| US 7,495,436 B2 | ||
| Rock properties prediction, categorization, and recognition from NMR echo-trains using linear and nonlinear regression | ||
| Mouin Hamdan, Pieta (Malta); and Holger Thern, Hannover (Germany) | ||
| Assigned to Baker Hughes Incorporated, Houston, Tex. (US) | ||
| Filed on Mar. 18, 2005, as Appl. No. 11/84,322. | ||
| Claims priority of provisional application 60/554121, filed on Mar. 18, 2004. | ||
| Prior Publication US 2005/0206378 A1, Sep. 22, 2005 | ||
| Int. Cl. G01V 3/00 (2006.01); G01R 33/20 (2006.01) | ||
| U.S. Cl. 324—303 | 21 Claims |
| 1. A method of estimating a value of a property of an earth formation, the method comprising:
(a) conveying a nuclear magnetic resonance (NMR) sensing apparatus into a borehole in the earth formation;
(b) measuring an NMR echo train indicative of the property from the earth formation;
(c) using a predictive model in order to estimate from the measured NMR echo train the value of the property; and
(d) recording the estimated value of the property on a computer readable medium; wherein the predictive model is obtained by a regression in which the dependent variable of the regression comprises a matrix
of amplitudes of a plurality of NMR echo trains obtained from one of: (i) a plurality of NMR echo trains obtained from a plurality
of rock samples, each rock sample having a known value of the property, and (ii) a plurality of simulated NMR echo trains
corresponding to a known value of the property, and in which the independent variable of the regression comprises the known
value of the property.
|