Sperrevik et. al. Empirical estimation of fault rock properties 1 EMPIRICAL ESTIMATION OF FAULT ROCK PROPERTIES Susanne Sperrevik 1^* , Paul A. Gillespie 2 , Quentin J. Fisher 3 , Trond Halvorsen 1 & Rob J. Knipe 3 1) Department of Geology, University of Bergen, Allegt. 41, N- 5007 Bergen, Norway 2) Norsk Hydro Research Centre, N- 5020 Bergen, Norway 3) Rock Deformation Research Group, Earth Sciences Departement, University of Leeds, UK ^ Present adress: Norsk Hydro Research Centre, N- 5020 Bergen, Norway * Corresponding author Susanne.Sperrevik@hydro.com 1 ABSTRACT Faults in clastic sequences are often significant barriers to single-phase fluid flow and can act as absolute barriers to the flow of non-wetting phases over geological time. Knowledge of the fault rock flow properties, as well as the width of the fault zone is required in order to conduct fluid flow simulations in faulted reservoirs. In this paper we present an equation for estimating fault zone thickness from fault throw based on outcrop data from Sinai (Knott et al. 1996) and Northumberland. These data show that the throw/thickness relationship is dependent on lithology, and can be related to the clay content of the fault zone. The permeability and threshold pressures of faultrocks are dependent on factors such as the mineralogical composition of the faulted rock, the effective stress conditions and the time-temperature history of the reservoir prior to, during and following deformation. A strong power law relationship is established between threshold pressure and permeability, which is insensitive to the faulting mechanisms. The permeability and the threshold pressures of both the host rocks and the fault rocks can be represented by functions which are dependent on the clay content and the maximum burial depth (i.e. time-temperature history), whereas for the fault rocks the depth (i.e. effective stress conditions) at the time of deformation also needs to be taken into account. The database from which these empirical relationships were derived contains core measurements from faults and their associated host rocks in siliciclastic sequences from the North Sea. Many types of fault rock are contained within the database (disaggregation zones, cataclastic faults, phyllosilicate-framework faults and clay smears) and these have experienced a wide range in their maximum burial depths (2000m-4500m). In reservoir simulation the sealing effect of the faults can be represented as transmissibility modifiers for each grid cell, calculated from knowledge of fault rock permeability, the width of the fault zone, the grid block permeabilities and the geometry of the simulation grid (Manzocchi et al. 1999). We have applied the technique to a number of North Sea reservoirs, using the new equation for calculating fault rock permeability. However, even if the new equation produced lower permeabilities than previously published relationships, in all cases the transmissibility modifiers generated by this technique proved consistently too high (1-2 orders of magnitude) in order to produce good history matches. In order to further improve the model, and to get better history match, we think that it is important to include capillary effects, relative transmissibility multipliers, the new equation for calculating fault zone width and to better constrain the clay content of the fault zone. However, better methods are still required for capturing complex fault geometries in the reservoir model.