Spatial Constraints for the Stochastic Modeling of Fault Networks in the Presence of Large Structural Uncertainties

in: 75th EAGE Conference and Exhibition incorporating SPE EUROPEC 2013

Abstract

Faults greatly impact the heterogeneities, the fluid flow and the geomechanical behavior of hydrocarbon reservoirs. However, significant uncertainty affects fault location, extent and connectivity due to limited quality and lack of subsurface data. In contrast to the standard practice of reservoir modeling which tends to represent only one deterministic fault network, we propose to sample structural uncertainty about fault networks (both geometrical and topological) by simulating a set of possible models all constrained by subsurface data. The conditioning of the faults may be done from fault sticks interpreted from 2D or 3D seismic surveys. Fault sticks correspond to one interpretation and must be used carefully especially when the uncertainties on the fault position are high. The objective of this paper is to introduce a new method to constrain the stochastic simulation of seismic-scale faults by negative constraints which condition the non-occurrence of fault. We propose to divide the simulation space into two subspaces: (1) a volume where the simulation of faults is allowed and (2) a volume where the occurrence of faults is excluded. These two volumes are separated by a closed surface, named envelope, used to constrain the position of faults. During stochastic fault simulation, we optimize the geometrical fault parameters (position of the center, dip, strike) so as to minimize a misfit function. The efficiency of this conditioning approach is tested on synthetic data sets. The described simulation method is applied in the cases of large-scale and high uncertainties which can be characterized by the envelope delimiting the exclusion zones. The algorithm simulates faults which honor both statistical input parameters and exclusion constraints. The high uncertainties lead to the simulation of various fault networks with different geometries and topologies. In the petroleum industry, the stochastic approach is widely used to manage the facies and petrophysical property uncertainties, but it is rarely applied to fault network modeling. The presented optimization algorithm belongs to data conditioning methods for stochastic fault simulation, with the distinctive feature to be adapted to dense and continuous data. The fault-exclusion zone is adequate to represent the continuous structural information recorded by the seismic data.

Download / Links

BibTeX Reference

@inproceedings{julio:hal-04068734,
 abstract = {Faults greatly impact the heterogeneities, the fluid flow and the geomechanical behavior of hydrocarbon reservoirs. However, significant uncertainty affects fault location, extent and connectivity due to limited quality and lack of subsurface data. In contrast to the standard practice of reservoir modeling which tends to represent only one deterministic fault network, we propose to sample structural uncertainty about fault networks (both geometrical and topological) by simulating a set of possible models all constrained by subsurface data. The conditioning of the faults may be done from fault sticks interpreted from 2D or 3D seismic surveys. Fault sticks correspond to one interpretation and must be used carefully especially when the uncertainties on the fault position are high. The objective of this paper is to introduce a new method to constrain the stochastic simulation of seismic-scale faults by negative constraints which condition the non-occurrence of fault. We propose to divide the simulation space into two subspaces: (1) a volume where the simulation of faults is allowed and (2) a volume where the occurrence of faults is excluded. These two volumes are separated by a closed surface, named envelope, used to constrain the position of faults. During stochastic fault simulation, we optimize the geometrical fault parameters (position of the center, dip, strike) so as to minimize a misfit function. The efficiency of this conditioning approach is tested on synthetic data sets. The described simulation method is applied in the cases of large-scale and high uncertainties which can be characterized by the envelope delimiting the exclusion zones. The algorithm simulates faults which honor both statistical input parameters and exclusion constraints. The high uncertainties lead to the simulation of various fault networks with different geometries and topologies. In the petroleum industry, the stochastic approach is widely used to manage the facies and petrophysical property uncertainties, but it is rarely applied to fault network modeling. The presented optimization algorithm belongs to data conditioning methods for stochastic fault simulation, with the distinctive feature to be adapted to dense and continuous data. The fault-exclusion zone is adequate to represent the continuous structural information recorded by the seismic data.},
 address = {London, United Kingdom},
 author = {Julio, Charline and Caumon, Guillaume},
 booktitle = {{75th EAGE Conference and Exhibition incorporating SPE EUROPEC 2013}},
 doi = {10.3997/2214-4609.20130865},
 hal_id = {hal-04068734},
 hal_version = {v1},
 month = {June},
 title = {{Spatial Constraints for the Stochastic Modeling of Fault Networks in the Presence of Large Structural Uncertainties}},
 url = {https://hal.univ-lorraine.fr/hal-04068734},
 year = {2013}
}