# US 12,000,976 B2 · Systems and methods for training a well model to predict material loss for a pipe string within a borehole > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 12,000,976 B2 - Document ID: US12000976B2 - Status: Granted - Kind: B2 - Application number: 17/006,110 - Filed: 2020-08-28 - Published: 2021-04-22 - Granted: 2024-06-04 - Earliest claimed: 2019-10-21 - Assignees: Landmark Graphics Corporation - Applicants: Landmark Graphics Corporation - Inventors: Sebastian Kroczka; Welton Danniel Souza; Chafaa Badis - Web page: https://patentgrep.com/patent/US12000976B2 ## CPC - G01V 20/00: Geomodelling in general - E21B 47/085: using radiant means, e.g. acoustic, radioactive or electromagnetic - G01V 1/40: specially adapted for well-logging - G06F 30/27: using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model - G06N 20/00: Machine learning - E21B 2200/20: Computer models or simulations, e.g. for reservoirs under production, drill bits - G06F 2113/08: Fluids - G06F 2113/14: Pipes - G01V: GEOPHYSICS; GRAVITATIONAL MEASUREMENTS; DETECTING MASSES OR OBJECTS; TAGS (means for indicating the location of accidentally buried, e.g. snow-buried, persons A63B29/02) - E21B: EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract A method for training a well model to predict material loss for a pipe string having a wall thickness and located within a borehole. The method may include measuring the wall thickness of a first pipe string at locations axially along the first pipe string with a logging tool at a first time. The method may also include measuring the wall thickness of the first pipe string at the locations with the logging tool at a second time. The method may further include training a first well model based on a machine learning (“ML”) algorithm to predict a predicted amount of material loss in the future for the first pipe string at a selected location using the wall thickness measurements at the first and second times and well operating condition information related to the first pipe string. ## Claims ### Claim 1 A method for training a well model to predict material loss for a pipe string having a wall thickness and located within a borehole, the method comprising: measuring the wall thickness of a first pipe string at locations axially along the first pipe string with a logging tool at a first time; measuring the wall thickness of the first pipe string at the locations with the logging tool at a second time; training a first well model based on a machine learning (“ML”) algorithm to predict a predicted amount of material loss in the future for the first pipe string at a selected location using the wall thickness measurements at the first and second times and well operating condition information related to the first pipe string; determining an amount of actual material loss for the first pipe string at the locations based on the wall thickness measurements; receiving input from a user relating to user-defined color-coded categories based on the amount of actual material loss for the pipe string; categorizing the locations into the user-defined color-coded categories at the respective locations; and generating a graphical user interface showing a visual representation of the amount of actual material loss for the first pipe string using the user-defined color-coded categories on a display. ### Claim 2 (depends on 1) The method of claim 1, further comprising determining when to plug and abandon the borehole based on the first well model. ### Claim 3 (depends on 1) The method of claim 1, further comprising: associating each of the locations with a tubular section of the first pipe string; and categorizing each tubular section into one of the user-defined color-coded categories based on the user-defined color-coded category of the location with largest amount of material loss for each tubular section, wherein generating the graphical user interface showing the visual representation of the loss of material from the first pipe string comprises generating the graphical user interface showing the visual representation of the largest amount of material lost from each tubular section using the user-defined color-coded categories. ### Claim 4 (depends on 1) The method of claim 1, further comprising: measuring wall thicknesses of the first pipe string at the locations with the logging tool at a third time; predicting a predicted amount of material loss at the locations at the third time with the first well model; determining an amount of actual material loss for the first pipe string at the locations at the third time; and comparing the actual amount of actual material loss at the third time to the predicted amount of material loss at the third time to determine if the first well model is accurate. ### Claim 5 (depends on 4) The method of claim 4, further comprising retraining the first well model using the wall thickness measurements at the third time and additional well operating condition information generated between the second time and the third time. ### Claim 6 (depends on 1) The method of claim 1, further comprising: training a second well model based on thickness measurements from and well operating condition information related to a second pipe string located in a second borehole in a same field as the first borehole; and training a field model using the first well model and the second well model to predict how well operating conditions affect pipe strings located in boreholes in the field. ### Claim 7 (depends on 6) The method of claim 6, further comprising training the field model to predict at least one of rework or abandonment of any borehole in the field. ### Claim 8 A system for training a well model to predict material loss for a pipe string having a wall thickness and located within a borehole, the system comprising: a logging tool operable to measure a wall thickness of a first pipe string at locations axially along the first pipe string; and a processor in electronic communication with the logging tool, the processor programmed to: receive, from the logging tool, wall thickness measurements of a first pipe string at locations axially along the first pipe string at a first time; receive, from the logging tool, wall thickness measurements of the first pipe string at the locations at a second time; train a first well model based on an ML algorithm to predict a predicted amount of material loss in the future for the first pipe string at a selected location using the wall thickness measurements at the first and second times and well operating condition information related to the first pipe string; determine an amount of actual material loss for the first pipe string at the locations based on the wall thickness measurements; receive input from a user relating to user-defined color-coded categories based on the amount of actual material loss for the pipe string; categorize the locations into the user-defined color-coded categories at the respective locations; and generate a graphical user interface showing a visual representation of the amount of actual material loss for the first pipe string using the user-defined color-coded categories on a display. ### Claim 9 (depends on 8) The system of claim 8, wherein: the processor is further programmed to: associate each of the locations with a tubular section of the first pipe string; and categorize each tubular section into a color-coded category based on the color-coded category of the location with largest amount of material loss for each tubular section; and generating the graphical user interface showing the visual representation of the loss of material from the first pipe string comprises generating the graphical user interface showing the visual representation of the largest amount of material lost from each tubular section using the user-defined color-coded categories. ### Claim 10 (depends on 8) The system of claim 8, wherein the processor is further programmed to: receive, from the logging tool, wall thickness measurements of the first pipe string at locations along the first pipe string at a third time; predict a predicted amount of material loss at the locations at the third time with the first well model; determine an amount of actual material loss for the first pipe string at the locations at the third time; and compare the actual amount of actual material loss at the third time to the predicted amount of material loss at the third time to determine if the first well model is accurate. ### Claim 11 (depends on 10) The system of claim 10, wherein the processor is further programmed to retrain the first well model using the wall thickness measurements at the third time and additional well operating condition information generated between the second time and the third time. ### Claim 12 (depends on 8) The system of claim 8, wherein the processor is further programmed to: train a second well model based on thickness measurements from and well operating condition information related to a second pipe string located in a second borehole in a same field as the first borehole; and train a field model using the first well model and the second well model to predict how well operating conditions affect pipe strings located in boreholes in the field. ### Claim 13 (depends on 12) The system of claim 12, wherein the processor is further programmed to train the field model to predict at least one of rework or abandonment of any borehole in the field. ### Claim 14 A non-transitory computer-readable medium comprising instructions which, when executed by a processor, enables the processor to perform a method for training a well model to predict material loss for a pipe string having a wall thickness and located within a borehole, the method comprising: receiving wall thickness measurements of a first pipe string at locations axially along the first pipe string at a first time; receiving wall thickness measurements of the first pipe string at the locations at a second time; training a first well model based on an ML algorithm to predict a predicted amount of material loss in the future for the first pipe string at a selected location using the wall thickness measurements at the first and second times and well operating condition information related to the first pipe string; determining an amount of actual material loss for the first pipe string at the locations based on the wall thickness measurements; receiving input from a user relating to user-defined color-coded categories based on the amount of actual material loss for the pipe string; categorizing the locations into the user-defined color-coded categories at the respective locations; and generating a graphical user interface showing a visual representation of the amount of actual material loss for the first pipe string using the user-defined color-coded categories on a display. ### Claim 15 (depends on 14) The non-transitory computer-readable medium of claim 14, wherein the method further comprises determining when to plug and abandon the borehole based on the first well model. ### Claim 16 (depends on 5) The non-transitory computer-readable medium of claim 5, wherein the method further comprises: associating each of the locations with a tubular section of the first pipe string; categorizing each tubular section into one of the user-defined color-coded categories based on the color-coded category of the location with largest amount of material loss for each tubular section; and generating the graphical user interface showing the visual representation of the loss of material from the first pipe string comprises generating the graphical user interface showing the visual representation of the largest amount of material lost from each tubular section using the user-defined color-coded categories. ### Claim 17 (depends on 14) The non-transitory computer-readable medium of claim 14, wherein the method further comprises: receiving wall thickness measurements of the first pipe string at the locations at a third time; predicting a predicted amount of material loss at the locations at the third time with the first well model; determining an amount of actual material loss for the first pipe string at the locations at the third time; and comparing the actual amount of actual material loss at the third time to the predicted amount of material loss at the third time to determine if the first well model is accurate. ### Claim 18 (depends on 17) The non-transitory computer-readable medium of claim 17, wherein the method further comprises retraining the first well model using the wall thickness measurements at the third time and additional well operating condition information generated between the second time and the third time. ### Claim 19 (depends on 14) The non-transitory computer-readable medium of claim 14, wherein the method further comprises: training a second well model based on thickness measurements from and well operating condition information related to a second pipe string located in a second borehole in a same field as the first borehole; and training a field model using the first well model and the second well model to predict how well operating conditions affect pipe strings located in boreholes in the field. ### Claim 20 (depends on 19) The non-transitory computer-readable medium of claim 19, wherein the method further comprises training the field model to predict at least one of rework or abandonment of any borehole in the field.