System and method for detecting downhole events
A method is provided to determine and mitigate one or more downhole events of a drill string in a wellbore. The method includes receiving, via a processor, drilling data from one or more surface sensors. The drilling data is processed into a plurality of segments. Each of the plurality of segments are processed by a pre-trained model such that one or more labels are determined. The one or more labels are relating to one or more downhole events corresponding to each of the plurality of segments.
1. A method comprising:
receiving drilling data captured at a surface of a wellbore using one or more surface sensors positioned at the surface of the wellbore;
separating the drilling data acquired by the one or more surface sensors into a plurality of segments using at least one processor, the drilling data separated such that a time axis corresponds to a sampling rate and each of the plurality of segments is of a desired number of seconds;
processing each of the plurality of segments using a pre-trained model by inputting the drilling data of each of the plurality of segments acquired by the one or more surface sensors into the pre-trained model, the pre-trained model trained to recognize one or more downhole events utilizing the drilling data acquired by the one or more surface sensors, the pre-trained model is a machine learning model using an artificial neural network having a plurality of layers;
determining, using the pre-trained model, one or more labels for each of the plurality of segments, the one or more labels relating to the one or more downhole events;
labeling each of the plurality of segments using the one or more labels corresponding to the one or more downhole events;
displaying, via a display of a computing device, a notification based on the labeling, the notification indicating the one or more downhole events can cause damage to a drill string; and
controlling the drill string based on the notification to mitigate the damage.
2. The method of claim 1 , wherein the one or more surface sensors are positioned on the drill string.
3. The method of claim 1 , wherein the one or more downhole events include at least one of stick-slip vibration, bit-bounce, whirl, kick, buckling, bit stalling, or motor failure.
4. The method of claim 1 , wherein the drilling data includes at least one of surface torque, weight on bit, lateral acceleration, rotations per minute, tension, flow rate, temperature, or pressure.
5. The method of claim 1 , wherein the machine learning model is a convolutional neural network, a recurrent neural network, a fully connected deep neural network, or a custom combination thereof.
6. The method of claim 1 , wherein the notification is displayed in real-time.
7. The method of claim 1 , wherein the pre-trained model is trained using sample data captured during a drilling operation at one or more sample drilling wells, the sample data including surface data and downhole data.
8. The method of claim 7 , wherein an input to the pre-trained model includes a plurality of sample segments and an output includes one or more sample labels corresponding to the plurality of sample segments, the plurality of sample segments obtained by processing the sample data.
9. The method of claim 8 , wherein the sample data is processed such that the surface data corresponds to the downhole data.
10. The method of claim 9 , wherein the downhole data includes at least one of lateral acceleration, axial acceleration, torque, rotation speed, flow rate, bending moment, or weight on bit.
11. The method of claim 1 , further comprising:
categorizing the one or more downhole events as severe or normal.
12. A system comprising:
one or more surface sensors positioned at a surface of a wellbore, the one or more surface sensors configured to measure drilling data at the surface of the wellbore; and
at least one computing device in communication with the one or more surface sensors, the at least one computing device including a processor, the at least one computing device configured to:
separate the drilling data acquired by the one or more surface sensors into a plurality of segments such that a time axis corresponds to a sampling rate and each of the plurality of segments is of a desired number of seconds,
determine one or more labels relating to one or more downhole events using a pre-trained model by inputting the drilling data of each of the plurality of segments acquired by the one or more surface sensors into the pre-trained model, the pre-trained model trained to recognize the one or more downhole events utilizing the drilling data acquired by the one or more surface sensors, the pre-trained model is a machine learning model using an artificial neural network having a plurality of layers,
label each of the plurality of segments using the one or more labels corresponding to the one or more downhole events,
display, via a display of the at least one computing device, the one or more labels to provide a notification that the one or more downhole events can cause damage a drill string,
display, via a display of the at least one computing device, a notification based on labeling of each of the plurality of segments using the one or more labels corresponding to the one or more downhole events, the notification indicating the one or more downhole events can cause damage to a drill string, and
control, via a controller, the drill string based on the notification to mitigate the damage.
13. The system of claim 12 , wherein the one or more surface sensors are positioned on the drill string.
14. The system of claim 12 , wherein the one or more downhole events include at least one of stick-slip vibration, bit-bounce, whirl, kick, buckling, bit stalling, or motor failure.
15. The system of claim 12 , wherein the drilling data includes at least one of surface torque, weight on bit, lateral acceleration, rotations per minute, tension, flow rate, temperature, or pressure.
16. The system of claim 12 , wherein the at least one computing device categorizes the one or more downhole events as severe or normal.
17. The system of claim 12 , wherein the at least one computing device is configured to display the notification in real-time.
18. The system of claim 12 ,
wherein,
the pre-trained model is trained using sample data captured during a drilling operation at one or more sample drilling wells,
the sample data is measured using one or more sample sensors,
the sample data includes surface data and downhole data,
the pre-trained model includes an input with a plurality of sample segments and an output with one or more sample labels corresponding to the plurality of sample segments.
19. The system of claim 18 , wherein the sample data is processed such that the surface data corresponds to the downhole data.
20. The system of claim 19 , wherein the downhole data includes at least one of lateral acceleration, axial acceleration, torque, rotation speed, flow rate, bending moment, or weight on bit.