Geologia

Solo disponible en BuenasTareas
  • Páginas : 5 (1139 palabras )
  • Descarga(s) : 0
  • Publicado : 8 de junio de 2009
Leer documento completo
Vista previa del texto
AAPG International Conference: October 24-27, 2004; Cancun, Mexico

Sismage and the 3D Visualization at Total
Sébastien Guillon, Naamen Keskes Total, CSTJ Av Laribau 64000 PAU FRANCE sebastien.guillon@total.com, naamen.keskes@total.com As any other seismic interpretation platform, Sismage® offers a 3D visualisation environment which allows the display all the available data needed for theinterpretation : seismic and attributes bloc (with or without volume rendering), horizons, faults, bodies, wells, satellite images, etc … All picking operations (manual or automatic) can obviously be done in this 3D View. The innovation offered by Sismage® concerns faults picking. Indeed we provide two new methodologies of picking called “Fault Peeling” and “Fault Tracker”. These two new techniquesallows to build very quickly a 3D triangulated faults network that can be exported directly to the geomodel. 1. The “Fault Peeling” methodology The “Fault Peeling” consists in converting a fault attribute block into a set of small independent fault planes in 3D (extracted along maximum values of the fault attribute) and the neighbouring relation between them (3D connectivity of these small planes).Then the process starts by posting seeds on one or several locations : each seed corresponds to an elementary fault plane. Semi-automatic picking can be done by merging all these elementary planes. Automatic propagation of these elementary planes can also be performed according to the neighbouring relations extracted before and to some 3D geometric properties such as planarity dip and azimuth. 1.1Fault Attribute As “Fault Peeling” has been developed using seismic attributes as input to the fault propagation, the first step consists in computing a seismic « fault » attribute that enhances faults structural directions. Up to now all available fault detection seismic attributes (amplitude, semblance, coherency) are often very sensitive to the sedimentary information: it introduces someorganised and chaotic residue which corrupts the attribute and overlays the true fault structural trends (fig 1). To extract noise-free structural trends, Sismage® computes a new fault attribute based on a statistical analysis of discontinuities in seismic images. It computes a confidence measurement of the seismic horizon continuity (which is low in case of fault presence). As faults could generallybe considered as a set of “pseudo vertical” discontinuities, the statistical analysis is performed in a set of “pseudo verticals” windows in order to enhance faults detection and then to remove sedimentary information which is not vertically consistent. This method allows to underline with greater details the faults pattern from field scale to reservoir scale including relays and segmented areas.The geological noise is greatly reduced with this new attribute (fig 1).

Fig 1 : Comparison between classical coherency and new fault attributes

Copyright © 2004, AAPG

2/3

AAPG International Conference: October 24-27, 2004; Cancun, Mexico

1.2 Fault Peeling Depending on the quality of the fault attribute, one can either manually interprete the fault cube or perform a second step,which consists in propagating the faults planes from this fault attribute : fault peeling technology(Sismage®). The peeling technology consists in converting the fault attribute into a set of small independent fault planes in 3D (extracted along maximum values of the fault attributes) and the neighbouring relation between them (3D connectivity of these small planes). Then the process is started byposting seeds on one or several locations in the fault plane in inline, xline, random line and/or timeslice: each seed corresponds to an elementary fault plane. Manual picking can be done by merging all these elementary planes, and/or automatic propagation of these elementary planes can also be performed according to the neighbouring relations extracted before and to some geometric properties such...
tracking img