is a diagram showing the recovery results and risk budget deduction for well section C-level testing in Embodiment 1 of the present invention.

[0025] Figure 5 This is a diagram showing the scanning and threshold classification of disturbance amplitude parameters from well section D to well section G in Embodiment 2 of the present invention.

[0026] Figure 6 This is a time-series evolution diagram of the initial scheduling sequence and the high and low disturbance states in Embodiment 2 of the present invention.

[0027] Figure 7 This is a comprehensive evaluation decision diagram of candidate scheduling sequences in Embodiment 2 of the present invention. Detailed Implementation

[0028] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0029] Combined with appendix Figure 1 This invention relates to a method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling. First, the entire borehole is segmented based on its actual depth to form multiple continuous well sections. The division of well sections can be determined according to changes in formation structure, stable drilling parameter ranges, or preset depth intervals, ensuring relative consistency in geological conditions and operational status for each well section, thus facilitating subsequent risk management by zone.

[0030] The risk budget item quantifies the tolerable degree of abnormal development in the current well section during the current operational phase and serves as an important basis for subsequent assessment of blockage risk evolution and scheduling decisions. The risk budget item includes at least one of the following: a precursor window frequency budget, a phase duration budget, and a low-reversibility event frequency budget. Multiple budgets can also be set simultaneously as needed to improve assessment accuracy. The precursor window frequency budget limits the cumulative number of precursor windows allowed within a well section. Precursor windows characterize early abnormal signals occurring during drilling, typically reflecting deviations in torque, pump pressure, return material parameters, or mechanical drilling rate parameters from the baseline. When the cumulative number of precursor windows in a well section gradually approaches or reaches this budget value, it indicates that the risk of blockage in that well section is increasing.

[0031] As the operation progresses, when a well section enters a certain risk stage, the system accumulates the duration of that stage. When the accumulated duration approaches or exceeds the preset stage duration budget, it indicates that the well section has been under adverse conditions for too long, further exacerbating the risk of blockage.

[0032] The sliding time window moves continuously along the time axis at a fixed time length. Within each time window, at least two parameters are collected from torque, pump pressure, returned material parameters, and mechanical drilling rate to ensure timely reflection of changes in downhole conditions. The collected data are compared with a pre-established baseline, which serves as a parameter reference range for the corresponding well section under stable drilling conditions.

[0033] When the collected parameters show a trend of deviation from the baseline allowable range, this change is taken as the primary evidence. Simultaneously, exclusionary evidence is used to verify the source of the anomaly. Exclusionary evidence is used to eliminate non-risk fluctuations caused by equipment operation adjustments or short-term disturbances. When the primary evidence is valid and not refuted by exclusionary evidence, the current time window is designated as a precursor window, used to characterize an early signal of congestion risk.

[0034] If a well segment is identified as a precursor window within multiple consecutive sliding time windows, and its consecutive occurrences reach a preset threshold, the risk status of the corresponding well segment is determined to have escalated, advancing it to a higher-level blockage stage. Subsequently, a predetermined exploratory action is immediately executed to verify the recoverability of the well segment's condition. This predetermined exploratory action is used to observe the response of key parameters through limited intervention, thereby determining whether the well segment shows a further deterioration trend.

[0035] After performing the predetermined exploratory actions, the changes in pump pressure or torque are continuously monitored. If the pump pressure or torque fails to recover to the baseline allowable range within the set time limit, it is determined that a low reversibility event has occurred in the well section, indicating that the well section's response to intervention measures is weak in the current state, and the risk of blockage is further increased.

[0036] Based on the identification results of the precursor windows, the budget for the corresponding precursor window frequency is deducted one by one according to the cumulative number of occurrences; based on the duration of the blockage stage in the well section, the duration of the stage is accumulated and the budget for the duration of the stage is deducted accordingly; for the identified low reversibility events, the budget for the low reversibility event frequency is deducted according to the number of times the event occurs.

[0037] When any type of risk budget is reduced to zero, the well section is deemed to have reached its risk tolerance limit, requiring borehole operations to be performed within a specified time to prevent further deterioration. Based on this determination, a borehole operation deadline is generated for that well section, representing the maximum allowable delay. Simultaneously, the urgency of the well section is calculated based on the budget consumption rate, remaining budget, and recent risk evolution trends, reflecting the priority of performing the borehole operation.

[0038] When multiple well sections simultaneously require borehole operations, they are prioritized based on their borehole operation deadlines, with earlier deadlines given priority. If the difference in deadlines between different well sections is significant, they are scheduled sequentially according to their deadlines. When the difference in deadlines across multiple well sections is no greater than a preset threshold, a comprehensive decision is made considering the urgency of each well section and preset scheduling constraints. Scheduling constraints may include equipment switching limitations, operational continuity requirements, and disturbance control requirements.

[0039] Under the premise of meeting scheduling constraints, priority should be given to performing borehole operations in well sections with higher urgency, and the operation sequence should be optimized to balance risk control and operation efficiency.

[0040] Once a time window is identified as a precursor window based on the primary evidence within the sliding time window, the identification result is immediately latched, and the identification results of subsequent time windows are continuously tracked while in the latched state. The latching state is only released when non-precursor windows appear consecutively for a first set number of times, thereby avoiding frequent interruptions to the precursor window due to short-term fluctuations and improving the reliability of identification continuity. At the same time, when exclusion evidence is triggered, the current anomaly is determined to be a non-risk disturbance and enters a cooling-off period. During the cooling-off period, the direct identification of regular precursor windows is suspended. The precursor window is only re-identified when the primary evidence meets the second set number of consecutive requirements in subsequent time windows, in order to prevent premature entry into the risk assessment process before the system has recovered and stabilized.

[0041] The first probing action is performed. This is a low-disturbance intervention used to observe the response of key parameters without significantly altering downhole conditions. After the first probing action, pump pressure or torque is continuously monitored. If it recovers to the baseline allowable range within a first time limit, the current well section is considered to be in a reversible state. If it does not recover within the first time limit, the second probing action is performed. This is a high-disturbance intervention used to enhance the clearing or unblocking effect.

[0042] After the second trial operation, the well section status is classified based on the time required for pump pressure or torque to recover to the baseline allowable range. When parameters recover within a short time, it is classified as a low-reversibility state; when they fail to recover or recover poorly within an extended observation period, it is classified as a near-blockage state. These status classifications are used to determine subsequent low-reversibility events, and also to adjust the risk budget deduction and calculate the urgency of borehole operations, thereby achieving refined management and dynamic control of well section risks.

[0043] Low-disturbance testing is used to make minor interventions without significantly changing the original drilling conditions. Specific forms include one or more combinations of short-term flow rate adjustments or short-range reciprocating motions. Short-term flow rate adjustments refer to making small adjustments to the circulating flow rate within a short period to observe changes in pump pressure and returned material; short-range reciprocating motions refer to moving the drill string up and down within a limited stroke range to determine if there is localized blockage or slight buildup in the well.

[0044] After the first trial operation is performed, the pump pressure or torque is continuously monitored and compared with the baseline allowable range. If the relevant parameters recover to the baseline allowable range within the first time limit, it indicates that the current well section has good self-recovery capability and no further enhanced intervention measures are needed, thereby avoiding unnecessary high-disturbance operations.

[0045] If, after the first trial action, the pump pressure or torque fails to recover to the baseline allowable range within the first time limit, the low-disturbance intervention is deemed insufficient, and a second trial action is executed. The second trial action is a high-disturbance trial, aimed at effectively clearing or cleaning the well section by increasing the intervention intensity. The second trial action includes at least one of enhanced cleaning circulation or short-range mechanical perforation. Enhanced cleaning circulation refers to increasing the circulation intensity to remove sediment or expanded material from the well; short-range mechanical perforation refers to mechanical perforation operations within a controlled range to directly eliminate local blockages.

[0046] Based on the degree of impact of various wellbore operations on well parameters, wellbore operations are classified into low-disturbance wellbore operations and high-disturbance wellbore operations. Low-disturbance wellbore operations refer to operations that have minimal impact on pump pressure, torque, and circulation status, and are suitable for well sections with lower risk or better recoverability. High-disturbance wellbore operations refer to operations that cause significant changes in the fluid and mechanical state of the well, and are typically used to deal with well sections with higher risk of blockage or poor recoverability.

[0047] During the scheduling plan generation process, candidate job sequences are subjected to constraint checks. If two adjacent high-disturbance borehole jobs are detected, the sequence is determined to not meet the disturbance control requirements. To avoid the adverse effects of continuous high disturbances on wellbore stability and the circulation system, a stabilization circulation buffer step is forcibly inserted between the two high-disturbance borehole jobs in the above situation. The stabilization circulation buffer step is used to restore the flow state in the well, reduce parameter fluctuations, and provide stable initial conditions for subsequent jobs.

[0048] Furthermore, if drilling or cyclic operations are immediately scheduled for the same well section after a high-disturbance borehole operation is completed, this is also considered a case of cumulative disturbance. In this case, a stabilizing cyclic buffer step should be inserted between the high-disturbance borehole operation and the subsequent operation to avoid the accumulation of risks caused by continuous high-intensity disturbances in the same well section.

[0049] At least two of the following parameters—pump pressure change, torque change, and circulating displacement change—are selected as indicators of disturbance amplitude. The deviation of each indicator from the baseline state is calculated and normalized. When the disturbance amplitude reaches or exceeds a preset first threshold, the through-hole operation is determined to be a high-disturbance through-hole operation; when the disturbance amplitude is below the first threshold, it is determined to be a low-disturbance through-hole operation.

[0050] During the stabilization and buffering process, closure conditions are set as criteria for ending the buffer to ensure sufficient recovery of the wellbore condition. These closure conditions include at least two of the following: pump pressure returning to and maintaining within the baseline allowable range, torque returning to and maintaining within the baseline allowable range, and return parameters recovering to a stable range. When any two of these conditions are simultaneously met, the buffering process is considered complete, and subsequent operations can proceed. If the closure conditions are not met within the second time limit, the buffering step is marked as unclosed, indicating that potential risks still exist in the well section. In this case, the urgency of subsequent related operations is increased, or the deadline for re-perforating the well section is moved forward to encourage the dispatch system to prioritize this section.

[0051] Furthermore, when multiple well sections simultaneously require through-hole operations and the difference in their deadlines does not exceed a preset threshold, a refined scheduling phase is initiated. Multiple candidate scheduling sequences are constructed, and the number of equipment switching operations for each sequence is calculated to reflect the efficiency and cost of scheduling execution. Based on this, a set of candidate scheduling sequences with the fewest equipment switching operations is selected, and further constraint screening is applied within this set. Priority is given to scheduling sequences corresponding to different well sections for adjacent through-hole operations to reduce the risk accumulation caused by continuous operations in the same well section. Simultaneously, two highly disruptive through-hole operations are not scheduled consecutively in the same well section to avoid the cumulative effect of high-intensity disturbances.

[0052] The candidate scheduling sequence is generated in a step-by-step expansion manner, that is, starting from an empty sequence or the initial well section, the operation tasks are gradually added to the end of the sequence according to certain rules until all well sections to be scheduled are included in the sequence.

[0053] During each expansion step, a borehole operation is selected from the unscheduled well segments and appended to the end of the current sequence. When appending, candidate well segments that allow two adjacent borehole operations to correspond to different well segments are prioritized to reduce the risk accumulation caused by continuous operations in the same well segment, while also helping to disperse the impact of disturbances and improve the overall operational stability.

[0054] When multiple candidate well sections satisfy the above conditions exist, further screening can be performed using urgency or other scheduling indicators to determine the optimal addition target. If no candidate well section in the current step satisfies the conditions for adjacent operations corresponding to different well sections, then a through-hole operation within the same well section can be selected for addition to ensure the complete generation of the scheduling sequence. In this case, the addition location is marked as a constraint conflict point to record the location where the priority constraint is violated.

[0055] Constraint conflict points can serve as important reference information in subsequent scheduling optimization processes, used to compare and filter different candidate scheduling sequences. For example, when multiple candidate sequences have the same number of device handovers, the sequence with fewer constraint conflict points can be prioritized to further improve the rationality and safety of the scheduling results.

[0056] During the generation of the scheduling sequence, when two consecutive high-disturbance perforation operations are detected in the same well section, a transition operation needs to be inserted between the two high-disturbance perforation operations to avoid the risk caused by the superposition of high-disturbance operations in the same well section. The state of the well section after the completion of the previous high-disturbance perforation operation is judged to determine whether it meets the closure condition. The closure condition is used to characterize whether the key parameters in the well have returned to a stable state.

[0057] When the closure conditions are met after a previous high-disturbance through-hole operation, it indicates that the well section has returned to a relatively stable state. In this case, a low-disturbance through-hole operation is inserted between the two high-disturbance through-hole operations. When the closure conditions are not met after a previous high-disturbance through-hole operation, it indicates that the well section is still unstable or has not fully recovered. In this case, it is not appropriate to directly arrange subsequent through-hole operations. Therefore, a stabilization circulation buffer step is inserted between the two high-disturbance through-hole operations. Through continuous circulation, the flow state in the well tends to stabilize, and key parameters return to the allowable range, thereby reducing the risk of subsequent operations.

[0058] After the insertion operation is completed, the number of device handovers is recalculated for the updated scheduling sequence. If the recalculated number of device handovers does not exceed the sum of the minimum number of handovers of the current candidate sequence and the preset allowable increment, the scheduling sequence is retained as a valid candidate; if it exceeds this range, the insertion operation is considered to have caused a significant decrease in scheduling efficiency, and the current candidate scheduling sequence is abandoned, and the next candidate scheduling sequence is selected for evaluation.

[0059] Example 1:

[0060] During drilling in a deep mudstone formation, well sections A, B, and C all exhibited relatively stable torque, pump pressure, returned material parameters, and mechanical drilling rate during the initial stable phase. Based on this, the system established a baseline database. Table 1 shows the basic parameters and initial risk budget for the three well sections. For ease of subsequent calculations, the returned material parameters are represented by the returned material coarse particle index; a higher value indicates a higher proportion of abnormal coarse particles in the returned material. The mechanical drilling rate is expressed as the rate of penetration per unit footage. Compared to well sections A and B, well section C has a higher baseline load level and lower budget redundancy, indicating that well section C is located in a more sensitive monitoring position. Figure 2 The steps of first establishing a baseline database and then configuring risk budgets according to well sections provide a unified reference for subsequent sliding window identification, recovery coefficient determination, and budget deduction.

[0061]

[0062] During continuous operation, well sections A and B only experienced minor fluctuations, with measured values ​​consistently remaining within the baseline allowable range, which the system identified as normal disturbances. Well section C, however, initially exhibited a slight anomaly, characterized by a sustained increase in torque, a slow rise in pump pressure, an increase in the coarse particle index of the returned material, and a slight decrease in the mechanical drilling rate. The system employs a sliding time window approach for judgment, comparing at least three indicators—torque, pump pressure, and returned material parameters—simultaneously within each sliding window, and combining this with exclusionary evidence for comprehensive identification. Exclusionary evidence includes a sudden drop in mechanical drilling rate without synchronous shift in pump pressure and torque, a brief surge in returned material followed by an immediate decline, and single-window disturbances caused by one-time manual parameter adjustments. To quantify the severity of precursors, this embodiment defines the precursor deviation as:

[0063]

[0064] in, For the current torque, The current pump pressure, The parameters of the current output item. , , These are the corresponding baseline values. When Furthermore, if at least two primary pieces of evidence simultaneously exceed their respective permissible deviation thresholds, the system will identify the sliding window as a precursor window.

[0065] Table 2 lists the main monitoring data and identification results for well section C during seven consecutive sliding windows. This data constitutes the core plot of this embodiment. It can be seen that well section C initially showed only a slight deviation, then the anomaly gradually accumulated, and intensified again after cooling, ultimately triggering the tiered exploration process. Figure 3 Further, the evolution of precursor deviation and latch-up cooling identification results for well section C during the sliding time window are presented. Among them, the precursor deviation of W2 is 0.104, which, although not reaching 0.12, already shows a significant accumulation of anomalies, and all three main evidence items simultaneously exceed the threshold. The precursor deviation of W3 is 0.141, which is higher than the threshold of 0.12, thus confirming that the anomaly in well section C has further intensified. The precursor deviation of W4 is 0.036. Although the value has decreased, due to the exclusion evidence hitting the target, the system does not simply eliminate the risk, but enters the cooling period. The precursor deviations of W6 and W7 are 0.175 and 0.229, respectively, which are again significantly higher than the threshold, indicating that the anomaly has intensified again after cooling, suggesting that well section C is evolving from a local deviation to a blockage stage.

[0066]

[0067] In the above data, well section C showed significant abnormal accumulation at W2, and W3 further met the precursor deviation threshold with synchronous enhancement of the main evidence. Based on this, the system determined that it had entered the precursor latching state. At this time, the precursor judgment was not immediately lifted due to a single drop in W4, but rather required two consecutive non-precursor windows before the latch could be lifted, thus avoiding the interruption of the precursor chain due to local fluctuations. W4 showed a situation where the drop in returned material parameters and pump pressure were synchronized, accompanied by short-term parameter adjustment traces, so it was excluded from the evidence, and the system entered the cooling period. The purpose of the cooling period is to suppress short-term repeated alarms. Only when the main evidence meets the conditions again with higher intensity after the cooling period will the precursor recognition be restored. Although W5 had a certain deviation, it was not identified as a new precursor window because it was still in the cooling period and had not reached the recovery threshold. W6 and W7 enhanced continuously again, and the deviation was significantly higher than the previous windows. Based on this, the system confirmed that the anomaly in well section C was not noise, but was evolving towards the blockage stage.

[0068] When the precursor window in well section C appeared consecutively for a set number of times, the system advanced to the blockage stage and performed a graded progressive test. The first test adopted a low-disturbance approach, including short-term flow rate adjustments and short-range reciprocating motions, aiming to verify the reversibility of the anomaly without significantly amplifying wellbore disturbance. After the first test, the torque in well section C decreased from 27.4 kN·m to 25.9 kN·m, the pump pressure decreased from 20.8 MPa to 19.6 MPa, and the returned material parameters decreased from 0.53 to 0.45. Although there was improvement, it still had not returned to the baseline allowable range. Based on this, the system concluded that a simple low-disturbance correction was insufficient to eliminate the flow resistance caused by the superposition of local diameter reduction and mudstone expansion, and therefore continued with the second test. The second test adopted a high-disturbance approach, including enhanced cleaning circulation and short-range mechanical through-hole drilling. After the second test, the torque dropped to 24.8 kN·m, the pump pressure dropped to 18.9 MPa, the return material parameters dropped to 0.40, and the mechanical drilling speed recovered to 4.8 m / h. However, the pump pressure and return material parameters were still higher than the baseline allowable upper limit, indicating that this section had not fully recovered to the reversible state. Figure 4 The results of the well section C-level test recovery and risk budget deduction are presented, reflecting that although the deviations of various parameters decreased after the first test, the recovery was not yet complete. The overall recovery effect was further enhanced after the second test, forming a progressive test chain from low-disturbance verification to high-disturbance treatment.

[0069] To quantify the trial results uniformly, a recovery coefficient is set:

[0070]

[0071] in, , , These are the recovery values ​​after the trial action. When When it is determined to be a reversible state, When it is judged to be a low reversible state, when The condition was determined to be near blockage. Substituting the data from the second test of well section C, we obtained:

[0072]

[0073]

[0074] Figure 4The corresponding calculation results show that, based on the instantaneous recovery coefficient after the second test, well section C is close to the reversible boundary. However, due to its experience of precursor latch-up, recurrence after cooling, and failure of the first test, the system does not directly classify it as reversible. Instead, it adjusts the state based on process evidence. If the same well section experiences precursor recurrence within a single treatment cycle and requires a second test for recovery, the state is downgraded by one level. Therefore, well section C is ultimately determined to be in a low-reversible state and recorded as a low-reversibility event. This approach is superior to simply judging based on the final recovery value because it incorporates the recovery difficulty during the evolution process into the decision, which is more consistent with the gradual characteristics of mudstone expansion blockage.

[0075] Regarding risk budget deductions, the number of precursor windows, duration of each phase, and low-reversibility events are mapped to budget consumption, defining the overall risk consumption as:

[0076]

[0077] in, The number of identified warning windows, The duration of the congestion phase, in minutes. For the number of reversible events, =1.0, =0.8, =2.0. In this round of treatment, well section C has identified a total of 4 precursor windows, a blockage phase lasting 15 minutes, and 1 low-reversibility event. Therefore:

[0078]

[0079] Accordingly, the budgeted number of precursor windows for well section C decreased from 4 to 0, the budgeted stage duration decreased from 24 minutes to 9 minutes, and the budgeted number of low-reversibility events decreased from 1 to 0. The system also marked well section C as a high-urgency section and generated an early warning for subsequent borehole operations. In contrast, well section A experienced only one minor deviation during the same period, without forming consecutive precursor windows, and its budget remained unchanged. Although well section B had two marginal deviations, both were excluded from evidence filtering, and the budgeted number of precursor windows was reduced by only 1; the budgeted stage duration and low-reversibility events were not substantially consumed.

[0080] Example 2:

[0081] During a continuous operation cycle of mudstone expansion blockage in a deep borehole, borehole access requests arose successively in well sections D, E, F, and G. After receiving the prediction results for each well section, the scheduling system formed a set of operations to be scheduled, recording their estimated deadlines, initial urgency, types of operations to be performed, and operational disturbance characteristics. To avoid relying solely on experience to determine different operational intensities, a disturbance amplitude index was constructed using pump pressure changes, torque changes, and circulation displacement changes. Based on this, borehole access operations were divided into low-disturbance and high-disturbance operations. Table 3 presents the initial data for the four well sections, where the deadline is expressed as relative remaining time, with smaller values ​​indicating greater urgency. Figure 5 The results of disturbance amplitude parameter scanning and threshold classification for well sections D to G are presented. It can be seen that each well section can be classified as high or low disturbance under a unified threshold. Well section F falls into the low disturbance classification zone, while well sections D, E, and G fall into the high disturbance classification zone. Therefore... Figure 5 This corresponds to the process of uniformly quantifying and hierarchically identifying the disturbance intensity of through-hole operations.

[0082]

[0083] The following disturbance amplitude calculation formula is used to uniformly quantify the operation of each through-hole, where the changes in pump pressure, torque, and circulating displacement are all taken as the absolute values ​​of their relative rates of change, denoted as ΔP, ΔT, and ΔQ, respectively:

[0084] Δ Δ Δ

[0085] when If the disturbance is high, the well section is considered a high-disturbance through-hole operation; otherwise, it is considered a low-disturbance through-hole operation. Substituting the data from Table 3, the disturbance amplitude of well section D is:

[0086]

[0087] The percentages for well section E are 16.45%, for well section F they are 7.9%, and for well section G they are 14.90%. Figure 5 The classification scatter plots and threshold shaded areas further indicate that well sections D, E, and G are classified as high-disturbance through-hole operations, while well section F is classified as a low-disturbance through-hole operation. This classification is based on the fact that high-disturbance operations are typically accompanied by more significant flow field reconstruction and mechanical load redistribution; if implemented continuously, they can easily lead to further instability in the mudstone swelling zone near the wellbore. Low-disturbance operations, on the other hand, are more suitable as transitional or interim steps.

[0088] The scheduling system then generated an initial sequence based on the remaining deadline time, from smallest to largest: well segment E, well segment G, well segment D, and well segment F. While this sequence satisfied the deadline priority principle, the system found during consistency checks that well segments E and G were both high-disturbance operations. Executing them directly adjacently would result in a superposition of high disturbances; if well segment G were followed by well segment D, the three consecutive high-disturbance operations would further amplify the risk, violating the constraint that disturbances cannot be adjacent. Therefore, the system did not directly adopt this initial sequence but instead used it as the initial framework for the first candidate sequence, attempting to insert stable cyclic buffer steps and low-disturbance operations. Figure 6 The temporal evolution of the initial scheduling sequence and the results of the high and low disturbance state ladder are presented. The cutoff remaining time for the initial sequence is 48 min, 50 min, 52 min and 55 min respectively. The state ladder shows that well segments E, G and D are in a high disturbance state, and well segment F is in a low disturbance state. This intuitively reveals the problem of high disturbance clustering in the first three positions of the initial sequence.

[0089] In the first round of rearrangement, the system prioritizes inserting the low-disturbance well segment F between well segments E and G, forming a candidate sequence S1, namely, high-disturbance through-hole in well segment E, low-disturbance through-hole in well segment F, high-disturbance through-hole in well segment G, and high-disturbance through-hole in well segment D. Although this sequence weakens the direct adjacency of the first two high-disturbance operations, well segments G and D are still continuously high-disturbance and require further processing. Therefore, the system inserts a stabilizing cycle buffer step after well segment G, resulting in the corrected sequence S1′. Meanwhile, another candidate sequence S2, which is gradually expanded, is established. The expansion method is to first select well segments E and G with the closest deadlines and the highest urgency. Since there is no second available low-disturbance well segment, only well segment D can be added and the position is marked as the constraint conflict point. Then, the feasibility of inserting buffers or adjusting the order is evaluated before and after the conflict point. Finally, the sequence S2 is obtained, which consists of high-disturbance through-hole in well segment E, stable circulation buffer, high-disturbance through-hole in well segment D, low-disturbance through-hole in well segment F, and high-disturbance through-hole in well segment G.

[0090] The stable cyclic buffering step does not wait for a fixed duration, but continues until at least two of the closure conditions are met. To improve the consistency of the judgment, a closure exponent is introduced to comprehensively characterize the stability after buffering. Let the pump pressure drop after buffering maintain the judgment value be... The torque drop remains the judgment value. The return parameter stability judgment value is If the condition is met, the value is 1; otherwise, it is 0. Therefore, the closure index is:

[0091]

[0092] when When the buffer is considered closed, The well section was considered not closed. Table 4 lists the monitoring results of the two actual insertion buffer steps. It can be seen that after the high-disturbance through-hole was executed in section E, the buffer pump pressure dropped from 15% above the baseline to 5% above, the torque dropped from 12% above to 4% above, and the return material fluctuation range converged from 0.18 to 0.07. All three items met the maintenance requirements, so the closure index was 3. After the high-disturbance through-hole was executed in section G, the buffer pump pressure was still 9% above, the torque was still 8% above, and although the return material fluctuation decreased, it was still unstable. The return material fluctuation after buffering was 0.12, which did not reach the stable threshold of 0.10. Therefore, all three items did not meet the maintenance requirements, the closure index was 0, and it was judged to be not closed.

[0093]

[0094] The closure of the buffer after well segment E indicates that the segment has returned to a relatively stable state after being affected by high-disturbance operations. Therefore, the system allows for subsequent low-disturbance operations or other well segment operations without imposing additional penalties. The failure to close the buffer after well segment G indicates that the flow and stress conditions around the segment are not yet sufficiently stable. Continuing to schedule high-disturbance boreholes or directly connecting to drilling and circulation operations in the same segment would significantly increase the risk of secondary disturbance. Based on this, the system increases the urgency of well segment G and moves the estimated remaining time for its re-borehole from 50 minutes to 42 minutes, giving it greater attention in subsequent rearrangements. Figure 6 The critical time limit broken line after the closed feedback corresponds to this adjustment process, indicating that the closed feedback is not a static marker, but directly affects the priority of subsequent time-series reordering. The significance of this closed decision mechanism is that it no longer understands the buffer as a fixed form of job filling, but as a measurable, judgmentable, and feedback-enabled dynamic stable step, thereby improving scheduling safety.

[0095] When the difference in deadlines among multiple well segments is no greater than a threshold, simply sorting by deadline order is insufficient to reflect the true priority. This is because, in the scenario of mudstone expansion and blockage, urgency is not solely determined by the remaining time, but is also closely related to the stability of the well segment after previous treatments, the cost of equipment switching, and the number of constraint conflicts. Therefore, a comprehensive scheduling evaluation value is further constructed to compare candidate sequences. Let the average urgency of a candidate sequence be U, the number of equipment switching times be M, the number of constraint conflict points be H, and the number of unclosed buffers be N, then the evaluation value is defined as:

[0096]

[0097] The higher the evaluation value, the better the overall sequence. For the corrected sequence S1′, executed according to well segment E, well segment F, well segment G, buffer, and well segment D, with 2 equipment switching times, 0 constraint conflict points, 1 unclosed buffer occurrence, and an average urgency of 0.827, we have:

[0098]

[0099] For sequence S2, executed according to well segment E, buffer, well segment D, well segment F, and well segment G, with 3 equipment switching times, 0 constraint conflict points, 0 unclosed buffer times, and an average urgency of 0.827, then:

[0100]

[0101] Figure 7 The comprehensive evaluation results of the candidate scheduling sequences are presented, with S1′ having a comprehensive evaluation value of 56.7 and S2 having a comprehensive evaluation value of 58.7. Although S2 has a slightly higher number of device switching operations, it avoids the risk of unclosed buffer legacy issues, resulting in a higher comprehensive evaluation and thus placing it in a better position within the comprehensive score band.

[0102] Before finalizing the plan, the system conducted an additional check on the insertability of two consecutive high-disturbance operations in the same well section. Since well section G failed to close after the first high-disturbance perforation, if another high-disturbance perforation is required in this section, another high-disturbance operation cannot be simply inserted between the two high-disturbance operations. Instead, a stable circulation buffer step should be inserted first. If the section had already closed after the previous high-disturbance operation, a low-disturbance perforation operation could be inserted to balance recovery and efficiency. The failure to close well section G directly ruled out the possibility of immediately scheduling the second high-disturbance operation. Therefore, the system retained the buffer step and abandoned an alternative sequence that superficially required fewer switching operations but allowed for exceeding incremental limits. This approach, although increasing switching actions locally, avoided the uncontrollable risks caused by the repeated superposition of high disturbances in the same well section.

[0103] After comprehensive comparison, the system ultimately selected sequence S2 as the execution plan. This involves first performing a high-disturbance through-hole operation on well segment E, and after confirming its subsequent buffer closure, then transitioning to a high-disturbance through-hole operation on well segment D. This is followed by a transition using a low-disturbance operation on well segment F, and finally processing well segment G. This result does not simply aim for the fewest switching operations, but rather balances initial urgency, buffer closure feedback, constraint conflict control, and overall sequence stability, all while ensuring close deadlines. In particular, the inclusion of information regarding the unclosed state of well segment G allows for dynamic correction of the scheduling results, rather than remaining fixed after a single scheduling iteration.

Claims

1. A method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling, characterized in that, include: The borehole is divided into sections according to depth, and a risk budget item is set for each section. The risk budget item includes at least one of the following: the budget for the number of precursor windows, the budget for the duration of the stage, and the budget for the number of low reversible events. Within the sliding time window, at least two of the following parameters are collected: torque, pump pressure, return material parameters, and mechanical drilling speed. These parameters are compared with the baseline to form primary evidence. The precursor window is determined by combining the excluded evidence. When the precursor window appears consecutively for a set number of times, the blockage phase is initiated, followed by a predetermined probing action. If the pump pressure or torque does not recover to the baseline allowable range within the set time limit, it is determined to be a low reversibility event. The corresponding risk budget is deducted based on the number of warning windows identified, the duration of the phase, and the number of low reversibility events. When any risk budget is exhausted, the deadline and urgency of the borehole operation for the corresponding well section are generated. The borehole operation requirements for multiple well sections are sorted according to the priority of the deadline, and a scheduling plan is generated by combining the urgency and scheduling constraints when the difference between the deadlines is not greater than the threshold.

2. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 1, characterized in that, The identification of the precursor window adopts an evidence latching and cooling-off determination mechanism; after the precursor window is identified, the latching is only released when a first set number of non-precursor windows appear consecutively; after excluding evidence triggering, a cooling-off period is entered, during which the precursor window is identified again only when the main evidence meets the second set number.

3. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 1, characterized in that, The predetermined testing action is a graded progressive testing, including a first testing action and a second testing action; the first testing action is executed first, and if the pump pressure or torque does not recover to the baseline allowable range within the first time limit, the second testing action is executed; and the well section status is divided into reversible, low reversible or near-blockage status according to the recovery time limit, so as to determine the low reversibility event and adjust the risk budget deduction range and urgency.

4. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 1, characterized in that, When generating the scheduling plan, a non-adjacent disturbance constraint is introduced to divide the through hole operation into low-disturbance through hole operation and high-disturbance through hole operation; when an adjacent high-disturbance through hole operation occurs, or when a high-disturbance through hole operation is followed immediately by a drilling operation or cyclic operation in the same well section, a stable cyclic buffer step is forcibly inserted.

5. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 3, characterized in that, The first probing action is a low-disturbance probing action, which includes at least one of short-term displacement adjustment and short-range reciprocating motion; the second probing action is a high-disturbance probing action, which includes at least one of enhanced cleaning cycle and short-range mechanical through hole; the second probing action is performed only if the pump pressure or torque fails to recover to the baseline allowable range within the first time limit after the first probing action is performed.

6. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 4, characterized in that, The low-disturbance through-hole operation and the high-disturbance through-hole operation are classified according to the disturbance amplitude of the operation parameters. The disturbance amplitude is characterized by at least two of the pump pressure change, torque change, and circulation displacement change. When the disturbance amplitude is not less than the first threshold, it is determined to be a high-disturbance through-hole operation; otherwise, it is determined to be a low-disturbance through-hole operation.

7. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 4, characterized in that, The stabilization cycle buffering step continues until at least two of the following closure conditions are met: pump pressure drops back to the baseline allowable range and remains there, torque drops back to the baseline allowable range and remains there, and return parameters recover to a stable range. If the closure conditions are not met within the second time limit, the operation is marked as not closed, and the urgency of subsequent operations is increased or the deadline for the second through-hole operation is moved forward.

8. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 4, characterized in that, When the difference in the cutoff time of the through-hole operation of multiple well sections is not greater than the threshold, the number of equipment switching times of the candidate scheduling sequence is calculated, and the scheduling sequence that satisfies the following constraints is selected from the candidate scheduling sequence with the minimum number of equipment switching times: adjacent through-hole operations are preferentially corresponding to different well sections, and two high-disturbance through-hole operations are not arranged consecutively in the same well section.

9. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 8, characterized in that, The candidate scheduling sequence is generated by a stepwise expansion method from a set of well segments whose deadline difference is not greater than the threshold. In each expansion step, an unscheduled well segment is added to the end of the sequence. When adding, priority is given to adjacent well segments corresponding to different well segments. If there is no addition option that meets this condition, it is allowed to add well segments with the same well segment, and the corresponding position is marked as a constraint conflict point.

10. The method for predicting mudstone expansion and blockage in deep boreholes and optimizing borehole operation scheduling according to claim 7 or 8, characterized in that, When an operation needs to be inserted between two consecutive high-disturbance through-hole operations in the same well section, if the closure condition is met after the previous high-disturbance through-hole operation, a low-disturbance through-hole operation is inserted; if the closure condition is not met after the previous high-disturbance through-hole operation, the stable cycle buffer step is inserted; after insertion, the number of equipment switching is recalculated. If the recalculated number of equipment switching exceeds the sum of the minimum number of switching and the allowable increment, the next candidate scheduling sequence is selected.

Images