Download
1 / 33
330 likes | 516 Views
SummaryIntroduction to GeomechanicsKey ingredients in Geomechanics todayFinal remarksFuture workReferencesContact Information. Outline. Summary. We are about to discuss some key ingredients to address geomechanical problems from both near borehole and reservoir level points of views. We're going to explain the geometry description, mesh generation, and also how to solve the governing equations with the Mortar FEM Method by using an Object Oriented approach based on Domain Decomposition. .34213
E N D
2. Summary
Introduction to Geomechanics
Key ingredients in Geomechanics today
Final remarks
Future work
References
Contact Information Outline
3. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
4. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
5. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
6. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
7. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
9. References: pore pressure Bowers, G. L., 1994, Pore Pressure Estimation From Velocity Data: Accounting For Overpressure Mechanisms Besides Undercompaction, IADC/SPE Drilling Conference, Dallas - USA, febrero 15 -18, SPE 27488, pp. 515 – 530 .
Bowers, Glenn. 2002. Detecting high overpressure, The Leading Edge. USA, febrero, pp. 174-177.
C.M, Sayers y G.M, Jhonson., 2000, Predill Pore Pressure Prediction Using Seismic Data. IADC/SPE Drilling Conference. New Orleans, Louisiana, 23-25, February, n° 59122 .
Eaton, Ben A., 1975, The Equation For Geopressure Prediction From Well Logs, SPE 50th Annual Fall Meeting, Society Petroleum Engineering (SPE), Dallas-USA, September 28 – October 1, SPE 5544 .
Economides, J. Michael., Watters, T Larry., Norman, Shari., 1988, Petroleum well construction. Jhon Wiley & Sons, New York- USA, pp. 91-173 .
Li Qiuguo Helliot., 2000, Abnormal Pressure Detection and Wellbore Stability Evaluation of East Sichuan, China. IADC/SPE Drilling Conference. Louisiana, 23-25. February, SPE 59125 .
Dutta, D. C., 2002, Geopressure Detection Using Reflection Seismic Data and Rock Physics Principles: Methodology and Case Histories From Deepwater Tertiary Clastics Basins, Society Petroleum Engineering (SPE) Asia Pacific Oil and Gas Conference and Exhibition held in Melbourne, Australia, 8-10 October .
10. References: unstructured meshes Thompson, J. F., Soni, B. K., Weatherill, N. P., 1999, Handbook of Grid Generation, CRC Pres, Boca Raton, pp. (II-1)-(II-4), pp. (14-1)-(14-22), pp. (16-1)-(16-11), pp. (19-1)-(19-20)
Baker, T. J., 1994, “Triangulations, mesh generation and point placement strategies”, Frontiers of Computational Fluid Dynamics, Caughey, D.A. and Hafez, M. M. (Eds.), John Wiley and Sons, pp. 101 .
George, P. L., Hecht, F. and Saltel, E., 1988, “Constraint of the boundary and automatic mesh generation”, Numerical Grid Generation in Computational Fluid Mechanics ‘88, Sengupta, S., Hauser, J., Eiseman, P. R. and Thompson, J. F. (Eds.), Pineridge Press Limited, U.K., pp. 589-597 .
Mavriplis, D., 1993, “An advancing front Delaunay triangulation algorithm designed for robustness”, AIAA Paper 93-0671 .
Lo, S. H., 1985, “A new mesh generation scheme for arbitrary planar domains”, Int. J. Num. Meth. End., pp. 1403-1426 .
Hermeline, F., 1982, “TRIANGULATION AUTOMATIQUE D'UN POLYÉDRE EN DIMENSION N”, R.A.I.R.O. Analyse numérique, volume 16, n? 3, pp. 211-242 .
Babeau, J. L., 1991, “Constrained Delaunay Triangulations with Local Mesh Size Control”, preprint, personal communication.
Flórez, H. A. and Manzanilla R., 2001, “Automatic Unstructured Mesh Generator for Arbitrary Planar Domains with a Boundary Description Based on B-Spline Curves”, Presented in the ASME International 2001 DETC Conference, Pittsburgh Pennsylvania
11. References: poroelasticity Liu R., 2004, “Discontinuous Galerkin Finite Element Solution for Poromechanics”, PhD thesis, The University of Texas at Austin .
Gai X., 2004, “A Coupled Geomechanics and Reservoir Flow Model on Parallel Computers”, PhD thesis, The University of Texas at Austin .
Han G. et al., 2002, “Semi-Analytical Solutions for the Effect of Well Shut Down on Rock Stability”, Canadian International Petroleum Conference, Calgary, Alberta .
Chen Z, et al, 2006, “Computational Methods for Multiphase Flows in Porous Media, SIAM, pp. 57; 247-258 .
Du J, and Olson J., 2001, “A poroelastic reservoir model for predicting subsidence and mapping subsurface pressure fronts”, Journal of Petroleum Technology & Science, Vol. 30, pp. 181-197.
Grandi, S. and Nafi M., 2001, “Geomechanical Modeling of In-situ Stresses around a Borehole”, MIT, Cambridge, MA.
Charlez A., 1999, “The concept of Mud Window Applied to Complex Drilling”, SPE Paper 56758 .
12. References: Geometry Walstrom, J. E., et al., 1972, “A comparison of various Directional survey Models And Approach To Model Error Analysis”, SPE 3379.
Gfrerrer, A. and Glaser, G.P., 2000, “A New Approach for Most Realistic Wellpath Computation”, SPE 62726.
Flórez, H. A., Rojas, L. R. and Kenyery, F., 2005, “Generation of Airfoil Geometries and CFD Structured Meshing using Bézier Curves”, Submitted in the International Journal of Fluid Dynamics.
Flórez, H. A., 2001, “A NEW METHOD FOR BUILDING B-SPLINE CURVES AND ITS APPLICATION TO GEOMETRY DESIGN AND STRUCTURED GRID GENERATION”, Presented in the ASME International 2001 DETC Conference, Pittsburgh Pennsylvania.
Flórez, H. A. and Manzanilla R., 2001, “AUTOMATIC UNSTRUCTURED MESH GENERATOR FOR ARBITRARY PLANAR DOMAINS WITH A BOUNDARY DESCRIPTION BASED ON B-SPLINE CURVES”, Presented in the ASME International 2001 DETC Conference, Pittsburgh Pennsylvania.
Thompson, J. F., Soni, B. K., Weatherill, N. P., 1999, Handbook of Grid Generation, CRC Pres, Boca Raton, pp. (28-1)-(28-14).
Farin, G., 1993, CURVES AND SURFACES FOR COMPUTER-AIDED GEOMETRIC DESIGN A Practical Guide, Fourth Edition, Academic Press, San Diego, pp. 96-138.
18. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
19. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
20. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
21. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
22. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
23. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
24. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
25. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
28. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
29. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
30. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
31. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
32. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
33. Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
Hector Klie 10/18/2003
So, this is what we can gather from these E&P problems:
Data comes with noise (equipment calibration, subjective observations, extrapolation, etc)
The problem has infinite solutions. They are generally underdetermined. However, notice that regressions are basically overdetermined problems.
They are unstable: little perturbations in the data generate big changes in the solution. Infinite and/or unstable solutions lead to ill/posed problems.
Besides they are nonlinear. These are the 3 ingredients for uncertainty.
And as it wasn't already enough, the problems are computationally expensive to solve (even if the uncertainty are within low bounds. As Luca Consentino says: ....
