Determining well productivity for hydraulically fractured wells
Methods and systems for determining well productivity include acquiring measurement data from a plurality of in-situ sensors located within a hydraulically fractured subterranean formation; classifying the noise degree for sensors of the plurality based on the acquired measurement data; selecting sensors from the plurality by minimizing noise degree while maintaining coverage of the subterranean formation above a user defined threshold; extracting data from the selected sensors; and estimating fracture half-length and well productivity potential based on the extracted data.
1 . A method for determining well productivity, the method comprising:
acquiring measurement data from a plurality of in-situ sensors located within fractures in a hydraulically fractured well in a subterranean formation;
classifying a noise degree for sensors of the plurality, the noise degree representing a quality of the acquired measurement data from the sensors of the plurality;
reducing effects of noise in the measurement data while maintaining coverage of the fractures above a threshold by selecting sensors from the plurality based on the noise degree classification and based on a physical distribution of the in-situ sensors in the fractures;
extracting data from the selected sensors having the physical distribution and having the noise degree satisfying the threshold; and
estimating fracture half-length and well productivity potential based on the extracted data.
2 . The method of claim 1 , wherein the minimizing comprises a mixed-integer programming framework.
3 . The method of claim 1 , further comprising: filtering measurement data from the plurality of in-situ sensors to remove noise from the measurement data, the filtering comprising artificial intelligence (AI) window filtering that is based on a radial basis function neural network.
4 . The method of claim 1 , further comprising: accessing, from a data store, data comprising at least one of rock property data, hydraulic fracturing parameter data, and sensor location data; combining the accessed data with the extracted data from the selected sensors; and estimating fracture half-length and well productivity potential based on the combined data.
5 . The method of claim 1 , further comprising: estimating a well productivity based on at least one of the estimated fracture half-length and well productivity potential, wherein the estimating comprises a decline curve analysis.
6 . The method of claim 5 , wherein the decline curve analysis comprises a long short-term memory framework to predict declines in production based on a time-series of well productivity.
7 . The method of claim 5 , further comprising: monitoring a production well to determine well leakage based on the estimated well productivity and the measured production.
8 . The method of claim 1 , wherein estimating fracture half-length and well productivity potential is based on a pretrained XGBoost machine learning model.
9 . The method of claim 1 , wherein the measurement data include at least one of temperature, pressure, and chemical concentration.
10 . A system for estimating well productivity, the system comprising:
a plurality of in-situ sensors;
a base station;
at least one processor; and
a memory storing instructions that when executed by the at least one processor cause the at least one processor to perform operations comprising:
acquiring measurement data from the plurality of in-situ sensors located within fractures in a hydraulically fractured well in a subterranean formation;
classifying a noise degree for sensors of the plurality, the noise degree representing a quality of the acquired measurement data from the sensors of the plurality;
reducing effects of noise in the measurement data while maintaining coverage of the fractures above a threshold by selecting sensors from the plurality based on the noise degree classification and based on a physical distribution of the in-situ sensors in the fractures;
extracting data from the selected sensors having the physical distribution and having the noise degree satisfying the threshold; and
estimating fracture half-length and well productivity potential based on the extracted data.
11 . The system of claim 10 , wherein the in-situ sensors comprise an energy harvesting module to harvest vibrational energy.
12 . The system of claim 10 , wherein the in-situ sensors induce vibrations within fractures of the subterranean formation larger than microseismic events emitted by the fracture to improve detectability of microseismic events by geophones.
13 . The system of claim 10 , the operations further comprising:
filtering measurement data from the plurality of in-situ sensors to remove noise from the measurement data, the filtering comprising artificial intelligence (AI) window filtering that is based on a radial basis function neural network.
14 . The system of claim 10 , the operations further comprising:
estimating a well productivity based on at least one of the estimated fracture half-length and well productivity potential, wherein the estimating comprises a decline curve analysis comprising a long short-term memory framework to predict declines in production based on a time-series of well productivity.
15 . The system of claim 10 , wherein estimating fracture half-length and well productivity potential is based on a pretrained XGBoost machine learning model.
16 . One or more non-transitory machine-readable storage devices storing instructions for determining well productivity, the instructions being executable by one or more processing devices to cause performance of operations comprising:
acquiring measurement data from a plurality of in-situ sensors located within fractures in a hydraulically fractured well in a subterranean formation;
classifying a noise degree for sensors of the plurality, the noise degree representing a quality of the acquired measurement data from the sensors of the plurality;
reducing effects of noise in the measurement data while maintaining coverage of the fractures above a threshold by selecting sensors from the plurality based on the noise degree classification and based on a physical distribution of the in-situ sensors in the fractures;
extracting data from the selected sensors having the physical distribution and having the noise degree satisfying the threshold; and
estimating fracture half-length and well productivity potential based on the extracted data.
17 . The non-transitory machine-readable storage devices of claim 16 , the operations further comprising:
filtering measurement data from the plurality of in-situ sensors to remove noise from the measurement data, the filtering comprising artificial intelligence (AI) window filtering that is based on a radial basis function neural network.
18 . The non-transitory machine-readable storage devices of claim 16 , the operations further comprising:
estimating a well productivity based on at least one of the estimated fracture half-length and well productivity potential, wherein the estimating comprises a decline curve analysis comprising a long short-term memory framework to predict declines in production based on a time-series of well productivity.
19 . The non-transitory machine-readable storage devices of claim 18 , the operations further comprising:
monitoring a production well to determine well leakage based on the estimated well productivity and the measured production.
20 . The non-transitory machine-readable storage devices of claim 16 , wherein estimating fracture half-length and well productivity potential is based on a pretrained XGBoost machine learning model.