METHOD AND SYSTEM TO INVERT TECTONIC BOUNDARY OR ROCK MASS FIELD IN IN-SITU STRESS COMPUTATION

Abstract

A method, system and computer program product for inverting boundary conditions for rock mass field in-situ stress computation for a geologic structure are disclosed. According to an embodiment, the current invention includes a method for inverting boundary conditions for rock mass field in-situ stress computation for a geologic structure, the method comprising: considering physical constraint of the geological structure; deriving and solving normal equations of penalized weighted least squares, said normal equations of penalized weighted least squares partially representing said physical constraint of the geological structure; and outputting boundary conditions and reproducing the rock mass field in-situ stress based on the result of said solved normal equations of penalized weighted least squares.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

The present invention is illustrated by way of example and not intended to be limited by the figures of the accompanying drawings in which like references indicate similar elements and in which:

FIG. 1 is a flowchart showing steps associated with the present method, apparatus, and article of manufacture;

FIG. 2 is a diagram showing one embodiment of the invention to simulate the tectonic effect on rock mass field by boundary loading;

FIG. 3 is a diagram showing computed Minimum horizontal stress contour of Appalachian Plateau as one example of the present method; and

FIG. 4 is a diagram showing the comparison of measured minimum horizontal stress (represented by dots) and the inverted minimum horizontal stress using the present method (represented by arrays) at the three wells in Appalachian Plateau.

DETAILED DESCRIPTION OF THE INVENTION

This invention develops a new technique to invert the boundary conditions for rock mass field in-situ stress computation. The proposed inverting procedure is based on the physical assumption that the in-situ stress can be decomposed into gravitational, tectonic, thermal and residual components, where the tectonic component can be simulated by applying the appropriate horizontal stress or displacement at the far-field boundary by using numerical tools, such as Finite Element Method (FEM) and Finite Difference Method (FDM). The Penalized Weighted Least Squares Method (PWLS), an optimization algorithm that includes both mathematical and physical fittings of the model prediction with the source data input for inversion, is developed to invert the tectonic boundary conditions and then to reproduce the rock mass field in-situ stress.

The optimization algorithm includes two parts: Weighted Least Squares Method, adopted to obtain a mathematical best-fit by taking into account the source data uncertainties, and Penalty Function, designed to ensure a geomechanical reasonable solution in complex tectonic geological structure. Corresponding iteration algorithm is also incorporated to address the nonlinear behavior of rock mass field.

FIG. 1 is a schematic flow chart of the steps of the system and method of this invention. According to the method and system of this invention, one needs to build a numerical model (grids) from a geological model, as shown as step 10 in FIG. 1. In one embodiment of the invention, when one is building a numerical model of domain under consideration, the geologic information, such as rock mass layers, faults and so on, should be included in the model. For example, the grid should be shaped according the geologic structure, and the mechanical properties of each element distinct by the grids should be assigned from lithology information. In one embodiment of the invention, in some simple cases, the numerical model can be created by the program itself. In another embodiment of the invention, the numerical model can be created by other reprocessing program.

Once the numerical model has been built from a geological model, the source data should be inputted as necessary information for further boundary condition inversion and in-situ stress computation. The receive source data step is shown as step 20 in FIG. 1. The set of source data typically includes stress related information measured from boreholes or from the field, and the inherent uncertainties of this information. Both direct stress measurement from borehole and indirect stress information from borehole (such as borehole breakout and drilling induced fractures, etc) can be input as source data for inversion. In one embodiment of the invention, source data includes data generated from MDT tools. In another embodiment of the invention, source data includes data measured from borehole using caliper and Fullbore Formation MicroImage (FMI).

Once the formation data have been received, then it would be possible to compute the gravitational stress component, which is indicated as step 30 in FIG. 1. In one embodiment of the invention, Finite Element Method (FEM) is applied to compute the gravitational stress component. In another embodiment of the invention, Finite Difference Method (FDM) is applied to compute the gravitational stress component.

Before we go to next step, as a normal approach in geomechanical modeling, we assume that the current in-situ stress can be decomposed into gravitational, tectonic, thermal and residual components, written as:

σ^tot=σ^gra+σ^tec+σ^the+σ^res+ . . .   (1)

where σ^gra, σ^tec, σ^the and σ^res are the gravitational, tectonic, thermal and residual stress components respectively. The gravitational component can be computed by gravitational loading. In general, thermal and residual components of the stress are less important as gravitational and tectonic components, and can be neglected or simplified in modeling, as disclosed by S. D. Mckinnon. See, S. D. MCKINNON, Analysis of Stress Measurements Using A Numerical Model Methodology, International Journal of Rock Mechanics & Mining Sciences, 12 Jul. 2001, Page 699-709. Therefore, the current in-situ stress can be simplified as:

σ^tot≈σ^gra+σ^tec  (2)

To calculate the current in-situ stress, it is needed to solve the tectonic stress component, which is indicated as step 40 in FIG. 1.

The tectonic stress component can be modeled by applying a set of appropriate horizontal stresses or displacements at the far-field boundary. For example, to simulate the tectonic effect on rock mass field by boundary loading, the complicated tectonic process can be decoupled into a set of simple boundary loading patterns, such as β₁, β₂ and β₃, as shown in FIG. 2. The set of tectonic effects applied to the system boundary have the unknown parameters β_i(i=1, 2, . . . , p) that need to be inverted. By using this pattern of tectonic boundary, the system will have the flexibility of modeling the complicated tectonic process, such as overburden deposition, repeated cycles of elevation and discontinued tectonic effect that occurred during the sedimentary history of the geological system, and gives reliable estimation of tectonic boundary.

In one embodiment of the invention, boundary loading patterns β₁, β₂, . . . , β_p are displacements that applied to the far-field boundary of the system to simulate the tectonic effects occurred during the sedimentary history of the geological system. In another embodiment of the invention, β₁, β₂, . . . , β_p can be the normal stress that applied to the far-field boundary of the system to simulate the tectonic effect.

In one embodiment of the invention, from linear elastic theory, the tectonic induced stress of the rock mass field can be written as:

$\begin{array}{cc}{\sigma }^{\mathrm{tec}}=\sum _{i=1}^p{{\beta }_i\left({\sigma }^{\mathrm{tec}0}\right)}_i& \left(3\right)\end{array}$

where (σ^tec0)_i is the tectonic stress induced by the i-th unit tectonic effect (β_i=1), which can be computed from FEM, FDM or other numerical tools. Here σ^tec is the sum of tectonic stress that induced by the tectonic history.

Assuming we have n-points of stress (stress element) measurements (σ^mea_i), j=1, 2, . . . n, the objective of our work is to invert parameters β_i in the following function:

$\begin{array}{cc}Y=X\beta +\delta Y=\left\{{\sigma }_j^{\mathrm{mea}}-{\sigma }_j^{\mathrm{gra}}\right\},X={\left({\sigma }^{\mathrm{tec}0}\right)}_{\mathrm{ij}},\beta =\left\{{\beta }_i\right\};\delta =\left\{{\delta }_j\right\}\begin{array}{cc}i=1,2,\dots p,& j=1,2,\dots n\end{array}& \left(4\right)\end{array}$

where (σ^tec0)_i, is the tectonic stress induced by i-th unit tectonic effect (β_i=1) at j-th measurement point and δ_i is its error. We assume that Var[δ]=v²V, where v²V is a n×n positive-definite matrix containing the uncertainty of measured in-situ stress. With taking into account of the uncertainty of measured data and neglecting thermal and residual effects, the objective of our work is to invert parameters β_i by minimizing the objective function:

ƒ(β)=δ′V⁻¹δ  (5)

The inversion procedure can be carried out mathematically by using Weighted Least Squares Method. But the problem with this procedure is the solution may show physically unreasonable results, especially for the case of complex geological structure. This is because the above optimization function is to minimize the error between the model and measurement in few points where the source data for inversion are located without consideration of the whole numerical model. In other words, it will produce the mathematically correct result at measurement points, but physically unreasonable result in the whole interested domain. A typical example of this problem can be explained as: Assuming in a structure that two sets tectonic boundaries (β₁ and β₂) must be applied to simulate the tectonic effect, and the stress measurements to invert β₁ and β₂ are σ^mea_i, j=1, 2, . . . , n>=3, then probably we will get the inverted solution of β₁ and β₂ that are two large numbers contrasted to each other, with their total contribution to tectonic stress, σ^tec=β₁σ^tec0₁+β₂σtec0₂ very close to the input measured data at the area where the input data are located (mathematically fitting), but from the physics this kinds of solution is not feasible.

To overcome this problem, physical consideration or constraint should be added to the optimization function. More generally, we hereby derive a new objective function of weighted least squares by adding a penalty term:

ƒ(β)=δ′V⁻¹δ+ω·g(β)  (6)

where δV⁻¹δ is still the sums of weighted square errors, g(β) is the penalty function including the constraints on boundary stress or displacements and ω is the penalty parameter to control the contribution of penalty function. This penalty function is used to prevent the solution from leaving physical feasible region during the optimization process. By solving the above normal equations of penalized weighted least squares, which is indicated as step 50 in FIG. 1, we can invert tectonic stress component while considering the physical constraint of the geological system.

Thus, from ∂ƒ(β)/∂β=0 we obtain the penalized weighted least squares estimation:

β=(X′V⁻¹X+D1)⁻¹(X′V⁻¹Y=D2)  (7)

where D1 and D2 are two matrixes containing elements of different parts of derivative of g(β) with respect to β.

Based on the general idea of penalty function, we now discuss how to select the penalty function and penalty parameter. The key idea of this penalty function is to add physical constraints to the optimization function, and the penalty parameter ω controls the contribution of the penalty function to the total objective function ƒ(β).

Depending on the geological structure and prior-information of a practical problem, the following three kinds of penalty functions can be used:

• • For relatively uniform geological structure, or complex structure that no other information about tectonic effect is available, the elastic energy applied to rock mass field by tectonic effects is used to define the penalty function:

$\begin{array}{cc}g\left(\beta \right)=\sum _{i=1}^p{\left(\frac{\sqrt{W_i}}{\sqrt{W_1}}-1\right)}^2& \left(8\right)\end{array}$

where W_i is the elastic energy applied by the tectonic effect β_i. Adding this function will control the solution from having big contrasted values, as discussed in the previously example.

• • For the cases that some prior information about tectonic effects can be guessed from geological information, such as faults slipping, the following penalty function can be used:

$\begin{array}{cc}g\left(\beta \right)=\sum _{i=0}^l{\left(\frac{{\gamma }_i}{{\Delta }_i}-1\right)}^2& \left(9\right)\end{array}$

where γ_i is the tectonic boundary displacement at a certain depth, that can be computed from β_i and its depth, Δ_i is the corresponding displacement guessed from prior information, and l is the number of guessed Δ_i

• • For the cases where the range of tectonic displacement at boundary could be given as prior-information, the following penalty function can be used:

$\begin{array}{cc}g\left(\beta \right)=\sum _{i=0}^l{\left(\frac{{\gamma }_{i,\max }}{{\gamma }_{i,\max }-{\gamma }_i}+\frac{{\gamma }_{i,\min }}{{\gamma }_i-{\gamma }_{i,\min }}\right)}^2& \left(10\right)\end{array}$

where γ_γi,min and γ_i,max are ranges of tectonic displacement at boundary at a certain depth.

Using these penalty functions, the selection of penalty parameter is very flexible and should be decided according to each actual application. The penalty parameter can also be solved in an iterative form as following equation.

ƒ(β^k)=(δ′V⁻¹δ)^k+α(δ′V⁻¹δ)^k-1g(δ^k)  (11)

where ε′V⁻¹ε is the sum of squares errors, and α is a small number to govern the amount of contribution from the penalty functions. The superscript k refers to the k-th iteration. In the iteration process to minimize the objective function, the penalty parameter α(δ′V⁻¹δ)^k-1 changes in each step to keep the percentage of contribution from penalty function at almost a same level.

It should be mentioned that the proposed inversion model and its weighted least squares estimation are based on the linear elastic theory and superposition rule of stress components. But in practical applications, nonlinear behaviors always appear in the domain of interest, and thus in the numerical model an iteration algorithm can be introduced to deal with this nonlinear behavior, which is indicated as step 70 in FIG. 1.

Once the normal equations (7) of penalized weighted least squares are solved, the results can be used to compare with the end criterion to indicate accuracy of the result, which is indicated as step 60 in FIG. 1. In one embodiment of the invention, the end criterion is

$\begin{array}{cc}{\hspace{0.17em}}^kR=\sqrt{\sum _j^n{\left(\xi _j-{\sigma }_j^{\mathrm{tec}-\mathrm{mea}}\right)}^2/\sum _j^n{\left({\sigma }_j^{\mathrm{tec}-\mathrm{mea}}\right)}^2}\le \varepsilon & \left(12\right)\end{array}$

where ^kξ_j is the computed tectonic stress of k-th iteration at j-th measurement point, σ_j^tec-mea is “measured tectonic stress” at j-th measurement point which can be obtained by σ_j^tec-mea=σ_j^mea−σ_j^gra, ^kR is total relative error of k-th iteration and ε is the specified threshold. Here the k-th iteration can be the iteration on nonlinear model behavior or the iteration on penalty function.

If the end criterion has not been reached, in one embodiment of the invention, it is necessary to update for iteration on nonlinear behavior of the numeric model and return to step 40, which is indicated as step 70 in FIG. 1. In another embodiment of the invention, if the end criterion has not been reached, it is necessary to update for iteration on penalty function and return to step 40, which is indicated as step 80 in FIG. 1. Still in another embodiment of the invention, if the end criterion has not been reached, it is necessary to update for iteration on nonlinear behavior of the numeric model (where an inner iteration is used) and to update for iteration on penalty function (where an outer iteration is used), which is indicated as step 70 to step 80 (or vice versa) then return to step 40.

If the end criterion has been reached, either after one or more times process of steps 70 or 80 or without processing steps 70 or 80, output boundary conditions and compute and output final stress field of the geologic structure, which is indicated as step 90 in FIG. 1.

Specifically, in one embodiment of the invention, g(β), the penalty function including the constraints on boundary stress or displacements, can be represented by either of equations (8)-(10) depending on the geological structure and prior-information of a practical problem. ε, the penalty parameter to control the contribution of penalty function, can be represented by α(δ′V⁻¹δ)^k-1 by solving equation (11).

Once g(β) and ε are decided, equation (7) can be derived and consequently equation (3) and equation (2) can also be solved. Therefore, by assuming that the in-situ stress can be decomposed into gravitational and tectonic components, we derived the Penalized Weighted Least Squares Method (PWLS) that includes both mathematical and physical fittings of the model prediction with the stress measurements and/or other field information, to invert the tectonic boundary conditions and then to reproduce the rock mass field in-situ stress. By applying the PWLS, the uncertainties of stress measurements and geological information can be considered. In addition, this model can be used in more general complex geological conditions and nonlinear properties of rock mass can be modeled.

To demonstrate the usability of the invention, we now provide some experimental results of applying the current invention to invert boundary conditions and compute the in-situ stress for Appalachian Plateau field. The data we used for this study were quoted from Evans and Plumb et al (Geophysical Log Responses and Their Correlation With “Bed-to-Bed Stress Contrasts in Paleozoic Rocks, Appalachian Plateau, New York) and Evans et al (Appalachian Stress Study 1. A Detailed Description of In Situ Stress Variations in Devonian Shales of the Appalachian Plateau). The plateau is an allochthonous sheet detached within the Silurian Salts below the Devonian section. A wealth of measured stress data is available in three wells that penetrating the formation rock mass to around 1000 meters depth: Wilkins, Appleton and O'Dell.

A 2D model is built on the plane that covered the three wells. The geological layers and formation rock mechanical properties are evaluated based on geological information and wireline data. Using the minimum horizontal stress measured at Wilkins well as input source data for inversion, the tectonic boundary of the rock mass field is inverted and the in-situ stress field computed from the inverted tectonic boundary. The resulted minimum horizontal stress field is shown in FIG. 3.

Measured stress at well Appleton and O'Dell is used to validate this invention, as shown in FIG. 4. The computed minimum horizontal stress inverted from this invention (represented by arrays) and the measured minimum horizontal stress (represented by dots) at well Appleton and O'Dell is displayed in the same plot for comparison. The plot shows a good match between the inverted results and the measurement.

Therefore, the current invention takes into account errors associated with different measurements and information. The current invention also introduces a penalty function to traditional weighted least squares to constrain the inversion parameters, thus keeps the solution from leaving the feasible region. The result of application of the current invention thus becomes fairly good mathematically as well as physically meaningful.

The flowchart in FIG. 1 illustrates the architecture, functionality, and operation of possible implementations of systems and methods according to various embodiments of the present invention. In this regard, each block in the flowchart may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It will also be noted that each block of the flowchart illustration, and combinations of blocks in the flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

While the disclosure has been particularly shown and described with reference to exemplary embodiments thereof, it will be understood by those of ordinary skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the claims. In addition, those of ordinary skill in the art appreciate that any arrangement which is calculated to achieve the same purpose may be substituted for the specific embodiments shown and that the invention has other applications in other environments. Therefore, it is intended that the scope of the invention be defined by the accompanying claims and their equivalents.

Claims

1. A method to invert boundary conditions for rock mass field in-situ stress computation for a geologic structure comprising: a. building a numerical model from a geological model of the geologic structure;b. inputting source data measured from the geologic structure into the numerical model;c. computing gravitational stress of the geologic structure by gravitational loading;d. computing at least one tectonic stress induced by unit tectonic effect;e. solving normal equations of penalized weighted least squares;f. determining whether the end criterion has been reached and if not return to step d, and if the end criterion has been reached, proceed to step g; andg. outputting boundary conditions and compute and output final in-situ stress field of the geologic structure.

2. The method as in claim 1 wherein the source data includes stress related measurement and information of the geologic structure and their uncertainties.

3. The method as in claim 1 wherein the source data includes data generated from MDT tools.

4. The method as in claim 1 wherein the source data includes data measured from borehole using caliper and Fullbore Formation MicroImage (FMI).

5. The method as in claim 1 wherein the penalized weighted least squares include Weighted Least Squares Method representing uncertainty of the measured data and Penalty Function representing geomechanical feature of the geological structure.

6. The method as in claim 1 wherein the penalized weighted Least squares method is represented by: ƒ(β)=δ′V−1δ+ω·g(β) where δ′V−1δ represents the sums of weighted square errors, g(β) represents the penalty function, and ω is the penalty parameter.

7. The method as in claim 6 further comprising selecting the penalty function and penalty parameter.

8. The method as in claim 6 further comprising selecting the penalty function by applying the elastic energy.

9. The method as in claim 6 further comprising selecting the penalty function by applying tectonic boundary displacement and the corresponding displacement.

10. The method as in claim 6 further comprising selecting the penalty function by applying range of tectonic displacement at boundary at a certain depth.

11. The method as in claim 1 wherein step f further includes updating for iteration on nonlinear behavior of the numeric model.

12. The method as in claim 6 wherein step f further includes updating for iteration on penalty function.

13. The method as in claim 1 further comprising assuming that the in-situ stress can be decomposed into gravitational, tectonic, thermal and residual components.

14. The method as in claim 13 further comprising assuming that thermal and residual components can be neglected so that the in-situ stress can be decomposed into gravitational and tectonic components.

15. The method as in claim 13 further comprising simulating the tectonic component by applying the appropriate horizontal stress or displacement at the far-field boundary by using numerical tools.

16. The method as in claim 1 wherein step d is performed using Finite Element Method.

17. The method as in claim 1 wherein step d is performed using Finite Difference Method.

18. The method as in claim 1 wherein step g is performed using Finite Element Method.

19. The method as in claim 1 wherein step g is performed using Finite Difference Method.

20. A method to invert boundary conditions for rock mass field in-situ stress computation for a geologic structure comprising: a. considering physical constraint of the geological structure;b. deriving and solving normal equations of penalized weighted least squares, said normal equations of penalized weighted least squares partially representing said physical constraint of the geological structure; andc. outputting boundary conditions and reproducing the rock mass field in-situ stress based on the result of said solved normal equations of penalized weighted least squares.

21. An article of manufacture, comprising: a computer usable medium having a computer readable program code means embodied therein for inverting boundary conditions for rock mass field in-situ stress computation for a geologic structure, the computer readable program code means in said article of manufacture comprising: a. computer-readable program means for building a numerical model from a geological model of the geologic structure;b. computer-readable program means for inputting source data measured from the geologic structure into the numerical model;c. computer-readable program means for computing gravitational stress of the geologic structure by gravitational loading;d. computer-readable program means for computing at least one tectonic stress induced by unit tectonic effect;e. computer-readable program means for solving normal equations of penalized weighted least squares;f. computer-readable program means for determining whether the end criterion has been reached and if not return to step d, and if the end criterion has been reached, proceed to step g; andg. computer-readable program means for outputting boundary conditions and compute and output final in-situ stress field of the geologic structure.

22. An article of manufacture, comprising: a computer usable medium having a computer readable program code means embodied therein for inverting boundary conditions for rock mass field in-situ stress computation for a geologic structure, the computer readable program code means in said article of manufacture comprising: a. computer-readable program means for considering physical constraint of the geological structure;b. computer-readable program means for deriving and solving normal equations of penalized weighted least squares, said normal equations of penalized weighted least squares partially representing said physical constraint of the geological structure; andc. computer-readable program means for outputting boundary conditions and reproducing the rock mass field in-situ stress based on the result of said solved normal equations of penalized weighted least squares.