Methods for predicting and monitoring downhole salinity variations
Methods and systems are provided for characterizing connate water salinity and resistivity of a subsurface formation. Well log data including resistivity and spontaneous potential (SP) log data are measured by at least one downhole tool disposed within a borehole. The resistivity and SP log data are inverted to determine a resistivity model and an SP model, which are used to determine connate water resistivity. The connate water resistivity is used to determine connate water salinity. The connate water salinity derived from the inversion of resistivity log data and SP log data (or derived from a trained ML system supplied with such log data) can be used as a baseline measure of connate water salinity, and this baseline measure can be evaluated together with the connate water salinity estimates derived from pulsed neutron tool measurements over time-lapsed periods of production to monitor variation in connate water salinity due to production.
1. A method for characterizing connate water salinity and resistivity of a subsurface formation, comprising:
i) obtaining, via a processing system, well log data measured by at least one downhole tool disposed within a borehole that traverses the subsurface formation, wherein the well log data is acquired during a drilling operation performed by a drill string within the borehole, and wherein the well log data includes resistivity log data and spontaneous potential (SP) log data;
ii) performing, via the processing system, inversion operations that invert the resistivity log data of i) to determine a resistivity model of the subsurface formation, wherein the inversion operations of ii) are based on an assumption of a step profile invasion for mud filtrate invasion;
iii) determining, via the processing system, a connate water resistivity of the subsurface formation for an interval of the borehole associated with a depth range by minimizing a cost function of the SP log data, wherein the cost function comprises the SP log data, the depth range of the borehole, a resistivity of mud in the subsurface formation, a resistivity of mud filtrate in the subsurface formation, the resistivity model, and a plurality of properties associated with the borehole obtained from the well log data;
iv) determining, via the processing system, a connate water salinity of the subsurface formation based on the connate water resistivity of the subsurface formation for the interval of the borehole as determined in iii), wherein the connate water salinity of the subsurface formation corresponds to an invaded zone of the subsurface formation associated with the drilling operation; and
v) using, via the processing system, the connate water salinity of the subsurface formation of iv) as training data for training a machine learning model, wherein the machine learning model predicts connate water salinity of an undisturbed zone of the subsurface formation for future drilling operations based on a result of the training and an input data vector associated with the undisturbed zone of the subsurface formation, and wherein the undisturbed zone of the subsurface formation is different from the invaded zone of the subsurface formation.
2. The method of claim 1 , wherein:
the inversion operations of ii) employ at least one parameter that affects spontaneous potential of the borehole and/or the connate water resistivity of the subsurface formation, wherein a data value for the at least one parameter is obtained from surface measurements or well log measurements.
3. The method according of claim 2 , wherein:
the at least one parameter is selected from the group consisting of: the resistivity of the mud (R m ), the resistivity of mud filtrate (R mf ), borehole diameter (D h ), formation temperature (T), formation thickness (h), and combinations thereof.
4. The method according of claim 1 , wherein:
the resistivity model includes a diameter of an invasion zone (D i ), resistivity of the invaded zone (R xo ), and resistivity of the undisturbed zone (R t ).
5. The method of claim 1 , wherein:
the cost function is in the form:
ζ
(
R
w
)
=
∫
dz
[
SP
M
(
z
)
-
SP
(
z
,
R
w
,
R
mf
,
R
m
,
T
,
R
xo
,
R
t
,
D
i
,
D
h
,
h
)
]
2
where SP M is the measured SP log data for the interval of the borehole specified by the depth range z,
R m is the resistivity of the mud, R mf is the resistivity of mud filtrate, D h is a borehole diameter, Tis a subsurface formation temperature, h is a subsurface formation thickness, and each is obtained from surface measurements or the well log data;
D i is diameter of an invasion zone, where D i , R xo , and R t are part of the resistivity model of the subsurface formation that is determined through inversion of the resistivity log data; and
R w is the connate water resistivity.
6. The method of claim 1 , wherein:
the inversion operations of ii) are based on an assumption that a linear relationship exists between resistivity and salinity for both mud filtrate and connate water.
7. The method of claim 1 , wherein:
the inversion operations of ii) are based on an assumption that a relationship between resistivity and salinity for both mud filtrate and connate water is dependent on temperature, salt concentration, and salt composition.
8. The method of claim 1 , wherein:
the inversion operations of ii) employ a parametric model or a pixel-based inversion.
9. The method of claim 1 , wherein:
the resistivity log data of i) is obtained from an array induction resistivity tool disposed within the borehole.
10. The method of claim 1 , wherein:
the at least one downhole tool of i) comprises a logging-while drilling tool or a wireline tool.
11. The method of claim 1 , wherein:
the input data vector further includes parameter data for at least one parameter that affects spontaneous potential of the borehole and/or the connate water resistivity of the subsurface formation, wherein the parameter data is obtained from surface measurements or well log measurements.
12. The method of claim 11 , wherein:
the at least one parameter is selected from the group consisting of: the resistivity of the mud (R m ), the resistivity of mud filtrate (R mf ), borehole diameter (D h ), formation temperature (T), formation thickness (h), and combinations thereof.
13. The method of claim 1 , further comprising:
monitoring one or more variations in the connate water salinity of the invaded zone by using the downhole tool to obtain pulsed neutron (PN) log data, wherein the PN log data indicates one or more additional variations in the connate water salinity over one or more time-lapsed periods of hydrocarbon production; and
using the PN log data as additional training data for training the machine learning model.
14. The method of claim 13 , wherein:
the one or more time-lapsed periods of hydrocarbon production involve the application of at least one enhanced oil recovery (EOR) technique.
15. The method of claim 14 , wherein:
the at least one EOR technique comprises natural gas driven production, water flooding, or other primary, secondary, and tertiary EOR techniques.
16. The method of claim 13 , wherein:
the PN log data comprises dielectric dispersion measurements and optionally pulsed-neutron sigma measurements of the subsurface formation.
17. The method of claim 13 , wherein:
the PN log data measures elemental yields for chlorine, oxygen, hydrogen, or any combination thereof in the subsurface formation.
18. The method of claim 13 , wherein:
the downhole tool employs a pulsed neutron generator with multiple scintillation gamma ray detectors to obtain the PN log data.
19. The method of claim 1 , wherein:
the machine learning model is trained using laboratory fluid analysis of physical samples of connate water obtained from the borehole.
20. The method of claim 1 , wherein:
the machine learning model is a nonlinear regression model or a deep learning neural network model.