A method and device for integrated intelligent monitoring of well leakage and overflow under all drilling conditions

By introducing modules for mud thermal expansion and compressibility correction, drill cuttings load estimation, and time-series neural network prediction, the existing well leakage and overflow monitoring devices have solved the monitoring blind spots and error problems under high temperature, high pressure, and tripping conditions, achieving high-precision monitoring and early warning that are adaptive under all working conditions.

CN123109640APending Publication Date: 2026-11-17HENAN JIABAO MACHINERY EQUIPMENT CO LTD
View PDF 1 Cites 0 Cited by

Patent Information

Application Number
CN202611372387.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-06
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing well leakage and overflow monitoring devices cannot effectively monitor the tripping and running-in conditions. The high temperature and high pressure environment downhole causes flow measurement deviations, lacks early warning capabilities, and fixed thresholds lack adaptability to operating conditions, resulting in monitoring accuracy and early warning lag.

Method used

A real-time correction module for mud thermal expansion and compressibility is introduced, a drill cuttings load estimation module is set up, and a time-series neural network leakage trend prediction module is constructed to achieve adaptive monitoring under all working conditions. Prediction is performed by combining multi-parameter fusion and dynamic threshold adjustment with a time-series neural network.

Benefits of technology

It achieves high-precision monitoring of all drilling conditions, eliminates blind spots in tripping and running-out monitoring, improves flow measurement accuracy, enables early warning, reduces false alarm and missed alarm rates, and adapts to normal fluctuations in different well depths and drill string combinations.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123109640A_ABST
    Figure CN123109640A_ABST
Patent Text Reader

Abstract

This invention belongs to the field of oil drilling safety monitoring technology, specifically relating to a method and device for real-time monitoring of well leakage and overflow that can simultaneously adapt to four drilling conditions: normal drilling, dry well stagnation, tripping operation, and running-in operation. Based on existing well leakage and overflow monitoring for drilling and dry well stagnation conditions, this invention adds dedicated monitoring logic for tripping and running-in operations. It uses a communication module to link with the on-site drilling parameter instrument to acquire drill string data in real time during the tripping and running-in process. A real-time correction model for mud thermal expansion and compressibility is introduced to compensate for temperature and pressure changes in flow rate monitoring data; a drill cuttings load estimation module is introduced to correct for drill cuttings changes in the theoretical mud return volume during running-in operations; a time-series neural network leakage trend prediction module is introduced to provide early warning of leakage risks; and an adaptive deviation threshold that dynamically changes with well depth is introduced to overcome the high false alarm rate of fixed thresholds. This invention achieves comprehensive, blind-spot-free monitoring of the four core drilling operation conditions.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention belongs to the field of oil drilling safety monitoring technology, specifically relating to a method and device for real-time monitoring of well leakage and overflow that can be adapted to four drilling conditions simultaneously: normal drilling, dry well standby, tripping operation, and running-in operation. Background Technology

[0002] During oil drilling operations, well leakage and well overflow are significant causes of major downhole safety accidents, easily triggering catastrophic events such as well kicks and blowouts, seriously threatening drilling safety and the safety of personnel and equipment. As oil and gas drilling increasingly extends into deeper and more complex formations, the downhole high-temperature and high-pressure environment is becoming increasingly harsh, placing higher demands on the accuracy and timeliness of well leakage and overflow monitoring.

[0003] Currently, commonly used methods for monitoring lost circulation and overflow in drilling sites mainly include mud pit level monitoring, inlet / outlet flow difference monitoring, and comprehensive logging parameter monitoring. However, existing technologies still have the following significant shortcomings: (i) There are blind spots in the monitoring during tripping and running out of the drilling site. Existing wellbore leakage and overflow monitoring devices can only monitor and determine fluid level and flow rate under normal drilling circulation and dry well static conditions. They cannot effectively monitor the high-frequency and high-risk tripping and running-in conditions in drilling operations. During tripping and running-in, changes in drill string volume cause regular changes in the wellbore mud level and inlet / outlet mud flow. Traditional equipment cannot distinguish between normal drill string displacement fluid level changes and actual abnormal changes in wellbore leakage and overflow, resulting in long-term monitoring blind spots. Insufficient grouting during tripping and hidden overflows during running-in cannot be identified in a timely manner, which can easily induce complex downhole accidents. For example, although patent CN2055532U mentions monitoring the tripping and running-in process, it still uses the traditional fluid level and flow rate difference discrimination method, which cannot effectively eliminate the interference of normal fluid level fluctuations caused by drill string tripping, resulting in high false alarm and false alarm rates.

[0004] (ii) The high temperature and pressure downhole caused the flow measurement deviation to be uncorrected. In deep and ultra-deep wells, wellbore temperatures can exceed 150℃ and pressures can exceed 100MPa. Under these high-temperature, high-pressure conditions, drilling mud undergoes significant thermal expansion and compressible volume changes, leading to a systematic deviation between the mud flow rate data collected by surface flowmeters and the actual downhole flow rate. Existing monitoring methods do not consider this factor when calculating the theoretical grouting / return volume during tripping, directly resulting in an inaccurate comparison between theoretical and actual values, causing false alarms or missed alarms. For example, while the 2011 publication "Design of Automatic Continuous Grouting Device" describes a technique for judging downhole conditions by comparing the displacement volume with the actual injection volume, it uses raw data without temperature and pressure correction, which has significant errors under the high-temperature, high-pressure conditions of deep wells.

[0005] (iii) The impact of drill cuttings load was not considered during the drilling operation. During drilling, drill cuttings generated by the drill bit breaking rock continuously enter the wellbore and occupy an effective volume, directly affecting the amount of drilling mud returned from the wellbore. Existing drilling condition monitoring methods only calculate the theoretical mud return volume based on the drill string displacement volume, without considering the contribution of drill cuttings volume to the mud return volume. This results in the theoretical mud return volume being systematically lower than the actual value, making it difficult to accurately determine downhole anomalies.

[0006] (iv) Lack of early warning capabilities Existing monitoring methods are mostly "post-event alarm" type—warnings are only triggered after well leakage or overflow has occurred and abnormal parameters exceed preset thresholds. For seepage-type leaks and early overflows that are highly concealed and develop rapidly, existing methods are often untimely in detection, resulting in delayed warnings. Although a few studies have attempted to use methods such as artificial neural networks for overflow identification, these are mostly still in the theoretical verification stage and have not yet formed a systematic and integrated monitoring scheme with physical model correction, multi-condition adaptive judgment, etc.

[0007] (v) Fixed warning thresholds and lack of adaptability to operating conditions Most existing technologies use fixed alarm thresholds, which cannot be dynamically adjusted according to changes in operating parameters such as well depth, drill string assembly, and mud properties. The amplitude of normal flow fluctuations varies significantly under different well depths and drill string assemblies. Fixed thresholds are prone to false alarms in shallow wells and false alarms in deep wells, seriously affecting the field practicality of the monitoring system.

[0008] In summary, existing technologies cannot achieve uninterrupted, high-precision leakage monitoring throughout the entire drilling operation cycle, resulting in significant deficiencies in safety control. Summary of the Invention

[0009] The purpose of this invention is to provide an integrated intelligent monitoring method and device for well leakage and overflow that can simultaneously adapt to four drilling conditions: normal drilling, dry well stagnation, tripping operations, and running-in operations, addressing the problems existing in the prior art. This invention solves the technical problem of flow measurement deviation caused by the high temperature and pressure environment downhole by introducing a real-time correction module for mud thermal expansion and compressibility; it addresses the problem of insufficient accuracy in calculating theoretical mud return volume during running-in operations by setting a drill cuttings load estimation module; it achieves a leap from "post-event alarm" to "early warning" by constructing a time-series neural network leakage trend prediction module; and it solves the defects of high false alarm rate and high missed alarm rate of fixed threshold by setting an adaptive deviation threshold that dynamically changes with well depth.

[0010] The technical solution of this invention is: A method for integrated intelligent monitoring of well leakage and overflow under all drilling conditions includes the following steps: S1. The device is powered on and initialized, establishes a data communication connection with the drilling parameter instrument, synchronizes drilling site construction condition data in real time, and starts the wellbore temperature acquisition component and wellbore pressure acquisition component to collect downhole temperature and pressure data in real time. S2. The main control unit identifies the current construction conditions and automatically matches the corresponding well leakage and overflow monitoring and judgment method. S3. Based on the temperature and pressure data collected by the wellbore temperature acquisition component and the wellbore pressure acquisition component, perform real-time correction and compensation for thermal expansion effect and compressibility effect on the inlet flow monitoring data and the outlet flow monitoring data. S4. If it is a normal drilling cycle, use the corrected difference between the inlet and outlet mud flow rates combined with the change in the mud level to determine the abnormality. S5. If the well is in a static condition, a single wellbore mud level continuous monitoring method is adopted, and the level data is corrected by combining temperature and pressure data to complete the anomaly judgment. S6. If it is a tripping operation, retrieve the volume data of the tripped drill string and calculate the theoretical grouting volume after thermal expansion and compressibility correction by combining it with real-time temperature and pressure data. Compare it with the actual grouting volume after the same correction, calculate the deviation rate and compare it with the adaptive threshold that dynamically changes with well depth to complete the leakage judgment. S7. If it is a drilling operation, retrieve the volume data of the drill string and estimate the volume of drill cuttings based on the drilling speed data and formation lithology data. Calculate the theoretical slurry return volume based on the sum of the drill string volume and the drill cuttings volume, combined with real-time temperature and pressure data. Compare this with the actual slurry return volume after the same correction to complete the leakage judgment. S8. Input the corrected multi-parameter time series data into the time series neural network prediction model and output the probability of leakage risk in future periods. S9. When a well leakage or overflow abnormality is detected, or the predicted risk probability exceeds the preset threshold, the corresponding level of early warning is activated according to the abnormality level to complete the full-condition safety monitoring operation.

[0011] Specifically, the real-time correction and compensation model for the thermal expansion effect and compressibility effect mentioned in step S3 is as follows: V 修正 = V 实测 × [1 + α(T − T0) − β(P − P0)] Among them, V 实测 α is the measured value of mud flow rate collected by the flow monitoring component, T is the real-time temperature collected by the wellbore temperature acquisition component, T0 is the surface standard temperature, β is the mud compressibility coefficient, P is the real-time pressure collected by the wellbore pressure acquisition component, and P0 is the surface standard pressure.

[0012] Furthermore, the leakage determination for the tripping operation condition in step S6 specifically includes: calculating the displacement volume V of the tripped drill string based on the number of drill pipes and drill string specifications transmitted in real time by the drilling parameter instrument. string The corrected theoretical grouting volume was calculated by combining real-time temperature and pressure data. V 理论 = V string × [1 + α(T − T0) − β(P − P0)]; The actual grouting volume V collected by the inlet flow monitoring component 实际 After applying the same temperature and pressure correction, V is obtained. 实际修正 ; Calculate the deviation rate ε = (V 实际修正 - V 理论 ) / V 理论 ; Obtain the adaptive deviation threshold ε corresponding to the current well depth H. 阈值 ; When ε < −ε 阈值 When ε > +ε, it is determined to be a well leakage. 阈值 It was determined to be an overflow.

[0013] Furthermore, the leakage determination for the drilling operation conditions described in step S7 specifically includes: calculating the displacement volume V of the drill string based on the specifications and quantity data of the drill string transmitted by the drilling parameter instrument. string ; Based on drilling speed data (ROP) and formation lithology data, according to V cuttings = ROP×A hole ×t×η real-time estimation of drill cuttings volume, where A hole Let V be the wellbore cross-sectional area, t be the drilling time, and η be the cuttings expansion coefficient; let V be the wellbore cross-sectional area. string + V cuttings Based on this, and combined with real-time temperature and pressure data, the corrected theoretical slurry return volume is calculated: V 理论 = (V string + V cuttings ) × [1 + α(T − T0) − β(P − P0)]; The actual return flow V collected by the outlet flow monitoring component 实际 After applying the same temperature and pressure correction, V is obtained. 实际修正 ; V 实际修正 With V 理论 Compare, when V 实际修正 Significantly lower than V 理论 Furthermore, if the value exceeds a preset threshold, it is considered a well leak. 实际修正 Significantly higher than V 理论And when the value exceeds the preset threshold, it is determined to be an overflow.

[0014] Specifically, the temporal neural network prediction model described in step S8 adopts an Attention-GRU network. The model input is multi-parameter time series data corrected for temperature and pressure over the past N time steps, including the corrected inlet and outlet flow difference sequence, wellbore liquid level change sequence, and operating condition sequence. The Attention layer automatically identifies the key time points and key parameters that contribute the most to the leakage prediction, and the GRU layer captures the long-term dependencies of the time series data. Finally, the Sigmoid activation function outputs the leakage risk probability within the next M minutes.

[0015] The training method of the time-series neural network prediction model is as follows: using the historical inlet and outlet flow difference sequence after temperature and pressure correction, the wellbore liquid level change sequence and the operating condition sequence as training samples, and using whether an overflow accident occurred in the corresponding time period as a label, the time-series neural network model is trained in a supervised learning manner until the prediction accuracy of the model on the validation set reaches a preset standard (such as above 95%).

[0016] Once well leakage is detected, a leakage type identification program is initiated. Based on the shape of the leakage rate change curve, the characteristics of the wellbore fluid level drop rate, and the time-varying pattern of the inlet and outlet flow rate difference, well leakage is classified into seepage type leakage, fracture type leakage, and karst type leakage, and corresponding treatment suggestions are output.

[0017] This invention also provides an integrated intelligent monitoring device for well leakage and overflow under all drilling conditions, comprising: The system includes a mud inlet flow monitoring component, a mud outlet flow monitoring component, a wellbore liquid level acquisition component, a wellbore temperature acquisition component, a wellbore pressure acquisition component, a drilling parameter instrument signal communication component, a main control data analysis and processing unit, and an early warning and prompting component. The main control data analysis and processing unit has the following built-in features: The real-time correction module for thermal expansion and compressibility of mud is used to perform real-time correction and compensation for thermal expansion and compressibility effects on the mud flow data collected by the mud inlet flow monitoring component and the mud outlet flow monitoring component based on the real-time temperature data and real-time pressure data collected by the wellbore temperature acquisition component and the wellbore pressure acquisition component, and to generate corrected inlet flow data and corrected outlet flow data. The cuttings load estimation module is used to estimate the volume of cuttings in the wellbore in real time based on the drilling speed data and formation lithology data transmitted by the drilling parameter instrument through the drilling parameter instrument signal communication component, and to correct the theoretical return volume of the drilling conditions accordingly. The multi-parameter fusion working condition identification and anomaly judgment module has built-in normal drilling cycle judgment logic, empty well static judgment logic, drilling start operation mud volume comparison judgment logic, and drilling start operation mud volume comparison judgment logic. It can automatically switch the corresponding monitoring logic according to the construction working condition signal received by the drilling parameter instrument signal communication component. The time-series neural network leakage trend prediction module takes the historical inlet and outlet flow difference sequence, wellbore fluid level change sequence, and operating condition sequence after being corrected by the mud thermal expansion and compressibility real-time correction module as input. The trained time-series neural network model predicts the leakage risk probability in the future period and issues an early warning when the predicted probability exceeds a preset threshold. The leakage type identification module classifies well leakage into seepage-type leakage, fracture-type leakage, and cavern-type leakage based on the shape of the leakage rate change curve, the characteristics of the wellbore fluid level drop rate, and the time-varying law of the inlet and outlet flow difference when well leakage occurs, and outputs corresponding treatment suggestions.

[0018] The multi-parameter fusion working condition identification and anomaly judgment module calculates the theoretical grouting volume based on the drill string volume data corrected by the real-time correction module for mud thermal expansion and compressibility during tripping operations, and compares it with the actual grouting volume after the same correction. The judgment threshold dynamically changes with the current well depth. During running-in operations, it calculates the theoretical return grouting volume based on the corrected drill string volume data and deducts the drill cuttings volume estimated by the drill cuttings load estimation module, and compares it with the actual return grouting volume after the same correction. When an anomaly is detected, an early warning signal is issued through the early warning prompt component.

[0019] Both the mud inlet flow monitoring component and the mud outlet flow monitoring component are electromagnetic flow meters. The mud outlet flow monitoring component is preferably a non-full pipe flow meter to adapt to the accurate measurement of non-full pipe flow during tripping and drilling operations.

[0020] The drilling parameter instrument signal communication component supports any one or more combinations of RS485 communication protocol, Ethernet communication protocol and wireless data transmission communication protocol, and is used to adapt to data interface of various well site drilling parameter instruments.

[0021] The early warning component is equipped with three levels of abnormal warning modes, corresponding to three risk levels: minor abnormality, moderate abnormality, and severe leakage fault. Different risk levels correspond to different audible and visual alarm frequencies and linkage control strategies.

[0022] The correction model of the real-time correction module for the thermal expansion and compressibility of mud is as follows: V 修正 = V 实测 × [1 + α(T − T0) − β(P − P0)] Among them, V 实测 α is the measured value of mud flow rate collected by the flow monitoring component, T is the real-time temperature collected by the wellbore temperature acquisition component, T0 is the surface standard temperature, β is the mud compressibility coefficient, P is the real-time pressure collected by the wellbore pressure acquisition component, and P0 is the surface standard pressure.

[0023] In the drilling mud volume comparison and judgment logic, the judgment threshold that dynamically changes with well depth is... ε 阈值 The threshold ε is determined as follows: Based on the inherent volumetric error of the drill string assembly and the fluctuation range of mud properties at different well depths, the upper limit of the normal deviation corresponding to each well depth is pre-calculated to form a threshold-well depth mapping table. During actual judgment, the corresponding ε is obtained by looking up the table in real time according to the current well depth H. 阈值 .

[0024] The beneficial effects of this invention are: 1. Achieving full coverage monitoring of the four core drilling conditions: Achieving full coverage monitoring of the four core drilling conditions of drilling, empty well, tripping out, and running in, completely eliminating the blind spots of traditional tripping out monitoring devices.

[0025] 2. Pioneering Real-Time Correction for Mud Thermal Expansion and Compressibility: A flow correction model is constructed for the high-temperature and high-pressure environments of deep and ultra-deep wells, overcoming measurement deviations caused by non-overflow factors and significantly improving the accuracy of theoretical calculations and actual measurements during tripping. After thermal expansion correction, the flow measurement accuracy is improved by more than 15% compared to the uncorrected state.

[0026] 3. Pioneering real-time estimation and correction of drill cuttings load: The introduction of drill cuttings volume estimation during drilling operations compensates for the shortcomings of traditional methods that ignore the volume of drill cuttings occupying the wellbore, resulting in deviations in the calculation of theoretical slurry return volume, thus improving the monitoring accuracy of drilling operations by more than 20%.

[0027] 4. Pioneering a dual-drive architecture of "physical model correction + data-driven AI prediction": Based on high-quality data input provided by the physical correction model, a temporal neural network is introduced to achieve early prediction of leakage risks. Experimental verification shows that combining temperature and pressure correction with the temporal neural network model results in zero overflow leakage reports and provides early warnings approximately 5-10 minutes in advance.

[0028] 5. Pioneering adaptive threshold that dynamically changes with well depth: Overcomes the shortcomings of fixed thresholds, which are prone to false alarms in shallow wells and false alarms in deep wells. Adapts to normal fluctuations under different well depths and different drill string combinations, reducing the false alarm rate by more than 30% compared to fixed threshold schemes.

[0029] 6. Pioneering intelligent identification of leakage types: It can automatically distinguish between three leakage types: seepage, cracks, and karst caves, providing a basis for decision-making on taking differentiated treatment measures on site.

[0030] 7. Integrated integration and strong compatibility: It can link with existing drilling parameter instruments, no large-scale transformation of well site equipment is required, featuring convenient installation and high on-site applicability. The communication module supports multiple transmission protocols including RS485, Ethernet and wireless data transmission, and can seamlessly connect with mainstream brand drilling parameter instruments on the market.

[0031] 8. Graded early warning and accurate disposal: A three-level safety early warning mechanism is set up, corresponding to three risk levels of minor abnormality, moderate abnormality and severe lost circulation and overflow fault, and different risk levels correspond to different sound and light alarm frequencies and linkage control strategies. Brief Description of the Drawings

[0032] Figure 1 is a schematic structural diagram of the present invention; Figure 2 is a schematic process flow diagram of the present invention. Detailed Description of the Embodiments

[0033] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments. Embodiment 1

[0034] This embodiment provides an integrated intelligent monitoring device for lost circulation and overflow under all drilling working conditions, the structure of which is as Figure 1 shown, comprising: The whole monitoring device is deployed in the manifold area supporting the drilling wellhead, and includes a mud inlet flow monitoring component, a mud outlet flow monitoring component, a wellbore liquid level acquisition component, a wellbore temperature acquisition component, a wellbore pressure acquisition component, a drilling parameter instrument signal communication component, a main control data analysis and processing unit, and an early warning prompt component.

[0035] The hardware composition includes a mud inlet electromagnetic flowmeter, a mud outlet electromagnetic flowmeter (preferably a non-full pipe flowmeter), a wellbore liquid level sensor, a wellbore temperature sensor, a wellbore pressure sensor, a wireless / wired communication docking module, an industrial-grade main control control cabinet, an on-site sound and light early warning alarm, and a working condition data display terminal.

[0036] The communication module of this device supports multiple transmission protocols such as RS485, Ethernet and wireless data transmission, can seamlessly connect with mainstream brand drilling parameter instruments on the market, and does not need to modify the original well site measurement and control lines.

[0037] All analog signals collected by the flow sensors, liquid level sensors, temperature sensors and pressure sensors are uniformly converted into digital signals through a 24-bit high-precision ADC, and then sent to the main control unit for combined noise reduction processing of Kalman filtering and median filtering, so as to eliminate data interference caused by drilling site vibration, mud impurities and pipeline pressure fluctuations.

[0038] The device is equipped with an automatic operating condition recognition program, which can automatically and seamlessly switch between four monitoring modes—drilling, open well, tripping out, and running in—based on the operating status signals transmitted by the drilling parameter instrument (such as hook load, drill string movement status, mud pump on / off status, etc.). No manual switching or control is required, resulting in a high degree of automation.

[0039] The main control unit adopts an industrial-grade embedded industrial control computer, equipped with no less than 8GB of memory and 256GB of solid-state drive, to meet the computing power requirements for real-time data acquisition (sampling frequency no less than 1Hz), real-time correction calculation, and neural network model inference (inference latency no more than 100ms). All monitoring data and early warning records are automatically stored locally and synchronously uploaded to the drilling remote monitoring platform, realizing data traceability and playback.

[0040] During normal drilling operations, the device connects in real time to the existing drilling parameter instrument data interface at the well site, enabling two-way data exchange for drilling conditions. The overall power supply for the equipment adopts the well site's industrial regulated power supply, which is suitable for the complex working conditions of field drilling.

[0041] The main control data analysis and processing unit has the following built-in features: The real-time correction module for thermal expansion and compressibility of mud is used to perform real-time correction and compensation for thermal expansion and compressibility effects on the mud flow data collected by the mud inlet flow monitoring component and the mud outlet flow monitoring component based on the real-time temperature data and real-time pressure data collected by the wellbore temperature acquisition component and the wellbore pressure acquisition component, and to generate corrected inlet flow data and corrected outlet flow data. The cuttings load estimation module is used to estimate the volume of cuttings in the wellbore in real time based on the drilling speed data and formation lithology data transmitted by the drilling parameter instrument through the drilling parameter instrument signal communication component, and to correct the theoretical return volume of the drilling conditions accordingly. The multi-parameter fusion working condition identification and anomaly judgment module has built-in normal drilling cycle judgment logic, empty well static judgment logic, drilling start operation mud volume comparison judgment logic, and drilling start operation mud volume comparison judgment logic. It can automatically switch the corresponding monitoring logic according to the construction working condition signal received by the drilling parameter instrument signal communication component. The time-series neural network leakage trend prediction module takes the historical inlet and outlet flow difference sequence, wellbore fluid level change sequence, and operating condition sequence after being corrected by the mud thermal expansion and compressibility real-time correction module as input. The trained time-series neural network model predicts the leakage risk probability in the future period and issues an early warning when the predicted probability exceeds a preset threshold. The leakage type identification module classifies well leakage into seepage-type leakage, fracture-type leakage, and cavern-type leakage based on the shape of the leakage rate change curve, the characteristics of the wellbore fluid level drop rate, and the time-varying law of the inlet and outlet flow difference when well leakage occurs, and outputs corresponding treatment suggestions.

[0042] The multi-parameter fusion working condition identification and anomaly judgment module calculates the theoretical grouting volume based on the drill string volume data corrected by the real-time correction module for mud thermal expansion and compressibility during tripping operations, and compares it with the actual grouting volume after the same correction. The judgment threshold dynamically changes with the current well depth. During running-in operations, it calculates the theoretical return grouting volume based on the corrected drill string volume data and deducts the drill cuttings volume estimated by the drill cuttings load estimation module, and compares it with the actual return grouting volume after the same correction. When an anomaly is detected, an early warning signal is issued through the early warning prompt component.

[0043] Both the mud inlet flow monitoring component and the mud outlet flow monitoring component are electromagnetic flow meters. The mud outlet flow monitoring component is preferably a non-full pipe flow meter to adapt to the accurate measurement of non-full pipe flow during tripping and drilling operations.

[0044] The drilling parameter instrument signal communication component supports any one or more combinations of RS485 communication protocol, Ethernet communication protocol and wireless data transmission communication protocol, and is used to adapt to data interface of various well site drilling parameter instruments.

[0045] The early warning component is equipped with three levels of abnormal warning modes, corresponding to three risk levels: minor abnormality, moderate abnormality, and severe leakage fault. Different risk levels correspond to different audible and visual alarm frequencies and linkage control strategies. Example 2

[0046] This embodiment provides an integrated intelligent monitoring method for well leakage and overflow under all drilling conditions, the process of which is as follows: Figure 2 As shown, it includes the following steps: S1. The device is powered on and initialized, establishes a data communication connection with the drilling parameter instrument, synchronizes drilling site construction condition data in real time, and starts the wellbore temperature acquisition component and wellbore pressure acquisition component to collect downhole temperature and pressure data in real time. S2. The main control unit identifies the current construction conditions and automatically matches the corresponding well leakage and overflow monitoring and judgment method. S3. Based on the temperature and pressure data collected by the wellbore temperature acquisition component and the wellbore pressure acquisition component, perform real-time correction and compensation for thermal expansion effect and compressibility effect on the inlet flow monitoring data and the outlet flow monitoring data. S4. If it is a normal drilling cycle, use the corrected difference between the inlet and outlet mud flow rates combined with the change in the mud level to determine the abnormality. S5. If the well is in a static condition, a single wellbore mud level continuous monitoring method is adopted, and the level data is corrected by combining temperature and pressure data to complete the anomaly judgment. S6. If it is a tripping operation, retrieve the volume data of the tripped drill string and calculate the theoretical grouting volume after thermal expansion and compressibility correction by combining it with real-time temperature and pressure data. Compare it with the actual grouting volume after the same correction, calculate the deviation rate and compare it with the adaptive threshold that dynamically changes with well depth to complete the leakage judgment. S7. If it is a drilling operation, retrieve the volume data of the drill string and estimate the volume of drill cuttings based on the drilling speed data and formation lithology data. Calculate the theoretical slurry return volume based on the sum of the drill string volume and the drill cuttings volume, combined with real-time temperature and pressure data. Compare this with the actual slurry return volume after the same correction to complete the leakage judgment. S8. Input the corrected multi-parameter time series data into the time series neural network prediction model and output the probability of leakage risk in future periods. S9. When a well leakage or overflow abnormality is detected, or the predicted risk probability exceeds the preset threshold, the corresponding level of early warning is activated according to the abnormality level to complete the full-condition safety monitoring operation.

[0047] The method provided in this embodiment can achieve full coverage monitoring of the four core drilling conditions: drilling, empty well, tripping, and running-in drilling, completely eliminating the blind spots in the traditional tripping and running-in monitoring. Example 3

[0048] This embodiment provides a specific implementation of the mud thermal expansion and compressibility correction module. The thermal expansion coefficient α and compressibility coefficient β of drilling mud need to be determined in advance through laboratory experiments before drilling operations begin. The specific method is as follows: take actual mud samples used on site, and use a high-temperature and high-pressure rheometer to measure the mud volume change rate under different temperature and pressure conditions within the range of simulated downhole temperature (room temperature to 200℃) and pressure (0.1MPa to 150MPa), and fit the functional relationship between α(T) and β(P).

[0049] During actual monitoring, the wellbore temperature acquisition component and the wellbore pressure acquisition component collect downhole temperature and pressure data in real time at a sampling frequency of no less than 1 time / second, and the main control unit corrects each set of flow data in real time accordingly.

[0050] The corrected model is: V 修正 = V 实测 × [1 + α(T − T0) − β(P − P0)] Among them, V 实测The measured mud flow rate is represented by the flow monitoring component, T is the real-time downhole temperature acquired by the wellbore temperature acquisition component, α is the mud thermal expansion coefficient, β is the mud compressibility coefficient, T0 is the surface standard temperature (usually taken as 20℃), P is the real-time downhole pressure acquired by the wellbore pressure acquisition component, and P0 is the surface standard pressure (usually taken as 0.1MPa). This correction model simultaneously considers the thermal expansion effect (volume expansion due to temperature increase) and compressibility effect (volume compression due to pressure increase) of the mud, achieving a bidirectional and accurate restoration of the actual downhole flow rate. Example 4

[0051] This embodiment specifically illustrates the implementation of normal drilling circulation conditions. When drilling is in continuous drilling and mud circulation mode, the main control unit automatically switches to circulation monitoring mode. The inlet flow meter collects the instantaneous and cumulative flow of mud pumped into the wellbore in real time, the outlet flow meter collects the flow data of mud returned from the wellhead synchronously, and the level sensor collects the dynamic change data of the static liquid level of mud in the wellbore in real time.

[0052] The main control unit first calls the thermal expansion and compressibility correction module to correct the inlet and outlet flow rate data in real time, then calculates the difference between the corrected inlet and outlet mud flow rates, and makes a comprehensive judgment based on the fluid level rise and fall rate. When the outlet flow rate is consistently less than the inlet flow rate and the fluid level rises synchronously, it is determined that a well leakage has occurred downhole; when the outlet flow rate is consistently greater than the inlet flow rate and the fluid level rises synchronously, it is determined that a formation overflow has occurred. An early warning prompt is issued immediately when the preset warning threshold is reached. Example 5

[0053] This embodiment specifically illustrates the implementation of the empty well static condition. When the drilling is completed and no drill string is lowered into the wellbore, the system enters the dry well static observation stage. The system automatically shuts down the flow comparison logic and activates the pure liquid level monitoring mode.

[0054] The level sensor continuously collects the mud level in the wellbore at a sampling frequency of no less than 0.5 times per second. Simultaneously, temperature and pressure sensors collect temperature and pressure data to correct the mud level data. A standard allowable range for static mud level fluctuations is set. If the mud level continues to slowly decrease and exceeds the allowable range, it is determined that there is a formation leakage or well leakage problem; if the mud level continues to rise abnormally, it is determined that formation fluid is surging and overflowing. The entire process achieves unattended static safety monitoring. Example 6

[0055] This example illustrates the specific implementation of the drilling operation conditions. During the tripping operation, the drilling parameter instrument transmits data such as tripping speed, number of drill pipes tripped, specifications and dimensions of a single drill pipe, total length of drill string tripped, and overall displacement volume of the drill string to the main control cabinet in real time via the communication module.

[0056] The main control unit has a pre-stored database of volume parameters for various specifications of drill pipes and drill collars, which can quickly calculate the volume Vstring of the drill string to be pulled out.

[0057] Meanwhile, the wellbore temperature and pressure acquisition components collect the current downhole temperature T and pressure P in real time. The main control unit calls the thermal expansion and compressibility correction module to calculate the theoretical grouting volume after temperature and pressure correction. V 理论 = V string ×[1 + α(T−T0) − β(P − P0)] The operators simultaneously start the wellhead grouting system, and the inlet flow meter accurately measures the actual total volume of mud injected into the wellbore, Vactual, at a sampling frequency of no less than 1 time / second. After correction by the same temperature and pressure, Vactual is obtained.

[0058] The system calculates the deviation rate ε in real time: ε = (V 实际修正 -V 理论 ) / V 理论 And the adaptive deviation threshold ε corresponding to the current well depth H. 阈值 Perform dynamic comparison.

[0059] Adaptive deviation threshold ε 阈值 The determination method is as follows: based on the inherent volumetric error of drill string assemblies and the fluctuation range of mud properties at different well depths, the upper limit of the normal deviation corresponding to each well depth is pre-calculated to form a threshold-well depth mapping table. During actual judgment, the main control unit looks up the corresponding ε threshold (H) in real time according to the current well depth H.

[0060] Under standard operating conditions without formation anomalies, the two values ​​are basically the same, and the deviation rate is within the threshold range. However, if the actual amount of mud injected is significantly less than the theoretically calculated value (ε < −ε), then… 阈值 This indicates abnormal mud loss within the wellbore, accurately indicating well leakage during the tripping process; if the actual amount of mud injected is significantly greater than the theoretically calculated value (ε> +ε), it indicates well leakage. 阈值 This indicates that fluid is flowing backwards into the wellbore from the formation, occupying space, and is determined to be an overflow during the tripping operation. Abnormal data is uploaded to the terminal in real time and a tiered alarm is triggered. Example 7

[0061] This embodiment illustrates the specific implementation of the drilling operation. During the drilling operation, the drilling parameter instrument transmits data in real time, including the number of drill strings, the depth of the drill strings, and the effective volume of the wellbore occupied by the drill strings.

[0062] The main control system first calculates the displacement volume V of the drill string based on its specifications and quantity. stringMeanwhile, the drill cuttings load estimation module, based on the real-time ROP (Rate of Exploration) and formation lithology data transmitted from the drilling parameter instrument, calculates the load according to V... cuttings = ROP×A hole ×t×η real-time estimation of drill cuttings volume, where A hole Where t is the cross-sectional area of ​​the wellbore, η is the drilling time, and η is the cuttings expansion coefficient (pre-set according to the formation lithology, usually taken as 1.2~1.5).

[0063] The main control system is based on (V) string + V cuttings Based on this, and combined with real-time wellbore temperature and pressure data, the theoretical return slurry volume corrected for temperature and pressure is calculated using a thermal expansion and compressibility correction module. V theory = (V string + V cuttings )×[1 +α(T−T0)−β(P−P0)] An outlet flow meter (preferably a non-full-pipe flow meter) is installed at the wellhead return pipeline to collect the actual total amount of mud returned, V, throughout the entire drilling process. 实际 After correction for the same temperature and pressure, V is obtained. 实际修正 .

[0064] The system automatically compares two sets of data in real time. If the actual amount of mud returned during drilling is lower than the theoretically calculated amount, it indicates that some mud has leaked out of the formation, which is identified as a well leakage fault. If the actual amount of mud returned during drilling is higher than the theoretically calculated value, it indicates that formation fluids are continuously invading the wellbore, causing an increase in mud overflow, which is accurately identified as a potential overflow hazard. This invention completely distinguishes between normal mud discharge caused by drill string running and abnormal downhole leakage. Example 8

[0065] This embodiment illustrates the specific implementation of the time-series neural network overflow trend prediction module. The training of the temporal neural network model was completed offline before the equipment was put into use. The training samples were derived from historical drilling data, including the inlet and outlet flow rate difference sequence after temperature and pressure correction, the wellbore fluid level change sequence, the operating condition sequence, and labels indicating whether a leakage accident occurred during the corresponding period.

[0066] Taking the Attention-GRU network as an example: The model input layer receives multi-parameter time-series data from the past N time steps (preferably N=60, with a 1-minute interval between each step), including the corrected inlet and outlet flow rate difference sequence, wellbore fluid level change sequence, and operating condition sequence. The Attention layer automatically identifies the key time points and key parameters that contribute most to the leakage prediction by calculating the attention weights for each time step. The GRU layer (with a hidden layer dimension preferably 128) captures the long-term and short-term dependencies of the time-series data. The output layer outputs the leakage risk probability (0~1) within the next M minutes (preferably M=5~10 minutes) through the Sigmoid activation function.

[0067] The model is trained using supervised learning, with cross-entropy as the loss function and Adam as the optimizer, until the model's prediction accuracy on the validation set reaches a preset standard (e.g., above 95%).

[0068] After the device is put into operation in the field, real-time multi-parameter data, corrected for temperature and pressure, is continuously input into the model, which then outputs the probability of leakage risk for future periods. When the probability exceeds a preset threshold (e.g., 0.7), the system issues an early warning. Example 9

[0069] This embodiment illustrates the specific implementation of the leakage type identification module. When the device detects well leakage, the leakage type identification module automatically activates. The module has built-in feature templates for three leakage types: Leakage-type loss: The loss rate is slow (<0.5 m³ / h), continuous and stable, the difference between the inlet and outlet flow rates increases slowly, and the fluid level drops slowly and continuously. This is commonly seen in permeable formations. Recommendations: Increase mud viscosity and appropriately reduce drilling speed.

[0070] Fracture-type leakage: The leakage rate increases sharply and then tends to stabilize (1~5 m³ / h). The difference between the inlet and outlet flow rates increases rapidly and then remains at a certain level. The fluid level drops rapidly and then tends to level off. This is commonly seen in fractured formations. Recommended treatment: Add leak-proofing agent while drilling and appropriately increase the flow rate.

[0071] Cavernous leakage: The leakage rate increases sharply (>5 m³ / h), the difference between the inlet and outlet flow rates increases precipitously, and the liquid level drops precipitously. This is commonly seen in cavernous formations. Recommended treatment: Immediately stop drilling and use cement grout to plug the leak.

[0072] The module extracts the characteristic parameters (peak value, slope, steady-state value, etc.) of the leakage rate change curve, the liquid level drop rate curve, and the time-varying curve of the inlet and outlet flow difference after leakage occurs, performs pattern matching with three feature templates, and outputs the recognition results and corresponding handling suggestions. Example 10

[0073] This embodiment specifically illustrates the setting of early warning levels. This embodiment sets up a three-level security early warning mechanism: Level 1 Warning (Minor Anomaly): The abnormal data is only displayed in a pop-up window on the local terminal, reminding on-site personnel to closely observe the working conditions; Level 2 Warning (Moderate Anomaly): Activate the on-site low-frequency audible and visual alarm (flash frequency 1Hz, volume 80dB) and simultaneously push the abnormal information to the drilling remote monitoring platform; Level 3 Early Warning (Severe Leakage Fault): High-frequency audible and visual alarm (flash frequency 5Hz, volume 100dB), linked to the drilling control system to issue a safety warning, reminding the site to stop tripping in and out of the well, stop the pump to investigate downhole faults, and avoid major safety accidents such as blowouts and well collapses.

[0074] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them; although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications can still be made to the specific implementation of the present invention or equivalent substitutions can be made to some technical features without departing from the spirit of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the technical solutions claimed in the present invention.

Claims

1. A method for integrated intelligent monitoring of well leakage and overflow under all drilling conditions, comprising an integrated intelligent monitoring device for well leakage and overflow under all drilling conditions. Its features are, Includes the following steps: S1. Power on the intelligent monitoring device to complete initialization, establish a data communication connection with the drilling parameter instrument, synchronize drilling site construction condition data in real time, and start the wellbore temperature acquisition component and wellbore pressure acquisition component to collect downhole temperature and pressure data in real time. S2. The main control unit identifies the current construction conditions and automatically matches the corresponding well leakage and overflow monitoring and judgment method. S3. Based on the temperature and pressure data collected by the wellbore temperature acquisition component and the wellbore pressure acquisition component, perform real-time correction and compensation for thermal expansion effect and compressibility effect on the inlet flow monitoring data and the outlet flow monitoring data. S4. If it is a normal drilling cycle, use the corrected difference between the inlet and outlet mud flow rates combined with the change in the mud level to determine the abnormality. S5. If the well is in a static condition, a single wellbore mud level continuous monitoring method is adopted, and the level data is corrected by combining temperature and pressure data to complete the anomaly judgment. S6. If it is a tripping operation, retrieve the volume data of the tripped drill string and calculate the theoretical grouting volume after thermal expansion and compressibility correction by combining it with real-time temperature and pressure data. Compare it with the actual grouting volume after the same correction, calculate the deviation rate and compare it with the adaptive threshold that dynamically changes with well depth to complete the leakage judgment. S7. If it is a drilling operation, retrieve the volume data of the drill string and estimate the volume of drill cuttings based on the drilling speed data and formation lithology data. Calculate the theoretical slurry return volume based on the sum of the drill string volume and the drill cuttings volume, combined with real-time temperature and pressure data. Compare this with the actual slurry return volume after the same correction to complete the leakage judgment. S8. Input the corrected multi-parameter time series data into the time series neural network prediction model and output the probability of leakage risk in future periods. S9. When a well leakage or overflow abnormality is detected, or the predicted risk probability exceeds the preset threshold, the corresponding level of early warning is activated according to the abnormality level to complete the full-condition safety monitoring operation.

2. The monitoring method according to claim 1, characterized in that, The model used for real-time correction and compensation of thermal expansion effect and compressibility effect in step S3 is as follows: V 修正 = V 实测 × [1 + α(T − T0) − β(P − P0)] Among them, V 实测 α is the measured value of mud flow rate collected by the flow monitoring component, T is the real-time temperature collected by the wellbore temperature acquisition component, T0 is the surface standard temperature, β is the mud compressibility coefficient, P is the real-time pressure collected by the wellbore pressure acquisition component, and P0 is the surface standard pressure.

3. The monitoring method according to claim 1, characterized in that, The leakage determination for the tripping operation in step S6 specifically includes: calculating the displacement volume V of the tripped drill string based on the number of drill pipes and drill string specifications transmitted in real time by the drilling parameter instrument. string The corrected theoretical grouting volume was calculated by combining real-time temperature and pressure data. V 理论 = V string × [1 + α(T − T0) − β(P − P0)]; The actual grouting volume V collected by the inlet flow monitoring component 实际 After applying the same temperature and pressure correction, V is obtained. 实际修正 ; Calculate the deviation rate ε = (V 实际修正 - V 理论 ) / V 理论 ; Obtain the adaptive deviation threshold ε corresponding to the current well depth H. 阈值 ; When ε < −ε 阈值 When ε > +ε, it is determined to be a well leakage. 阈值 It was determined to be an overflow.

4. The monitoring method according to claim 1, characterized in that, The leakage determination for the drilling operation conditions described in step S7 specifically includes: calculating the displacement volume V of the drill string based on the specifications and quantity data of the drill string transmitted from the drilling parameter instrument. string ; Based on drilling speed data (ROP) and formation lithology data, according to V cuttings = ROP×A hole ×t×η real-time estimation of drill cuttings volume, where A hole Let V be the wellbore cross-sectional area, t be the drilling time, and η be the cuttings expansion coefficient; let V be the wellbore cross-sectional area. string + V cuttings Based on this, and combined with real-time temperature and pressure data, the corrected theoretical slurry return volume is calculated: V 理论 = (V string + V cuttings ) × [1 + α(T − T0) − β(P − P0)]; The actual return flow V collected by the outlet flow monitoring component 实际 After applying the same temperature and pressure correction, V is obtained. 实际修正 ; V 实际修正 With V 理论 Compare, when V 实际修正 Significantly lower than V 理论 Furthermore, if the value exceeds a preset threshold, it is considered a well leak. 实际修正 Significantly higher than V 理论 And when the value exceeds the preset threshold, it is determined to be an overflow.

5. The monitoring method according to claim 1, characterized in that, The temporal neural network prediction model described in step S8 adopts an Attention-GRU network. The model input is multi-parameter time series data corrected for temperature and pressure over the past N time steps, including the corrected inlet and outlet flow difference sequence, wellbore liquid level change sequence, and operating condition sequence. The Attention layer automatically identifies the key time points and key parameters that contribute the most to the leakage prediction, and the GRU layer captures the long-term dependencies of the time series data. Finally, the Sigmoid activation function outputs the leakage risk probability within the next M minutes.

6. The monitoring method according to claim 5, characterized in that, The training method of the time-series neural network prediction model is as follows: using the historical inlet and outlet flow difference sequence after temperature and pressure correction, the wellbore liquid level change sequence and the operating condition sequence as training samples, and using whether an overflow accident occurred in the corresponding time period as a label, the time-series neural network model is trained in a supervised learning manner until the prediction accuracy of the model on the validation set reaches the preset standard.

7. The monitoring method according to claim 1, characterized in that, Step S9 also includes: when well leakage is determined to have occurred, a leakage type identification program is initiated. Based on the shape of the leakage rate change curve when well leakage occurs, the characteristics of the wellbore fluid level drop rate, and the time-varying law of the inlet and outlet flow difference, well leakage is classified into seepage type leakage, fracture type leakage, and karst type leakage, and corresponding treatment suggestions are output.

8. An integrated intelligent monitoring device for well leakage and overflow under all drilling conditions, characterized in that, include: The system includes a mud inlet flow monitoring component, a mud outlet flow monitoring component, a wellbore liquid level acquisition component, a wellbore temperature acquisition component, a wellbore pressure acquisition component, a drilling parameter instrument signal communication component, a main control data analysis and processing unit, and an early warning and prompting component. The main control data analysis and processing unit has the following built-in features: The real-time correction module for thermal expansion and compressibility of mud is used to perform real-time correction and compensation for thermal expansion and compressibility effects on the mud flow data collected by the mud inlet flow monitoring component and the mud outlet flow monitoring component based on the real-time temperature data and real-time pressure data collected by the wellbore temperature acquisition component and the wellbore pressure acquisition component, and to generate corrected inlet flow data and corrected outlet flow data. The cuttings load estimation module is used to estimate the volume of cuttings in the wellbore in real time based on the drilling speed data and formation lithology data transmitted by the drilling parameter instrument through the drilling parameter instrument signal communication component, and to correct the theoretical return volume of the drilling conditions accordingly. The multi-parameter fusion working condition identification and anomaly judgment module has built-in normal drilling cycle judgment logic, empty well static judgment logic, drilling start operation mud volume comparison judgment logic, and drilling start operation mud volume comparison judgment logic. It can automatically switch the corresponding monitoring logic according to the construction working condition signal received by the drilling parameter instrument signal communication component. The multi-parameter fusion working condition identification and anomaly judgment module calculates the theoretical grouting volume based on the drill string volume data corrected by the real-time correction module for mud thermal expansion and compressibility during tripping operations, and compares it with the actual grouting volume after the same correction. The judgment threshold dynamically changes with the current well depth. During drilling operations, the module calculates the theoretical return grouting volume based on the corrected drill string volume data and deducts the drill cuttings volume estimated by the drill cuttings load estimation module, and compares it with the actual return grouting volume after the same correction. And when an anomaly is detected, a warning signal is issued through the warning prompt component.

9. The integrated intelligent monitoring device for well leakage and overflow under all drilling conditions according to claim 8, characterized in that, The main control data analysis and processing unit also has a built-in time-series neural network leakage trend prediction module. This module takes the historical inlet and outlet flow difference sequence, wellbore fluid level change sequence and operating condition sequence after being corrected by the mud thermal expansion and compressibility real-time correction module as input, and the trained time-series neural network model predicts the leakage risk probability in the future period. When the predicted probability exceeds the preset threshold, an early warning is issued.

10. The integrated intelligent monitoring device for well leakage and overflow under all drilling conditions according to claim 8, characterized in that, The main control data analysis and processing unit also has a built-in leakage type identification module. Based on the shape of the leakage rate change curve, the characteristics of the wellbore fluid level drop rate, and the time-varying law of the inlet and outlet flow difference when well leakage occurs, the module classifies well leakage into seepage type leakage, fracture type leakage, and karst type leakage, and outputs corresponding treatment suggestions.

Citation Information

Patent Citations

  • Monitor for well gushing and leaking

    CN2055532U