Full text loading...
- E-ISSN: 1365-2478
-
Modelling the seismic attributes of overpressured siliciclastic rocks, with a genetic annealing (GAN) algorithm
- Source: Geophysical Prospecting, Volume 50, Issue 2, Nov 2002, p. 169 - 182
-
- 23 Nov 2002
Abstract
ABSTRACT A genetic annealing (GAN) algorithm is used to derive an empirical model which predicts compressional‐wave velocity values for overpressured siliciclastic rocks. The algorithm involves non‐linear random searching and mutation techniques and its annealing component imposes a very strict control over the rate of convergence of the search. This technique provides an alternative to the standard calculations involving the effective stress coefficient (n). The pore pressure is introduced into the model as an explicit variable and as part of an overpressure coefficient, (Pp/Pc) − the ratio of pore to confining pressure. Empirical model‐derived data and known laboratory data are compared and their differences are shown to be within statistically acceptable error limits. The empirical equation fits all under‐ and overpressured data simultaneously, irrespective of pore fluid pressure level, with the same parameters. It is used to predict seismic velocities very accurately for extreme levels of overpressure, starting from normally pressured experimental data. The model highlights the effect of pore pressure on the compressional‐wave velocity of fully saturated samples with different clay contents. It can be used when the experimental data available are sparse and particularly when a prediction of material behaviour is necessary at specific pore fluid pressure and depth conditions.