# US 11,441,411 B2 · Optimal drilling parameter machine learning system and methods > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 11,441,411 B2 - Document ID: US11441411B2 - Status: Granted - Kind: B2 - Application number: 16/601,960 - Filed: 2019-10-15 - Published: 2021-04-15 - Granted: 2022-09-13 - Earliest claimed: 2019-10-15 - Assignees: Nabors Drilling Technologies USA, Inc. - Applicants: Nabors Drilling Technologies USA, Inc. - Inventors: Suresh Venugopal; Namitha Vinay - Web page: https://patentgrep.com/patent/US11441411B2 ## CPC - E21B 44/00: Automatic control systems specially adapted for drilling operations, i.e. self-operating systems which function to carry out or modify a drilling operation without intervention of a human operator, e.g. computer-controlled drilling systems; Systems specially adapted for monitoring a plurality of drilling variables or conditions - E21B 41/00: Equipment or details not covered by groups E21B15/00 - E21B40/00 - G06N 3/0499: Feedforward networks - G06N 3/09: Supervised learning - G06N 7/01: Probabilistic graphical models, e.g. probabilistic networks - G06N 20/00: Machine learning - E21B 2200/20: Computer models or simulations, e.g. for reservoirs under production, drill bits - E21B 2200/22: Fuzzy logic, artificial intelligence, neural networks or the like - G06N 3/08: Learning methods - E21B: EARTH OR ROCK DRILLING; OBTAINING OIL, GAS, WATER, SOLUBLE OR MELTABLE MATERIALS OR A SLURRY OF MINERALS FROM WELLS - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract A method of optimizing drilling and drilling instructions for different segments of a wellbore that includes: calculating mathematical approximations of energy transfer loss between a top drive and a BHA using a neural network and a plurality of drilling records for segments of wellbores that are similar to the segments of the wellbore to be drilled; calculating drilling instructions based on each mathematical approximation; drilling the wellbore using the drilling instructions; monitoring the energy transfer loss; and optimizing the drilling instructions using one of the mathematical approximations to minimize the energy transfer loss. Optimizing the drilling instructions includes using Bayesian optimization techniques. ## Claims ### Claim 1 A method of drilling a first segment of a wellbore that has a first set of distinctive characteristics, the method comprising: training, using a computing system, a first machine learning model using a first plurality of historical wellbore drilling records; wherein each of the historical wellbore drilling records of the first plurality of historical wellbore drilling records comprises historical data obtained during drilling of a wellbore and upon which a delta mechanical specific energy (“MSE”) is calculated for a given depth of the wellbore; wherein the delta MSE is a difference between a surface MSE and a bottom hole assembly (“BHA”) MSE for the given depth; wherein each of the wellbores associated with the first plurality of historical wellbore drilling records has the first set of distinctive characteristics; and wherein an output of the first machine learning model is a first mathematical expression representing a relationship between a first set of optimization parameters and the delta MSE when drilling a wellbore that has the first set of distinctive characteristics; generating, using the computing system and based on the first mathematical expression, first target values for the first set of optimization parameters for drilling of the wellbore; drilling a first portion of the first segment of the wellbore using the first target values for the first set of optimization parameters; monitoring, using the computing system, drilling parameters during the drilling of the first segment of the wellbore; calculating, based on the monitored drilling parameters and using the computing system, the delta MSE associated with the first portion of the first segment of the wellbore; optimizing, using the computing system and based on the first mathematical expression, the first target values for the first set of optimization parameters to reduce the delta MSE; and drilling a second portion of the first segment of the wellbore using the optimized first target values for the first set of optimization parameters. ### Claim 2 (depends on 1) The method of claim 1, wherein the first set of optimization parameters comprise one or more of: weight on bit, rotations per minute, standpipe pressure, and torque. ### Claim 3 (depends on 1) The method of claim 1, wherein the first set of distinctive characteristics comprises one or more of: a client name, a geographical area, an expected formation, a shape of a portion of the wellbore, equipment information, and a target wellbore diameter. ### Claim 4 (depends on 3) The method of claim 3, wherein the geographical area comprises a field name; wherein the equipment information comprises information identifying equipment that forms the BHA; and wherein the shape of the portion of the wellbore comprises one or more of: a straight shape, an L-shape, and a S-shape. ### Claim 5 (depends on 1) The method of claim 1, which further comprises training, using the computing system, a second machine learning model using a second plurality of historical wellbore drilling records; wherein each of the historical wellbore drilling records of the second plurality of historical wellbore drilling records comprises historical data obtained during drilling of a wellbore and upon which a delta MSE is calculated for a given depth of the wellbore; wherein each of the wellbores associated with the second plurality of historical wellbore drilling records has a second set of distinctive characteristics; and wherein an output of the second machine learning model is a second mathematical expression representing a relationship between a second set of optimization parameters and the delta MSE when drilling a wellbore that has the second set of distinctive characteristics. ### Claim 6 (depends on 5) The method of claim 5, wherein the first mathematical expression is different from the second mathematical expression. ### Claim 7 (depends on 5) The method of claim 5, wherein the first set of optimization parameters are different from the second set of optimization parameters. ### Claim 8 (depends on 5) The method of claim 5, which further comprises: generating, using the computing system and based on the second mathematical expression, second target values for the second set of optimization parameters for drilling a second segment of the wellbore; drilling a first portion of the second segment of the wellbore using the second target values for the second set of optimization parameters; wherein the second segment of the wellbore has the second set of distinctive characteristics; monitoring, using the computing system, drilling parameters during the drilling of the first portion of the second segment of the wellbore; calculating, based on the monitored drilling parameters and using the computing system, the delta MSE associated with the first portion of the second segment of the wellbore; optimizing, using the computing system and based on the second mathematical expression, the second target values for the second set of optimization parameters to reduce the delta MSE; and drilling a second portion of the second segment of the wellbore using the optimized second target values for the second set of optimization parameters. ### Claim 9 (depends on 1) The method of claim 1, wherein optimizing the first target values for the first set of optimization parameters comprises using Bayesian optimization techniques. ### Claim 10 (depends on 1) The method of claim 1, wherein optimizing the first target values for the first set of optimization parameters occurs after a first stand of drill pipe is coupled to the BHA that creates the wellbore and before a second consecutive stand of drill pipe is coupled to the BHA. ### Claim 11 An apparatus for drilling a first segment of a wellbore that has a first set of distinctive characteristics, comprising: a non-transitory computer readable medium having stored thereon a plurality of instructions, wherein the instructions are executed with at least one processor so that the following steps are executed: training a first machine learning model using a first plurality of historical wellbore drilling records; wherein each of the historical wellbore drilling records of the first plurality of historical wellbore drilling records comprises historical data obtained during drilling of a wellbore and upon which a delta mechanical specific energy (“MSE”) is calculated for a given depth of the wellbore; wherein the delta MSE is a difference between a surface MSE and a bottom hole assembly (“BHA”) MSE for the given depth; wherein each of the wellbores associated with the first plurality of historical wellbore drilling records has the first set of distinctive characteristics; and wherein an output of the first machine learning model is a first mathematical expression representing a relationship between a first set of optimization parameters and the delta MSE when drilling a wellbore that has the first set of distinctive characteristics; generating, based on the first mathematical expression, first target values for the first set of optimization parameters for drilling of the wellbore; drilling a first portion of the first segment of the wellbore using the first target values for the first set of optimization parameters; monitoring drilling parameters during the drilling of the first segment of the wellbore; calculating, based on the monitored drilling parameters, the delta MSE associated with the first portion of the first segment of the wellbore; optimizing the first target values for the first set of optimization parameters, using the first mathematical expression and in response to the calculated delta MSE, to reduce the delta MSE; and drilling a second portion of the first segment of the wellbore using the optimized first target values for the first set of optimization parameters. ### Claim 12 (depends on 11) The apparatus of claim 11, wherein the first target parameters comprise one or more of: weight on bit, rotations per minute, standpipe pressure, and torque. ### Claim 13 (depends on 11) The apparatus of claim 11, wherein the first set of distinctive characteristics comprises one or more of: a client name, a geographical area, an expected formation, a shape of a portion of the wellbore, equipment information, and a target wellbore diameter. ### Claim 14 (depends on 13) The apparatus of claim 13, wherein the geographical area comprises a field name; wherein the equipment information comprises information identifying equipment that forms a BHA used to create the wellbore; and wherein the shape of the portion of the wellbore comprises one or more of: a straight shape, an L-shape, and a S-shape. ### Claim 15 (depends on 11) The apparatus of claim 11, wherein execution of the instructions results in the following steps also being executed: training a second machine learning model using a second plurality of historical wellbore drilling records; wherein each of the historical wellbore drilling records of the second plurality of historical wellbore drilling records comprises historical data obtained during drilling of a wellbore and upon which a delta MSE is calculated for a given depth of the wellbore; wherein each of the wellbores associated with the second plurality of historical wellbore drilling records has a second set of distinctive characteristics; and wherein an output of the second machine learning model is a second mathematical expression representing a relationship between a second set of optimization parameters and the delta MSE when drilling a wellbore that has the second set of distinctive characteristics. ### Claim 16 (depends on 15) The apparatus of claim 15, wherein the first mathematical expression is different from the second mathematical expression. ### Claim 17 (depends on 15) The apparatus of claim 15, wherein the first set of optimization parameters are different from the second set of optimization parameters. ### Claim 18 (depends on 15) The apparatus of claim 15, wherein execution of the instructions results in the following steps also being executed: generating second target parameters for drilling of a second segment of the wellbore based on the second mathematical expression; drilling a first portion of the second segment of the wellbore using the second target parameters; wherein the second segment of the wellbore has the second set of distinctive characteristics; monitoring drilling parameters during the drilling of the first portion of the second segment of the wellbore; calculating, based on the monitored drilling parameters, the delta MSE associated with the first portion of the second segment of the wellbore; optimizing the second target values for the second set of optimization parameters, using the second mathematical expression and in response to the calculated delta MSE, to minimize the delta MSE; and drilling a second portion of the second segment of the wellbore using the optimized second target values for the second set of optimization parameters. ### Claim 19 A method of optimizing drilling instructions for segments of a wellbore, the method comprising: calculating a first and second mathematical approximations of energy transfer loss between a top drive and a bottom hole assembly coupled to the top drive using a neural network and a plurality of drilling records for segments of wellbores that are similar to the segment of the wellbore to be drilled; calculating a first set of drilling instructions based on the first mathematical approximation; calculating a second set of drilling instructions based on the second mathematical approximation; wherein the second mathematical approximation is different from the first mathematical approximation; drilling a first portion of a first segment of the wellbore using the first set of drilling instructions; monitoring the energy transfer loss between a top drive associated with the wellbore and a bottom hole assembly associated with the wellbore while drilling the first portion of the first segment of the wellbore; optimizing the first set of drilling instructions using the first mathematical approximation to minimize the energy transfer loss between the top drive and the bottom hole assembly; drilling a second portion of the first segment of the wellbore using the optimized first set of drilling instructions; drilling a first portion of a second segment of the wellbore using the second set of drilling instructions; monitoring the energy transfer loss between the top drive and the bottom hole assembly while drilling the first portion of the second segment of the wellbore; optimizing the second set of drilling instructions using the second mathematical approximation to minimize the energy transfer loss between the top drive and the bottom hole assembly; and drilling a second portion of the second segment of the wellbore using the optimized second set of drilling instructions. ### Claim 20 (depends on 19) The method of claim 19, wherein optimizing the first set of drilling instructions comprises using Bayesian optimization techniques.