DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Response to Amendment
Claims 1, 3-7, 9-11, 13-17, and 19-20 are amended, claims 2 and 12 are cancelled, and claims 21-22 are new. Claims 1, 3-11, and 13-22 are pending.
Response to Arguments
Applicant's arguments filed 5/11/2026 have been fully considered.
Regarding the rejections of independent claims 1 and 11 under 101, Examiner agrees with Applicant’s argument on page 9 of the response that the claims as a whole integrate any alleged abstract idea into a practical application because the claims as a whole recite a technological solution to a technological problem by scaling a flow rate based on predicted measurement when real-time downhole equipment measurement data is unavailable (e.g., cannot be received). Such a practical application is evident based on “downhole” equipment being implementing for material extraction and the efficiency and head capacity related performance curve that is generated based on measurement data identifies a material extraction rate. Furthermore, the practical application is evident in terms of updating the performance curve by activity including “predicting, in real time, via a machine learning model and in response to the determining that the second equipment is unavailable, second measurement data associated with the second equipment during a time that the second equipment is unavailable, and updating the performance curve based on the predicted second measurement data” and “automatically scaling, in real time, a flow rate of the material extracted from the downhole environment based on the performance curve.”
Therefore, the rejections of claims 1, 3-11, and 13-20 under 101 are withdrawn.
Regarding the rejections of independent claims 1 and 11 under 102 as anticipated by Moricca, Examiner acknowledges that as contended by Applicant on pages 11-12 of the response that the amendments overcome the rejections, which are withdrawn. However, in view of further search and analysis new grounds for rejecting claims 1 and 11 under 103 are set forth herein.
Regarding amended claim 1 and with reference to the rejection of claim 2 under 103, Applicant contends on pages 13-14 that Jaaskelainen in combination with Moricca does not teach the combined elements of amended claim 1. In support, Applicant notes on page 14 that amended claim 1 recites that the predicted second measurement data is used to “update the performance curve,” that the performance curve is configured for “identifying an extraction rate of the material with respect to efficiency and head capacity,” and that per claim 1 the system automatically scales “in real time, a flow rate of the material extracted from the downhole environment based on the performance curve to extend a duration of the run.” Applicant then asserts that Jaaskelainen does not describe routing such synthetic data into an extraction-rate-vs-efficiency-and-head-capacity performance curve to update the curve and further does not describe automatically scaling flow rate of extracted material based on such a curve, such that Morrica in view of Jaaskelainen does not describe such a performance curve updated using the predicted measurement data.
Examiner acknowledges as asserted by Applicant on page 14 that Jaaskelainen does not describe routing the ML-generated substitute data into an extraction-rate-vs-efficiency-and-head-capacity performance curve to update the curve nor does Jaaskelainen appear to teach automatically scaling a flow rate of extracted material based on such a curve. Examiner submits that Moricca teaches updating a performance curve (e.g., [0023] real-time measurement data used for updating model) and further teaches adjusting (scaling) pump flow rate in real time based on the performance curve (FIG. 2E performance curves 244 including a Best Efficiency Curve including a liquid rate (bb/day) corresponding to the head value and table 246 providing user-selectable/adjustable configuration including flow rates. Performance curve is generated/updated during/reactive to ongoing dynamic online operations (e.g., as described in Abstract, [0018]) and therefore performance curve and operations performed with respect to performance curve are performed “in real time”; FIG. 2E showing tracking of selected flow rates corresponding/following performance curves; [0023]). Moricca further teaches generating/updating the performance curve using multiple sensor instruments ([0014] measurement data may be collected using multiple different downhole instruments; [0023] performance curve(s) generated/updated based on nodal model using real-time (dynamically changing) data, which is based on the measured data that per [0014] may include data from multiple downhole instruments).
Jaaskelainen discloses a method for monitoring and managing well operations including pump operations (Abstract) for circumstances in which it is determined that sensing data is unavailable during the monitoring operations ([0002] well may not be tooled with particular sensors that would be useful for AI processing; [0016] synthetic model used for circumstances in which certain sensors may be unavailable. Examiner notes the unavailability of the sensor(s) is inherently entailed in the determination to configure a synthetic model to obtain corresponding “predicted” data). Jaaskelainen further teaches predicting additional measurement data associated with other (second) equipment during a time that the other equipment is unavailable by combining actual sensor measurement values and “synthetic” values that are values predicted (in terms of what an actual sensor value is expected to be) using a machine learning model ([0016] and [0033] AI synthetic data model generates unavailable sensor data with respect to pump operations; [0015] and [0030] AI model is trained (machine learning); claim 1).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of using a machine learning model to predict, in association with a determination that sensor equipment is/will be unavailable, pump operation related measurements to the method taught by Moricca in which multiple sensor measurements are used for generating the performance curve such that in combination the method includes determining that a second equipment submersed into the downhole environment becomes unavailable during the run and predicting, in real time (as part of the reactive modeling updating taught by Moricca) second measurement data associated with the second equipment (second equipment is or includes a sensor) during a time of equipment unavailability during the run by using a machine learning model and applying the second measurement data to the performance curve updating disclosed by Moricca.
A motivation would have been to enable use of potentially useful sensor data that may not be otherwise available by actual sensor measurement as disclosed by Jaaskelainen ([0016]). Furthermore, such a combination would amount to applying a known design option for providing additional sensor data for pump monitoring to achieve predictable results.
On page 14 of the response Applicant further contends that “[e]ven substituting Jaaskelainen’s synthetic data for Moricca’s measured-data inputs, the combination yields at most a nodal-physics performance curve incorporating synthetic sensor values, not a performance curve updated ‘based on the predicted second measurement data’ generated by the machine learning model…” and further contends that neither Moricca nor Jaaskelainen teach or suggest scaling the flow rate “in real time” based on the performance curve.
Examiner submits that Jaaskelainen’s contribution to the combination lay in using machine learning to determine otherwise “missing” sensor measurement data, which as combined with Moricca which teaches updating the performance curve using one or more of multiple possible sensor inputs collected in real time, would result in a modification to Moricca that results in the performance curve being updated “based on the predicted second measurement data” and consequently that the flow rate, that as disclosed by Moricca is controlled based on the performance curve, is consequently also adjusted/scaled in real time according to updating of the performance curve.
Claim Objections
Claims 7 and 10 are objected to because of the following informalities:
In claim 7 line 3, “Predicting” should not be capitalized.
In claim 10 line 2, “deterioration at least one” should read “deterioration of at least one.”
Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
Claims 7 and 17 are rejected under 35 U.S.C. 112(a) as failing to comply with the written description requirement. The claim(s) contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, at the time the application was filed, had possession of the claimed invention.
Claim 7 recites “determining, via the machine learning model, a confidence indicator of the actual value of the efficiency of the first equipment based at least on the minimum value and the maximum value,” which does not appear to be disclosed by Applicant’s original disclosure with sufficient clarity such that one of ordinary skill could reasonably conclude that the inventor was in possession of the claimed invention at the time of filing.
As disclosed by Applicant’s specification and drawings (e.g., FIGS. 4A and 4B, [0020], [0036], [0046], and [0048]) a confidence indicator is determined for a given efficiency curve segment that also includes minimum, maximum, and actual efficiency values such that the confidence indicator is disclosed as coinciding with minimum and maximum efficiencies. However, Applicant’s original disclosure appears to be largely silent regarding the manner in which the confidence indicator is actually determined including using the minimum and maximum efficiency values as a basis for such determination. Therefore, while the possibility of determining a confidence of actual efficiency using minimum and maximum efficiency values may or may not have been apparent to one of ordinary skill, Applicant’s original disclosure does not convey this feature with sufficient clarity such that one of ordinary skill would understand that the invention actually described includes such a feature (i.e., that the inventor(s) had possession of the invention including this feature at the time of filing).
Claim 17 includes substantially similar features and is likewise rejected for the same reasons.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 3-4, 6, 8, 10-11, 13-14, 16, 18, and 20 are rejected under 35 U.S.C. 103 as being unpatentable over Moricca (US 2014/0039836 A1) in view of Jaaskelainen (US 2021/0123431 A1).
As to claim 1, Moricca teaches “[a] method of providing performance feedback of an extraction facility (Abstract disclosing method for monitoring to optimize operations of ESP; FIG. 4A and 4B; method performed by system 300 in FIG. 3) comprising:
receiving first measurement data from a first equipment submersed into a downhole environment during a run for extracting material from the downhole environment (Abstract disclosing collecting measured data representative of state of ESP within a well; [0012] and [0014] monitoring ESP operations including collecting downhole measurement data; [0023] data acquisition system 310 acquired the measurement data);
determining, in real time, a performance curve based on the first measurement data (Abstract performance curve for ESP generated using updated model that is based on the measured data; [0023] performance curve(s) generated based on nodal model, which is based on the measured data which may be real-time data. Performance curve is generated/updated during/reactive to ongoing dynamic online operations (e.g., as described in Abstract, [0018], and [0023]) and therefore is performed “in real time”), the performance curve identifying an extraction rate of the material with respect to efficiency and head capacity (FIG. 2E performance curve (as interpreted in view of Applicant’s specification to entail a collection of curves) relates liquid rate (extraction rate of ESP) to head in feet (capacity) and efficiency (ops range including best efficiency));”
“providing the performance curve to an operator of the run (FIG. 3 depicting graphical display that may be viewed by a user; [0023] data generated by matched model) for real-time feedback associated with performance of equipment in the downhole environment ([0023] ESP data may include real-time monitoring data such that the feedback provided by the displayed curved effectively constitutes real-time feedback); and
automatically scaling, in real time, a flow rate of the material extracted from the downhole environment (FIG. 2E performance curves 244 including a Best Efficiency Curve including a liquid rate (bb/day) corresponding to the head value and table 246 providing user-selectable/adjustable configuration including flow rates. Performance curve is generated/updated during/reactive to ongoing dynamic online operations (e.g., as described in Abstract, [0018]) and therefore performance curve and operations performed with respect to performance curve are performed “in real time”) based on the performance curve (FIG. 2E showing tracking of selected flow rates corresponding/following performance curves; [0023]) to extend a duration of the run (Best Efficiency Curve in FIG. 2E conveys an optimal operating efficiency that would inherently minimize equipment deterioration (extend run)).”
Regarding “determining, in real time, that a second equipment submersed into the downhole environment is unavailable during the run,” and “predicting, in real time, via a machine learning model and in response to the determining that the second equipment is unavailable, second measurement data associated with the second equipment during a time that the second equipment is unavailable, and updating the performance curve based on the predicted second measurement data,” Moricca discloses providing measurement data from multiple sensor instruments (first and second equipment) that are also used for generating/updating the performance curve ([0014] measurement data may be collected using multiple different downhole instruments; [0023] performance curve(s) generated/updated based on nodal model using real-time (dynamically changing) data, which is based on the measured data that per [0014] may include data from multiple downhole instruments).
Moricca however does not appear to disclose determining that a downhole equipment (e.g., sensor) is unavailable and “predicting” the additional sensor/measurement data “via a machine learning model” and in response to determining equipment unavailability.
Jaaskelainen discloses a method for monitoring and managing well operations including pump operations (Abstract) for circumstances in which it is determined that sensing data is unavailable during the monitoring operations ([0002] well may not be tooled with particular sensors that would be useful for AI processing; [0016] synthetic model used for circumstances in which certain sensors may be unavailable. Examiner notes the unavailability of the sensor(s) is inherently entailed in the determination to configure a synthetic model to obtain corresponding “predicted” data). Jaaskelainen further teaches predicting additional measurement data associated with other (second) equipment during a time that the other equipment is unavailable by combining actual sensor measurement values and “synthetic” values that are values predicted (in terms of what an actual sensor value is expected to be) using a machine learning model ([0016] and [0033] AI synthetic data model generates unavailable sensor data with respect to pump operations; [0015] and [0030] AI model is trained (machine learning); claim 1).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of using a machine learning model to predict, in association with a determination that sensor equipment is/will be unavailable, pump operation related measurements to the method taught by Moricca in which multiple sensor measurements are used for generating the performance curve such that in combination the method includes determining that a second equipment submersed into the downhole environment becomes unavailable during the run and predicting, in real time (as part of the reactive modeling updating taught by Moricca) second measurement data associated with the second equipment (second equipment is or includes a sensor) during a time of equipment unavailability during the run by using a machine learning model and applying the second measurement data to the performance curve updating disclosed by Moricca.
A motivation would have been to enable use of potentially useful sensor data that may not be otherwise available by actual sensor measurement as disclosed by Jaaskelainen ([0016]). Furthermore, such a combination would amount to applying a known design option for providing additional sensor data for pump monitoring to achieve predictable results.
As to claim 3, the combination of Moricca and Jaaskelainen teaches “[t]he method of claim 1,” and Jaaskelainen further teaches “wherein the machine learning model is at least partially trained based on measurement data from the downhole environment ([0015] synthesis model trained using historical downhole sensing data or from offset (other) wells; [0028]-[0030] downhole sensor data used in training process).”
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of training the machine learning modeling using measurement data from the downhole environment to the method taught by Moricca as modified by Jaaskelainen that uses machine learning modeling to provide additional sensor data for generating the performance curve, such that in combination the method includes use of a machine learning model that has been trained based on measurement data from the downhole environment.
The motivation would have been to select/utilize a model that is trained in a manner that accounts for downhole operations/conditions to optimize accuracy of modeling output as suggested by Jaaskelainen.
As to claim 4, the combination of Moricca and Jaaskelainen teaches “[t]he method of claim 1,” and Jaaskelainen further teaches “wherein the machine learning model is further trained based on downhole environments having different characteristics ([0015] synthesis model trained using historical downhole sensing data or from offset (other) wells).”
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of training the machine learning modeling using measurement data different downhole environments (downhole environments having different characteristics) to the method taught by Moricca as modified by Jaaskelainen that uses machine learning modeling to provide additional sensor data for generating the performance curve, such that in combination the method includes use of a machine learning model that has been trained based on downhole environments having different characteristics.
The motivation would have been to select/utilize a model that is trained in a manner that accounts for downhole operations/conditions to optimize accuracy over a broader range of operational/environmental conditions, resulting a more robust modeling for complex and dynamic downhole operations and conditions as suggested by Jaaskelainen.
As to claim 6, the combination of Moricca and Jaaskelainen teaches “[t]he method of claim 1,” and Moricca teaches “wherein the performance curve is updated using” “third measurement data ([0023] performance curve(s) generated/updated based on nodal model, which is based on the measured data that per [0014] may include data from multiple downhole instruments)” but does not appear to teach
“determining that a third equipment becomes unavailable during the run; and
predicting third measurement data associated with the third equipment during a time that the third equipment is unavailable,
wherein the performance curve is updated using “the predicted” third measurement data.
Jaaskelainen discloses a method for monitoring and managing well operations including pump operations (Abstract) for circumstances in which it is determined that sensing data is unavailable during the monitoring operations ([0002] well may not be tooled with particular sensors that would be useful for AI processing; [0016] synthetic model used for circumstances in which certain sensors may be unavailable (Examiner note the unavailability of the sensor(s) is inherently entailed in the determination to configure a synthetic model to obtain corresponding “predicted” data). Jaaskelainen further teaches predicting additional measurement data associated with other (third) equipment during a time that the other equipment is unavailable by combining actual sensor measurement values and “synthetic” values that are values predicted (in terms of what an actual sensor value is expected to be) using a machine learning model ([0016] and [0033] AI synthetic data model generates unavailable sensor data with respect to pump operations; [0015] and [0030] AI model is trained (machine learning); claim 1).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of using a machine learning model to predict, in association with a determination that sensor equipment is/will be unavailable, pump operation related measurements to the method taught by Moricca in which multiple sensor measurements are used for generating the performance curve such that in combination the method includes determining that a third equipment becomes unavailable during the run and predicting an additional (third) measurement from additional (third) equipment during the run by using a machine learning model (predicting), such that the performance curve is updated using the “predicted” third measurement data.
A motivation would have been to enable use of potentially useful sensor data that may not be otherwise available by actual sensor measurement as disclosed by Jaaskelainen ([0016]). Furthermore, such a combination would amount to applying a known design option for providing additional sensor data for pump monitoring to achieve predictable results.
As to claim 8, the combination of Moricca and Jaaskelainen teaches “[t]he method of claim 1, wherein, before beginning the run, an initial performance curve is generated (Moricca: FIG. 4 blocks 410 (model is updated), 412 (corresponding performance curves generated that per block 414 constitute updated performance curves), and the performance curve changes based on measurement data during the run (Moricca: FIG. 4 blocks 402, 404, 410 and 412; [0022] performance curves generated to correspond to analysis models; [0023] analysis models updated by on measured conditions (ESP data collected via data acquisition subsystem 310)).”
As to claim 10, the combination of Moricca and Jaaskelainen teaches “[t]he method of claim 1, further comprising:
estimating a recommended flow rate for the material (Moricca: FIG. 2E performance curves 244 including a Best Efficiency Curve including a liquid rate (bb/day) corresponding to the head value and table 246 providing user-selectable configuration including flow rates) to minimize deterioration at least one of the first equipment or the second equipment (Moricca: Best Efficiency Curve in FIG. 2E conveys an optimal operating efficiency that would inherently minimize equipment deterioration) based on the downhole environment (Moricca: FIG. 2E performance curves including Best Efficiency Curve associate pump head (downhole operation parameter is part of downhole environment) with flow rate. Examiner notes that the correspondence/intersection of the flow rate with pump head constitutes interdependence of head/flow rate in terms of operation efficiency).”
The Examiner notes that “to minimize deterioration” conveys an intended result/purpose that does not further limit the function of the recited method and is therefore not given patentable weight.
As to claim 11, Moricca teaches “[a] system for providing performance feedback of an extraction facility (Abstract disclosing system for monitoring to optimize operations of ESP; FIG. 4A and 4B; FIG. 3 system 300), comprising:
a storage configured to store instructions (claim 13 system includes memory for storing ESP monitoring, diagnosing and optimizing software; FIG. 1 computer system 45 (inherently includes memory for storing executable instructions));
a processor configured to execute the instructions (claim 13 system includes a processor for executing instructions; FIG. 1 computer system 45 (inherently includes a processor for executing instructions)) and cause the processor to:
receive first measurement data from a first equipment submersed into a downhole environment during a run for extracting material from the downhole environment (Abstract disclosing collecting measured data representative of state of ESP within a well; [0012] and [0014] monitoring ESP operations including collecting downhole measurement data; [0023] data acquisition system 310 acquired the measurement data);
determine a performance curve based on the first measurement data (Abstract performance curve for ESP generated using updated model that is based on the measured data; [0023] performance curve(s) generated based on nodal model, which is based on the measured data), the performance curve identifying an extraction rate of the material with respect to efficiency and head capacity (FIG. 2E performance curve (as interpreted in view of Applicant’s specification to entail a collection of curves) relates liquid rate (extraction rate of ESP) to head in feet (capacity) and efficiency (ops range including best efficiency));”
“provide the performance curve to an operator of the run (FIG. 3 depicting graphical display that may be viewed by a user; [0023] data generated by matched model) for real-time feedback associated with performance of equipment in the downhole environment ([0023] ESP data may include real-time monitoring data such that the feedback provided by the displayed curved effectively constitutes real-time feedback); and
automatically scale a flow rate of the material extracted from the downhole environment (FIG. 2E performance curves 244 including a Best Efficiency Curve including a liquid rate (bb/day) corresponding to the head value and table 246 providing user-selectable/adjustable configuration including flow rates) based on the performance curve (FIG. 2E showing tracking of selected flow rates corresponding/following performance curves; [0023]) to extend a duration of the run (Best Efficiency Curve in FIG. 2E conveys an optimal operating efficiency that would inherently minimize equipment deterioration (extend run)).
Regarding “determine that a second equipment submersed into the downhole environment is unavailable during the run” and “predict, via a machine-learning model and in response to the determining that the second equipment becomes unavailable, second measurement data associated with the second equipment during a time that the second equipment is unavailable, and update the performance curve based on the predicted second measurement data,” Moricca discloses providing measurement data from multiple sensor instruments (first and second equipment) that are also used for generating/updating the performance curve ([0014] measurement data may be collected using multiple different downhole instruments; [0023] performance curve(s) generated/updated based on nodal model, which is based on the measured data that per [0014] may include data from multiple downhole instruments).
Moricca however does not appear to disclose determining that a downhole equipment (e.g., sensor) is unavailable and “predicting” the additional sensor/measurement data “via a machine learning model” and in response to determining equipment unavailability.
Jaaskelainen discloses a method/system for monitoring and managing well operations including pump operations (Abstract) for circumstances in which it is determined that sensing data is unavailable during the monitoring operations ([0002] well may not be tooled with particular sensors that would be useful for AI processing; [0016] synthetic model used for circumstances in which certain sensors may be unavailable (Examiner note the unavailability of the sensor(s) is inherently entailed in the determination to configure a synthetic model to obtain corresponding “predicted” data). Jaaskelainen further teaches predicting additional measurement data associated with other (second) equipment during a time that the other equipment is unavailable by combining actual sensor measurement values and “synthetic” values that are values predicted (in terms of what an actual sensor value is expected to be) using a machine learning model ([0016] and [0033] AI synthetic data model generates unavailable sensor data with respect to pump operations; [0015] and [0030] AI model is trained (machine learning); claim 1).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of using a machine learning model to predict, in association with a determination that sensor equipment is/will be unavailable, pump operation related measurements to the system taught by Moricca in which multiple sensor measurements are used for generating the performance curve such that in combination the system is configured for determining that a second equipment submersed into the downhole environment becomes unavailable during the run and predicting second measurement data associated with the second equipment (second equipment is or includes a sensor) during a time of equipment unavailability during the run by using a machine learning model and applying the second measurement data to the performance curve updating disclosed by Moricca.
A motivation would have been to enable use of potentially useful sensor data that may not be otherwise available by actual sensor measurement as disclosed by Jaaskelainen ([0016]). Furthermore, such a combination would amount to applying a known design option for providing additional sensor data for pump monitoring to achieve predictable results.
As to claim 13, the combination of Moricca and Jaaskelainen teaches “[t]he system of claim 11,” and Jaaskelainen further teaches “wherein the machine learning model is at least partially trained based on measurement data from the downhole environment ([0015] synthesis model trained using historical downhole sensing data or from offset (other) wells; [0028]-[0030] downhole sensor data used in training process).”
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of training the machine learning modeling using measurement data from the downhole environment to the system taught by Moricca as modified by Jaaskelainen that uses machine learning modeling to provide additional sensor data for generating the performance curve, such that in combination the system uses a machine learning model that has been trained based on measurement data from the downhole environment.
The motivation would have been to select/utilize a model that is trained in a manner that accounts for downhole operations/conditions to optimize accuracy of modeling output as suggested by Jaaskelainen.
As to claim 14, the combination of Moricca and Jaaskelainen teaches “[t]he system of claim 11,” and Jaaskelainen further teaches “wherein the machine learning model is further trained based on downhole environments having different characteristics ([0015] synthesis model trained using historical downhole sensing data or from offset (other) wells).”
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of training the machine learning modeling using measurement data different downhole environments (downhole environments having different characteristics) to the system taught by Moricca as modified by Jaaskelainen that uses machine learning modeling to provide additional sensor data for generating the performance curve, such that in combination the system includes uses a machine learning model that has been trained based on downhole environments having different characteristics.
The motivation would have been to select/utilize a model that is trained in a manner that accounts for downhole operations/conditions to optimize accuracy over a broader range of operational/environmental conditions, resulting a more robust modeling for complex and dynamic downhole operations and conditions as suggested by Jaaskelainen.
As to claim 16, the combination of Moricca and Jaaskelainen teaches “[t]he system of claim 11,” and Moricca teaches “wherein the performance curve is updated using” “third measurement data ([0023] performance curve(s) generated/updated based on nodal model, which is based on the measured data that per [0014] may include data from multiple downhole instruments)” but does not appear to teach
“determine that a third equipment becomes unavailable during the run; and
predict third measurement data associated with the third equipment during a time that the third equipment is unavailable,
wherein the performance curve is updated using “the predicted” third measurement data.
Jaaskelainen discloses a method for monitoring and managing well operations including pump operations (Abstract) for circumstances in which it is determined that sensing data is unavailable during the monitoring operations ([0002] well may not be tooled with particular sensors that would be useful for AI processing; [0016] synthetic model used for circumstances in which certain sensors may be unavailable (Examiner note the unavailability of the sensor(s) is inherently entailed in the determination to configure a synthetic model to obtain corresponding “predicted” data). Jaaskelainen further teaches predicting additional measurement data associated with other (third) equipment during a time that the other equipment is unavailable by combining actual sensor measurement values and “synthetic” values that are values predicted (in terms of what an actual sensor value is expected to be) using a machine learning model ([0016] and [0033] AI synthetic data model generates unavailable sensor data with respect to pump operations; [0015] and [0030] AI model is trained (machine learning); claim 1).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching of using a machine learning model to predict, in association with a determination that sensor equipment is/will be unavailable, pump operation related measurements to the system taught by Moricca as modified by Jaaskelainen in which multiple sensor measurements are used for generating the performance curve such that in combination the system is configured for determining that a third equipment becomes unavailable during the run and predicting an additional (third) measurement from additional (third) equipment during the run by using a machine learning model (predicting), such that the performance curve is updated using the “predicted” third measurement data.
A motivation would have been to enable use of potentially useful sensor data that may not be otherwise available by actual sensor measurement as disclosed by Jaaskelainen ([0016]). Furthermore, such a combination would amount to applying a known design option for providing additional sensor data for pump monitoring to achieve predictable results.
As to claim 18, the combination of Moricca and Jaaskelainen teaches “[t]he system of claim 11, wherein an initial performance curve is generated (Moricca: FIG. 4 blocks 410 (model is updated), 412 (corresponding performance curves generated that per block 414 constitute updated performance curves), and the performance curve changes based on measurement data during the run (Moricca: FIG. 4 blocks 402, 404, 410 and 412; [0022] performance curves generated to correspond to analysis models; [0023] analysis models updated by on measured conditions (ESP data collected via data acquisition subsystem 310)).”
As to claim 20, the combination of Moricca and Jaaskelainen teaches “[t]he system of claim 11, wherein the processor is configured to execute the instructions and cause the processor to:
estimate a recommended flow rate for the material (Moricca: FIG. 2E performance curves 244 including a Best Efficiency Curve including a liquid rate (bb/day) corresponding to the head value and table 246 providing user-selectable configuration including flow rates) to minimize deterioration of at least one of the first equipment or the second equipment (Moricca: Best Efficiency Curve in FIG. 2E conveys an optimal operating efficiency that would inherently minimize equipment deterioration) based on the downhole environment (Moricca: FIG. 2E performance curves including Best Efficiency Curve associate pump head (downhole operation parameter is part of downhole environment) with flow rate. Examiner notes that the correspondence/intersection of the flow rate with pump head constitutes interdependence of head/flow rate in terms of operation efficiency).”
The Examiner notes that “to minimize deterioration” conveys an intended result/purpose that does not further limit the function of the recited method and is therefore not given patentable weight.
Claims 5 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Moricca in view of Jaaskelainen as applied to claims 1 and 11 above, and further in view of Zhou (US 2022/0188775 A1) and Davidson (US 2021/0123443 A1).
As to claim 5, the combination of Moricca and Jaaskelainen teaches “[t]he method of claim 1,” and Jaaskelainen further teaches that the machine learning model may be a label-classifier type model ([0052] modeled trained via supervised learning; claim 5) “configured to identify classifications of the downhole environment ([0015] synthesis model trained using downhole environment features such as downhole pressure, microseismic data, etc. that per [0052] are used in supervised learning and therefore are labelled for classification)” “and predict the second measurement data based on training data associated with downhole environments having the identified classifications ([0015] synthesis model trained using historical downhole sensing data or from offset (other) wells; [0028]-[0030] downhole sensor data used in training process. Examiner notes that the training process inherently entails application of feature classifications that would be applied for offset well data.).”
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching that the machine learning model may be a label-classifier type model configured to identify classifications of the downhole environment and predict the second measurement based on training data similar to the downhole environment to the method taught by Moricca as modified by Jaaskelainen that uses machine learning modeling to provide additional sensor data for generating the performance curve, such that in combination the method includes using a label-classifier type model configured to identify classifications of the downhole environment and predict the second measurement based on training data similar to the downhole environment.
The motivation would have been to select/utilize a model that is trained in a manner that accounts for downhole operations/conditions to optimize accuracy of modeling output as suggested by Jaaskelainen. The use of a supervised learning model (a label-classifier type model configured to identify classifications) as the particular type of model would amount to selecting a known design option for implementing machine learning modeling to ascertain predictive outputs based on input features to achieve predictable results.
Neither Moricca nor Jaaskelainen expressly teaches that the classifier model is a “multi-label classifier.”
Prior to the effective filing date, multi-label classifiers were a well-known type of learning classifier option applicable for a variety of applications including pump operational conditions predictions. For example, Zhou discloses a method/system for applying multi-label classification for pump management ([0001]) that includes using a multi-label classifier for determining downhole pump operational conditions (FIG. 2 output of model aggregator 206 used for multi-label classification; FIG. 5 model template 500 including dynamic inputs 108 (per [0038] may include sensor monitoring data) processed by aggregation model to provide multi-label outputs, [0046]; [0034]-[0035] model applied for predicting pump operating conditions (e.g., failure conditions)) for submersible pumps ([0036]).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Zhou’s teaching of using a multi-label classifier for determining downhole pump operational conditions for predicting pump operating conditions to the method taught by Moricca as modified by Jaaskelainen to including using machine learning to provide additional performance curve information such that in combination the method includes implementing the classifier as a multi-label classifier.
The motivation would have been to leverage the more comprehensive labeling capability of a multi-label classifier to enable more accurate and precise feature labeling to ultimately enable more accurate and precise machine learning output as suggested by Zhou.
Regarding the model being configured to identify classifications of the downhole environment “according to at least two of viscosity of fluid, fluid density, presence of multi-phase flow, presence of scale, and presence of particles,” Jaaskelainen discloses using multiple types of classification data ([0015] that may correspond to any one of a variety of different sensor data ([0027]-[0028]) and in which the sensor data (that may be modeled using machine learning) may include fluid characteristics ([0028] and [0034] sensor data may include fluid types/volumes).
The use of sensed fluid characteristics such as viscosity and density for monitoring pump operating condition/health was known prior to the effective filing date as disclosed by Davidson (Abstract monitoring pump using multiple sensors; [0038] environmental sensors may include slurry density meters, viscometers, and/or other fluid-related sensors).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Davidson’s teaching of using a combination of fluid viscosity and fluid density sensor measurements for monitoring pump performance to the method taught by Moricca as modified by Jaaskelainen, which teaches using multi-classification machine learning modeling for replacing absent sensor data, such that in combination at least two of the sensor classifications implemented via machine learning modeling/classification are fluid viscosity and fluid density.
The motivation would have been to leverage the flexibility of machine learning classification flexibility to provide downhole parameters that are known useful metrics for monitoring pump operations.
As to claim 15, the combination of Moricca and Jaaskelainen teaches “[t]he system of claim 11,” and Jaaskelainen further teaches that the machine learning model may be a label-classifier type model ([0052] modeled trained via supervised learning; claim 5) “configured to identify classifications of the downhole environment ([0015] synthesis model trained using downhole environment features such as downhole pressure, microseismic data, etc. that per [0052] are used in supervised learning and therefore are labelled for classification)” “and predict the second measurement based on training data associated with downhole environments having the identified classifications ([0015] synthesis model trained using historical downhole sensing data or from offset (other) wells; [0028]-[0030] downhole sensor data used in training process. Examiner notes that the training process inherently entails application of feature classifications that would be applied for offset well data.).”
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Jaaskelainen’s teaching that the machine learning model may be a label-classifier type model configured to identify classifications of the downhole environment and predict the second measurement based on training data similar to the downhole environment to the method taught by Moricca as modified by Jaaskelainen that uses machine learning modeling to provide additional sensor data for generating the performance curve, such that in combination the method includes using a label-classifier type model configured to identify classifications of the downhole environment and predict the second measurement based on training data similar to the downhole environment.
The motivation would have been to select/utilize a model that is trained in a manner that accounts for downhole operations/conditions to optimize accuracy of modeling output as suggested by Jaaskelainen. The use of a supervised learning model (a label-classifier type model configured to identify classifications) as the particular type of model would amount to selecting a known design option for implementing machine learning modeling to ascertain predictive outputs based on input features to achieve predictable results.
Neither Moricca nor Jaaskelainen expressly teaches that the classifier model is a “multi-label classifier.”
Prior to the effective filing date, multi-label classifiers were a well-known type of learning classifier option applicable for a variety of applications including pump operational conditions predictions. For example, Zhou discloses a method/system for applying multi-label classification for pump management ([0001]) that includes using a multi-label classifier for determining downhole pump operational conditions (FIG. 2 output of model aggregator 206 used for multi-label classification; FIG. 5 model template 500 including dynamic inputs 108 (per [0038] may include sensor monitoring data) processed by aggregation model to provide multi-label outputs, [0046]; [0034]-[0035] model applied for predicting pump operating conditions (e.g., failure conditions)) for submersible pumps ([0036]).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Zhou’s teaching of using a multi-label classifier for determining downhole pump operational conditions for predicting pump operating conditions to the method taught by Moricca as modified by Jaaskelainen to including using machine learning to provide additional performance curve information such that in combination the method includes implementing the classifier as a multi-label classifier.
The motivation would have been to leverage the more comprehensive labeling capability of a multi-label classifier to enable more accurate and precise feature labeling to ultimately enable more accurate and precise machine learning output as suggested by Zhou.
Regarding the model being configured to identify classifications of the downhole environment “according to at least two of viscosity of fluid, fluid density, presence of multi-phase flow, presence of scale, and presence of particles,” Jaaskelainen discloses using multiple types of classification data ([0015] that may correspond to any one of a variety of different sensor data ([0027]-[0028]) and in which the sensor data (that may be modeled using machine learning) may include fluid characteristics ([0028] and [0034] sensor data may include fluid types/volumes).
The use of sensed fluid characteristics such as viscosity and density for monitoring pump operating condition/health was known prior to the effective filing date as disclosed by Davidson (Abstract monitoring pump using multiple sensors; [0038] environmental sensors may include slurry density meters, viscometers, and/or other fluid-related sensors).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Davidson’s teaching of using a combination of fluid viscosity and fluid density sensor measurements for monitoring pump performance to the method taught by Moricca as modified by Jaaskelainen, which teaches using multi-classification machine learning modeling for replacing absent sensor data, such that in combination at least two of the sensor classifications implemented via machine learning modeling/classification are fluid viscosity and fluid density.
The motivation would have been to leverage the flexibility of machine learning classification flexibility to provide downhole parameters that are known useful metrics for monitoring pump operations.
Claims 9 and 19 are rejected under 35 U.S.C. 103 as being unpatentable over Moricca in view of Jaaskelainen as applied to claims 8 and 18, and in further view of Fowler (US 2021/0017999 A1).
As to claim 9, the combination of Moricca and Jaaskelainen teaches “[t]he method of claim 8,” but does not appear to teach initializing modeling using standardized performance testing and therefore does not teach “wherein the initial performance curve of the first equipment is based on a standardized performance test of the first equipment using a calibration material.”
Fowler discloses a method for controlling/configuring pump operations (Abstract) by that includes using initial pump performance curves (FIG. 3 depicting manufacturer’s pump curves; FIG. 4 block 402, [0059]; FIG. 4 block 418 depicting generated curves based on processing of initial curve data (manufacturer specification curves per [0059]) input at block 402; FIG. 8 block 802 (receive specification curve), block 808 (receive sensor data using pump ops), block 810 (update model including, per [0093], adjusting corresponding performance curve according to the sensor data ) that is based on standardized testing/evaluation of the pump ([0055] manufacturer’s pump curves from a pump manufacturer’s specifications), and further that manufacturer’s specification curves may be based on performance testing using a calibration material ([0023] manufacturer specification curves may describe performance for just water).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Fowler’s teaching of using a manufacturer’s specification curve(s) as an initial curve for pump performance monitoring in which the specification curve is based on performance testing using a calibration material to the method taught by Moricca as modified by Jaaskelainen such that in combination the method uses an initial performance curve of the first equipment that is based on a standardized performance test of the first equipment using a calibration material.
Such a combination would amount to selecting a known design option for initializing performance curve monitoring of pump operations to achieve predictable results.
As to claim 19, the combination Moricca and Jaaskelainen teaches “[t]he system of claim 18,” but does not appear to teach initializing modeling using standardized performance testing and therefore does not teach “wherein the initial performance curve of the first equipment is based on a standardized performance test of the first equipment using a calibration material.”
Fowler discloses a method for controlling/configuring pump operations (Abstract) by that includes using initial pump performance curves (FIG. 3 depicting manufacturer’s pump curves; FIG. 4 block 402, [0059]; FIG. 4 block 418 depicting generated curves based on processing of initial curve data (manufacturer specification curves per [0059]) input at block 402; FIG. 8 block 802 (receive specification curve), block 808 (receive sensor data using pump ops), block 810 (update model including, per [0093], adjusting corresponding performance curve according to the sensor data ) that is based on standardized testing/evaluation of the pump ([0055] manufacturer’s pump curves from a pump manufacturer’s specifications), and further that manufacturer’s specification curves may be based on performance testing using a calibration material ([0023] manufacturer specification curves may describe performance for just water).
It would have been obvious to one of ordinary skill in the art before the effective filing date, to have applied Fowler’s teaching of using a manufacturer’s specification curve(s) as an initial curve for pump performance monitoring in which the specification curve is based on performance testing using a calibration material to the system taught by Moricca as modified by Jaaskelainen such that in combination the system uses an initial performance curve of the first equipment that is based on a standardized performance test of the first equipment using a calibration material.
Such a combination would amount to selecting a known design option for initializing performance curve monitoring of pump operations to achieve predictable results.
Claims 21-22 are rejected under 35 U.S.C. 103 as being unpatentable over Moricca in view of Jaaskelainen as applied to claims 1 and 11, and in further view of Linnell (US 2026/0174949 A1) (publication provided as an attachment because not currently available on Google Patents).
Regarding claims 21 and 22, the combination of Moricca and Jaaskelainen teaches the method of claim 1 and system of claim 11, “wherein the second equipment includes a downhole sensor operable to provide measured second measurement data (Moricca: a second of the multiple sensors described in [0014] measurement data may be collected using multiple different downhole instruments; [0023] performance curve(s) generated/updated based on nodal model, which is based on the measured data that per [0014] may include data from multiple downhole instruments) over a communication link (Morrica: FIG. 1 depicting signal path 128 between sensors 118 and control panel 132, [0015]; FIG. 1 depicting wireless link between sensors 118 (via control panel 132) and computer system 45),
and determining that the second equipment is unavailable (Jaaskelainen as combined with Moricca for claim 1 teaches determining unavailability of sensors),” but neither Moricca nor Jaaskelainen teaches that the determination of second equipment unavailability “includes detecting a failure of the communication link.”
Sensor/sensor data unavailability via a loss of a communication link would have been an evident possibility to one of ordinary skill in the art prior to the effective filing date. For example, Linnell discloses a pump control method (Abstract) in which control operations may be affected by a determined “offline” condition in which a loss of a sensor communication link is detected ([0125] control process loses communications (detection inherent) to the primary remote pressure sensor).
It would have been obvious to one of ordinary skill in the art before the effective filing date, in view of Linnell’s teaching of the possibility of and determining of a loss of a sensor communication link, to have modified the method/system taught by Moricca as modified by Jaaskelainen which teaches determining/awareness of lack of a sensor data input to include that the method/system is configured to determine second equipment unavailability via detecting loss of a sensor communication link.
Such a combination would amount to selecting a known design option for determining sensor data unavailability to achieve predictable results.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MATTHEW W BACA whose telephone number is (571)272-2507. The examiner can normally be reached Monday - Friday 8:00 am - 5:30 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Andrew Schechter can be reached at (571) 272-2302. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/MATTHEW W. BACA/Examiner, Art Unit 2857
/ANDREW SCHECHTER/Supervisory Patent Examiner, Art Unit 2857
Prosecution Insights