Patent Application
20200319282
Understanding the characteristics of geologic formations and fluids located therein is important for effective hydrocarbon exploration and production. Operations such as drilling, formation evaluation and production rely on accurate petrophysical interpretation derived from a diverse set of logging technologies.
Nuclear Magnetic Resonance (NMR) is a technique that is used in the energy industry to estimate properties of subterranean formations. NMR typically involves applying a constant magnetic field and an oscillating magnetic field to a formation region, and detecting NMR signals due to voltage induced in a detector by precession of the nuclear spins of atomic nuclei in the region. NMR signals can be analyzed to estimate formation properties such as porosity, permeability and fluid saturations.
An embodiment of a method of estimating properties of a resource bearing formation includes receiving, by a processor, a measured echo train generated by a nuclear magnetic resonance (NMR) measurement device deployed in a region of interest, and a measured noise of the measured echo train, and calculating a T2 distribution subject to a nonlinear equality constraint, the nonlinear equality constraint dependent on the measured noise and a fitting error between the measured echo train and a modeled echo train. Calculating the T2 distribution includes estimating a solution for the T2 distribution that is closest to satisfying the nonlinear equality constraint.
An embodiment of a system for estimating properties of a resource bearing formation includes an input device configured to receive a measured echo train generated by a nuclear magnetic resonance (NMR) measurement device deployed in a region of interest, and a measured noise of the measured echo train. The system also includes a processing device configured to calculate a T2 distribution subject to a nonlinear equality constraint, the nonlinear equality constraint dependent on the measured noise and a fitting error between the measured echo train and a modeled echo train. Calculating the T2 distribution includes estimating a solution for the T2 distribution that is closest to satisfying the nonlinear equality constraint.
The subject matter which is regarded as the invention is particularly pointed out and distinctly claimed in the claims at the conclusion of the specification. The foregoing and other features and advantages of the invention are apparent from the following detailed description taken in conjunction with the accompanying drawings in which:
Methods, systems and apparatuses for calculating or estimating properties of resource bearing formations using magnetic resonance techniques are described herein. Systems and methods are provided for performing nuclear magnetic resonance measurements, and estimating or calculating properties such as porosity, permeability and fluid saturation based on inversion of T2 distributions.
An embodiment of an inversion method is provided to compute an optimal solution for a T2 distribution based on echo train data and measured noise. The inversion method utilizes a new smoothing factor for suppressing the sensitivity of the solution for the T2 distribution to the measured noise of echo trains, and a new objective function to optimize the solution for the T2 distribution. The objective function is defined as the squared Euclidean distance between one or more modeled echo trains and one or more measured echo trains, and includes a regularization factor or term for suppressing the noise effects on the one or more modeled echo trains. The objective function has the value of half of the total sum of squared measured noise of an echo train (or echo trains) and squared fitting errors between one or more measured echo trains and one or more modeled echo trains, if the estimated T2 distribution is convergent to the true T2 distribution. The optimal solution for the T2 distribution is the unique solution that minimizes the objective function and satisfies a new nonlinear equality constraint. The smoothing factor, in one embodiment, is defined as a ratio of the L2-norm of measured noise to the L2-norm of the optimal solution for the T2 distribution.
In one embodiment, a processing device is configured to calculate the optimal solution for the T2 distribution according to a nonlinear optimization algorithm. According to the algorithm, an initial solution or selected value of the smoothing factor is proposed, and an initial search range for the smoothing factor is determined based on the initial selected value of the smoothing factor. The algorithm directs an iterative process where, for each iteration, a new solution for the T2 distribution is calculated and a new search range is determined. In addition, a new value for the smoothing factor is determined and the new search range is determined by reducing the search range of a previous iteration. The iteration process continues until the search range for the smoothing factor is below a selected range.
Embodiments described herein provide a number of advantages and technical effects. For example, the embodiments provide an efficient, robust and data-oriented method for inversion of echo trains, including noisy echo trains. In addition, the embodiments provide a more accurate estimation than conventional techniques, which can be executed with a smaller computational cost than conventional techniques.
The system 10 includes a magnetic resonance device such as an NMR tool 18. The tool 18 is configured to generate magnetic resonance data for use in estimating fracture characteristics of a formation, and/or other characteristics such as porosity, pore size, irreducible water saturation, permeability, hydrocarbon content, fluid viscosity and others. The tool 18 may be configured as a component of various subterranean systems, such as wireline well logging and logging-while-drilling (LWD) systems.
In the embodiment of
The NMR tool 18 includes a static magnetic field source 24, such as a permanent magnet assembly, that magnetizes formation materials and a transmitter and/or receiver assembly 26 (e.g., an antenna or antenna assembly) that transmits radio frequency (RF) energy or pulsed energy that provides an oscillating magnetic field in the formation, and detects NMR signals as voltages induced in the receiver. The assembly 26 may serve the receive function, or distinct receiving antennas may be used for that purpose.
In one embodiment, the NMR tool 18 and/or other downhole components are equipped with transmission equipment to communicate with other downhole components and/or with surface devices or components. For example, transmission equipment such as a cable 28 extends from the BHA 22 to surface equipment such as a drill rig 30. The transmission equipment may take any desired form, and different transmission media and methods may be used, such as wired, fiber optic, mud pulse telemetry and/or other wireless transmission methods.
The cable 28 or other transmission equipment can be used to communicate with one or more processing devices, such as a downhole electronics unit 32 and/or a surface processing unit 34. The processing device or devices include various electronic components to facilitate receiving signals and collecting data, transmitting data and commands, and/or controlling one or more aspects of an energy industry operation such as an LWD operation.
For example, the processing unit 34 includes an input/output device 36, a processor 38, and a data storage device 40 (e.g., memory, computer-readable media, etc.) for storing data 42 such as measurement data (e.g., echo train data and/or noise measurements), models, T2 distributions and others. The data storage device 40 also stores computer programs or software 44 that cause the processor to perform aspects of methods and processes described herein. The surface processing unit 34 (and/or the downhole electronics unit 32) may be configured to perform functions such as controlling drilling and steering, controlling the pumping of borehole fluid and/or cement injection, transmitting and receiving data, processing measurement data, and/or monitoring operations of the system 10. The surface processing unit 34 and/or an operator can control operational parameters such as drilling direction and fluid flow, and may also facilitate planning other operations, such as production and hydraulic fracturing operations. For example, the surface processing unit 34 is configured as a surface control unit which controls various parameters such as rotary speed, weight-on-bit, fluid flow parameters (e.g., pressure and flow rate) and others.
The surface processing unit 34, electronics unit 32 and/or other suitable processing device includes a processor configured to perform NMR measurements of a region or volume of interest in a formation, analyze NMR measurements and/or estimate properties of the formation based on the NMR measurements.
In one embodiment, the surface processing unit 34 is configured to perform a method that includes receiving NMR measurement data (e.g., echo train data), estimating a T2 distribution, and estimating one or more properties of a formation or formation region. Although the methods described herein are discussed as being performed by a processor or processing device, the methods are not so limited. For example, one or more aspects of the methods may be performed by a human operator or performed by a human operator in conjunction with the processor.
Magnetic resonance measurements are performed by the NMR tool 18, which generates a static magnetic field (B0) in a volume or region within the formation (a “region of interest”) using one or more magnets (e.g., the magnetic field source 24). An oscillating magnetic field (B1) is generated, which is at least substantially perpendicular to the static magnetic field in the volume of interest. The region of interest may be circular or toroidal around the borehole, and/or focused or directed toward a specific angular region (i.e., side-looking).
When exposed to the magnetic field B0, the spin axes of hydrogen nuclei in the formation precess around the direction of the B0 field with the Larmor frequency, which is proportional to the strength of the magnetic field B0. Over time, the spin axes align themselves at distinct angles along the B0 field and create a net magnetization (i.e., polarization), which will build up with the time constant T1, referred to as a longitudinal relaxation or spin lattice relaxation time. T2 is a time constant of the transversal relaxation, which describes the loss of magnetization in the plane orthogonal to the B0 field.
The NMR measurements include generating NMR signal echoes from atomic nuclei, such as hydrogen nuclei, of the formation. The NMR signal echoes are recorded as measured echo trains. The NMR measurements yield transverse relaxation times T2, which are exponential decay time constants that correspond to a characteristic or property of formation material. NMR measurements are analyzed by performing an inversion of measurement data to generate a quantitative distribution of T2 values, referred to as a T2 distribution. The process of obtaining this T2 distribution is commonly called T2 inversion. Properties such as total porosity, partial porosities, pore size and fluid type in the formation may be determined from the T2 distribution.
Conventional inversion methods attempt to minimize the sum of squared fitting errors between measured and modeled echo trains by a process of regularization or smoothing. Smoothing is performed to reduce or remove the effects of noise on the optimal solution of the T2 distribution. Generally, inversion methods attempt to minimize such errors by including a regularization term, referred to as a regularization factor or smoothing factor. The smoothing factor reduces the sensitivity of the T2 distribution solution to noise, and can increase the precision of peaks in the T2 distribution.
Conventional methods assume that measured echo trains have white noise, and do not explicitly take into account measured noises in echo trains as part of the inversion process. In downhole applications, the assumption of white noise may not be valid, which results in an incorrect selection of the smoothing factor. Although variants of inversion processes have been developed for optimizing the smoothing factor, the smoothing factor calculated in such processes depends on the selection of optimization parameters, such as the cutoff of fitting errors and levels of smoothing factors to be searched. Incorrect selection of the smoothing factor comprises the accuracy of estimated properties such as porosity, permeability and fluid saturations.
In contrast to conventional methods, which select a value for the smoothing factor and then calculate a fitting error, methods described herein utilize a new supplementary nonlinear equality constraint to find a unique and optimal smoothing factor and provide an optimal solution of the T2 distribution that is unaffected or substantially unaffected by noise. Estimation of the T2 distribution and the smoothing factor is a non-linear optimization problem; the value of the objective function is minimized and the smoothing factor is searched over a set of feasible alternatives. The nonlinear equality constraint is based on the fitting errors and measured noise.
Performing an inversion of echo train data includes using a forward model that relates echo train data to properties to be estimated, such as porosity, permeability, fluid saturation and others. In the following discussion, the forward model relates echo train data to porosity. However, the forward model is not so limited and can be adapted for estimation of other properties.
In one embodiment, the forward model used for the inversion of an echo train for long waiting time is represented by the following equation:
where tk is the sampling time of the kth echo of the echo train, and the total number of echoes is Nt. G is the gradient of the magnetic field B0, and TW is the waiting time. In one embodiment, the waiting time is selected to be sufficient for full polarization of spins under the static magnetic field B0. This waiting time is referred to as a long waiting time or long TW. s(tk) is the calibrated magnetic momentum signal of spins at sampling time tk, also referred to as an NMR measurement signal or echo train signal. T2,j is the jth bin of T2 transverse relaxation times, where the total number of relaxation time bins is NT2. p is the incremental porosity (expressed in porosity units or p.u.), and p(T2,j) is the incremental porosity of the jth bin. εk is the fitting error of the kth echo at time tk.
The forward model can also be represented in matrix format as:
where s is a column vector [s1, s2, s3 . . . , sNt-1, sNt]T, and p is a column vector [p1, p2, p3 . . . pNT2-1,pNT2]T representing the T2 distribution. E is a column vector [ε1, ε2, ε3 . . . εNt-1, εNt]T, and A is the matrix with elements e−tk/T2,j. It is noted that an element of A is not limited to the exponential function shown above; other functions such as a gamma function can be utilized.
In one embodiment, the objective function is defined as the squared Euclidean distance between the modeled echo trains (e.g., echo trains modeled according to equations (1) and (2)) and the measured echo trains, and includes the regularization term for suppressing the noise effects on the modeled echo train. The objective function (OBJ) can be represented by the following equation:
The optimal solution for the T2 distribution is the unique solution that minimizes the objective function, satisfying the non-negative constraint for the T2 distribution (i.e., p>0) and with the optimized smoothing factor.
The smoothing factor is denoted as β, and is used for reducing the sensitivity of the optimal solution of the T2 distribution (which in one embodiment is the incremental porosity distribution in the T2 domain) to the noise n of a measured echo train. In one embodiment, the smoothing factor β is defined as the ratio of the L2-norm of measured noise to the L2-norm of the solution for the T2 distribution.
In one embodiment, the smoothing factor β is calculated by determining a value for the smoothing factor that satisfies a nonlinear equality constraint. The nonlinear equality constraint is that the L2 norm of the fitting error of the modeled echo train is equal to the L2 norm of noise of the measured echo train, i.e., ∥ε∥2=∥n∥2.
In one embodiment, the smoothing factor β is defined as the ratio of the L2-norm of measured noise to the L2-norm of the solution of T2 distribution (the incremental porosity) p. The smoothing factor β can be represented by the equation:
where n is a vector of measured noise. ∥ ∥2 is the L2 norm of a vector. The objective function OBJ can be rewritten as a function of measured noise n and the fitting error ε as follows:
OBJ=½[(s−Ap)T(s−Ap)+β2pTp]=½(∥ε∥22+∥βp∥22) (5)
where T is the transpose operator.
In one embodiment, the processing device is configured to compute the optimal solution for the T2 distribution using an iterative algorithm, which utilizes an initially selected value for the smoothing factor (also referred to as a guess solution or initial guess value). The initial guess solution may be based on, for example, smoothing factor solutions taken from previous measurements or simulations.
In one embodiment, the initial guess value is based on the measured echo train and its measured noise. The initial guess value for the smoothing factor (denoted as βinit) is calculated as follows:
The constant c is selected to improve the accuracy of initial guess value of the smoothing factor. The constant c can be selected as any suitable value. For example, the value for the constant c can be selected as 4. In one embodiment, c can be selected to make the initial guess value of β be close to the value of β as defined in Eq. (4).
The initial guess value βinit is used to calculate its initial search range. The initial search range is used to calculate solutions for a new value of the smoothing factor and a new solution of T2 distribution (p), as well as a new search range for the smoothing factor. For each iteration, a new solution for the T2 distribution is found. In addition, for each iteration, the new search range is found by shrinking or reducing the previous search range, and a new value of the smoothing factor (βi) is determined based on the new search range. For example, the new value of the smoothing factor is selected as the median point in the new search range. The algorithm iterates until the search range for a selected smooth factor is below a selected error tolerance ε. At that point, the smooth factor is considered as the optimum smooth factor and the resultant T2 distribution can be calculated and is considered the optimum solution.
The method 60 is described in conjunction with the system 10 of
At stage 61, echo train data and measured noise data (noise train) are input to a processing device such as the surface processing unit 34.
At stage 62, an initial guess value for the smoothing factor (βinit), and an initial searching range for the smoothing factor are determined. The initial range is bounded by a maximum smoothing factor (βmax) and a minimum smoothing factor (δmax). The goal of the method 60 is to determine an optimum value of the smoothing factor that minimizes the objective function. In this embodiment, the smoothing factor is selected that results in a solution of the T2 distribution that comes closest to satisfying the normalized nonlinear equality constraint, i.e., results in a value of a normalized equality condition (Eβ) that is closest to one.
The optimal solution (popt) of the T2 distribution is the solution that minimizes the objective function OBJ, described as follows:
It is equivalent to the solution:
p
opt
=p|
min{[∥ε|
+∥βp∥
]}. (8)
This optimal solution satisfies the non-negative constraint:
p>0 (9)
and the nonlinear equality constraint:
∥n∥2=∥ε∥2. (10)
As shown, the solution p is selected to be non-negative, and the L2-norm of the fitting error ε is considered equal to the L2-norm of the measured noise n. This optimization problem is a non-negative least squares problem with an additional nonlinear equality constraint.
In one embodiment, an iterative algorithm is used to solve this nonlinear optimization problem. The initial minimum and maximum values of the smoothing factor are set as:
and the normalized equality condition Eβ values (also referred to as error norm ratio values) corresponding to the maximum and minimum searched smoothing factor are computed by:
where Eβ,max is the normalized equality condition corresponding to βmax, and Eβ,min is the normalized equality condition corresponding to βmin. The solutions of T2 distributions pmin and pmax, corresponding to βmin and βmax, are calculated with the iterative non-negative least squares algorithm to solve the non-negative least squares problem.
At stage 63, the processing device determines whether Eβ,min is greater than one. If so, at stage 64, βmax is reduced to be equal to the initial selected βmin, and βmin is reduced by half.
At stage 65, the processing device determines whether Eβ,max is less than one. If so, at stage 66, βmin is increased to be equal to the initial selected βmax, and βmax is doubled.
β (βi) at the first iteration is then calculated based on βmin and βmax.
At stage 67, the range of the normalized equality condition value Eβ is compared to a selected tolerance range for linearly predicting a new smoothing factor value βi at iteration i. For example, it is determined whether the difference between Eβ,min and Eβ,max is within a selected tolerance range, e.g., a range of about 0.002. At stage 68, if the difference falls within the tolerance range, βi is linearly predicted as follows:
At stage 69, if the difference falls outside of the tolerance range for linearization, βi is calculated as follows:
βi=βmin+0.365(βmax−βmin).
Referring to
At stage 71, the slope values of the normalized equality condition value Eβ are determined, indicated by smax and smin. smax and smin can be calculated according to the following equations:
At stage 72, the normalized equality condition value Eβ,i is compared to one, and the search range of βi is adjusted. At stage 73, if Eβ,i is less than one, βmin is updated with βi. At stage 74, if Eβ,i is greater than one, βmax is updated with βi. A new value for β (βi+1) and a new slope (si+1) of the normalized equality condition value Eβ can now be calculated.
Once the value of βi and the search range of the smoothing factor is determined, at stage 75, the ratio of smax to smin is calculated and compared to a selected value, e.g., a value of three. At stage 76, if the ratio is less than the selected value, βi+1 is calculated as:
βi+1=βmin+(1−Eβ,min)/si+1, where (16)
s
i+1=(Eβ,max−Eβ,min)/(βmax−βmin). (17)
At stage 77, if the ratio is greater than three, βi+1 is calculated as:
βi+1=βmin+0.365(βmax−βmin). (18)
At stage 78, the search range is compared to an error tolerance. For example, the processing device determines whether the difference between βmin and βmax is less than an error tolerance, e.g., a tolerance of 0.1. If not, stages 70-78 are repeated using the smoothing factor value of βi+1.
At stage 79, if the search range of the smoothing factor is within the error tolerance, the final iteration is reached to obtain the final optimal solutions of the smoothing factor and the T2 distribution.
Referring to
The T2 distributions shown in the above examples are plotted as amplitude values in arbitrary units (a.u.) as a function of T2 values. The amplitude values can be representative of one or more properties of a formation, such as incremental porosity (e.g., as porosity units or p.u.).
In the first stage 121, the NMR tool 18 or other magnetic resonance measurement tool is deployed into a borehole to a selected location. In one embodiment, the NMR tool 18 is deployed as part of a wireline operation, or during drilling as part of an LWD operation. The speed at which the NMR tool 18 is advanced is referred to as logging speed.
In the second stage 122, when the NMR tool 18 is located at the selected location, NMR measurements are performed by generating a static magnetic field B0 in a volume or region of interest around the selected location (a formation or formation region), and transmitting pulsed signals from at least one transmitting antenna into the region of interest. At least one receiving antenna detects NMR signals from the region of interest in response to the interaction between the nuclear spins of interest and the static and oscillating magnetic fields, and generates echo trains that may be measured at a plurality of depths or locations.
In the third stage 123, the NMR measurement data including raw time domain echo trains are processed to calculate a T2 distribution by inverting the echo train data into the T2 domain according to methods described herein (e.g., the method 60).
In the fourth stage 124, properties of the region of interest are estimated based on the T2 distribution. For example, the T2 distribution can be associated with porosity. Other properties that can be estimated include permeability, fluid saturation, fluid types and others.
In the fifth stage 125, aspects of an energy industry operation are performed based on the estimated properties (e.g., operational parameters are planned and/or adjusted). Examples of energy industry operations include drilling, stimulation, formation evaluation, measurement and/or production operations. For example, estimations of porosity can be used to control drilling direction and find productive zones that can be targeted for production.
In connection with the teachings herein, various analyses and/or analytical components may be used, including digital and/or analog subsystems. The system may have components such as a processor, storage media, memory, input, output, communications link (wired, wireless, pulsed mud, optical or other), user interfaces, software programs, signal processors and other such components (such as resistors, capacitors, inductors, etc.) to provide for operation and analyses of the apparatus and methods disclosed herein in any of several manners well-appreciated in the art. It is considered that these teachings may be, but need not be, implemented in conjunction with a set of computer executable instructions stored on a computer readable medium, including memory (ROMs, RAMs), optical (CD-ROMs), or magnetic (disks, hard drives), or any other type that when executed causes a computer to implement the method of the present invention. These instructions may provide for equipment operation, control, data collection and analysis and other functions deemed relevant by a system designer, owner, user, or other such personnel, in addition to the functions described in this disclosure.
One skilled in the art will recognize that the various components or technologies may provide certain necessary or beneficial functionality or features. Accordingly, these functions and features as may be needed in support of the appended claims and variations thereof, are recognized as being inherently included as a part of the teachings herein and a part of the invention disclosed.
While the invention has been described with reference to exemplary embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications will be appreciated by those skilled in the art to adapt a particular instrument, situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed as the best mode contemplated for carrying out this invention.
Set forth below are some embodiments of the foregoing disclosure:
A method of estimating properties of a resource bearing formation, the method comprising: receiving, by a processor, a measured echo train generated by a nuclear magnetic resonance (NMR) measurement device deployed in a region of interest, and a measured noise of the measured echo train; and calculating a T2 distribution subject to a nonlinear equality constraint, the nonlinear equality constraint dependent on the measured noise and a fitting error between the measured echo train and a modeled echo train, wherein calculating the T2 distribution includes estimating a solution for the T2 distribution that is closest to satisfying the nonlinear equality constraint.
The method as in any prior embodiment, wherein estimating the solution includes minimizing an objective function, the objective function including a regularization factor configured to reduce a sensitivity of the estimated solution to noise in the measured echo train.
The method as in any prior embodiment, wherein the objective function is minimized based on a constraint specifying that the T2 distribution is non-negative.
The method as in any prior embodiment, wherein the regularization factor is defined as a ratio of an L2-norm of the measured noise to an L2-norm of a solution for the T2 distribution.
The method as in any prior embodiment, wherein the nonlinear equality constraint is that the L2-norm of the fitting error is equal to the L2-norm of the measured noise.
The method as in any prior embodiment, wherein estimating the solution includes: selecting an initial value (βinit) of the regularization factor based on the measured noise and the measured echo train, calculating an initial search range based on the initial value βinit, and calculating a T2 distribution; iteratively calculating a solution of the T2 distribution, a search range and a solution of the regularization factor until the search range is below a selected error tolerance; and selecting the solution of the regularization factor associated with the search range that is below the selected error tolerance, and estimating the solution for the T2 distribution based on the selected solution of the regularization factor.
The method as in any prior embodiment, wherein the objective function (OBJ) is defined as:
OBJ=½(∥ε∥22+∥βp∥22),
wherein p is a solution of the T2 distribution, ε is the fitting error and β is a solution of the regularization factor.
The method as in any prior embodiment, wherein the solution is a non-negative solution (popt) that minimizes the objective function and is based on:
p
opt
=p|
min{[∥ε∥
+∥βp∥
]},
wherein the non-negative solution popt satisfies the non-linear equality constraint represented by:
∥n∥2=∥ε∥2.
The method as in any prior embodiment, wherein each search range of the regularization factor is based on a minimum value and a maximum value of a normalized equality condition, the normalized equality condition is defined as a ratio of the L2-norms of the fitting error and the measured noise.
The method as in any prior embodiment, wherein the minimum value (Eβ,min) of the normalized equality condition is based on:
wherein βmin is a selected minimum value of the regularization factor and n is the measured noise, and the maximum value (Eβ,max) of the normalized equality condition is based on:
wherein βmax is a selected maximum value of the regularization factor.
The method as in any prior embodiment, further comprising controlling an energy industry operation based on the estimated solution.
A system for estimating properties of a resource bearing formation, the system comprising: an input device configured to receive a measured echo train generated by a nuclear magnetic resonance (NMR) measurement device deployed in a region of interest, and a measured noise of the measured echo train; and a processing device configured to perform: calculating a T2 distribution subject to a nonlinear equality constraint, the nonlinear equality constraint dependent on the measured noise and a fitting error between the measured echo train and a modeled echo train, wherein calculating the T2 distribution includes estimating a solution for the T2 distribution that is closest to satisfying the nonlinear equality constraint.
The system as in any prior embodiment, wherein estimating the solution includes minimizing an objective function, the objective function including a regularization factor configured to reduce a sensitivity of the estimated solution to noise in the measured echo train.
The system as in any prior embodiment, wherein the objective function is minimized based on a constraint specifying that the T2 distribution is non-negative, and the regularization factor is defined as a ratio of an L2-norm of the measured noise to an L2-norm of the solution for the T2 distribution.
The system as in any prior embodiment, wherein the nonlinear equality constraint is that the L2-norm of the fitting error is equal to the L2-norm of the measured noise.
The system as in any prior embodiment, wherein estimating the solution includes: selecting an initial value (βinit) of the regularization factor based on the measured noise and the measured echo train, calculating an initial search range based on the initial value βinit, and calculating a T2 distribution; iteratively calculating a solution of the T2 distribution, a search range and a solution of the regularization factor until the search range is below a selected error tolerance; and selecting the solution of the regularization factor associated with the search range that is below the selected error tolerance, and estimating the solution for the T2 distribution based on the selected solution of the regularization factor.
The system as in any prior embodiment, wherein the objective function (OBJ) is defined as:
OBJ=½(∥ε∥22+∥βp∥22)
wherein p is a solution of the T2 distribution, ε is the fitting error and β is a solution of the regularization factor.
The system as in any prior embodiment, wherein the solution is a non-negative solution (popt) that minimizes the objective function and is based on:
p
opt
=p|
min{[∥ε∥
+∥βp∥
]},
wherein the non-negative solution popt satisfies the non-linear equality constraint represented by:
∥n∥2=∥ε∥2.
The system as in any prior embodiment, wherein each search range is based on a minimum value and a maximum value of a normalized equality condition, the normalized equality condition is defined as a ratio of the L2-norms of the fitting error and the measured noise.
The system as in any prior embodiment, wherein the minimum value (Eβ,min) of the normalized equality condition is based on:
wherein βmin is a selected minimum value of the regularization factor and n is the measured noise, and the maximum value (Eβ,max) of the normalized equality condition is based on:
wherein βmax is a selected maximum value of the regularization factor.
The teachings of the present disclosure may be used in a variety of well operations. These operations may involve using one or more treatment agents to treat a formation, the fluids resident in a formation, a wellbore, and/or equipment in the wellbore, such as production tubing. The treatment agents may be in the form of liquids, gases, solids, semi-solids, and mixtures thereof. Illustrative treatment agents include, but are not limited to, fracturing fluids, acids, steam, water, brine, anti-corrosion agents, cement, permeability modifiers, drilling muds, emulsifiers, demulsifiers, tracers, flow improvers etc. Illustrative well operations include, but are not limited to, hydraulic fracturing, stimulation, tracer injection, cleaning, acidizing, steam injection, water flooding, cementing, etc.
While the invention has been described with reference to an exemplary embodiment or embodiments, it will be understood by those skilled in the art that various changes may be made and equivalents may be substituted for elements thereof without departing from the scope of the invention. In addition, many modifications may be made to adapt a particular situation or material to the teachings of the invention without departing from the essential scope thereof. Therefore, it is intended that the invention not be limited to the particular embodiment disclosed as the best mode contemplated for carrying out this invention, but that the invention will include all embodiments falling within the scope of the claims. Also, in the drawings and the description, there have been disclosed exemplary embodiments of the invention and, although specific terms may have been employed, they are unless otherwise stated used in a generic and descriptive sense only and not for purposes of limitation, the scope of the invention therefore not being so limited.
| Number | Date | Country | |
|---|---|---|---|
| 62831002 | Apr 2019 | US |
