Evolution of dilatant fracture networks in a normal fault — Evidence from 4D model experiments Marc Holland a,b, ⁎, Heijn van Gent a , Loïc Bazalgette c , Najwa Yassir c , Eilard H. Hoogerduijn Strating d , Janos L. Urai a a Structural Geology, Tectonics and Geomechanics, RWTH Aachen University, Lochnerstrasse 4-20, D-52066 Aachen, Germany b Baker Hughes, RDS, Emmerich-Josef Str. 5, D-55116 Mainz, Germany c Shell International Exploration and Production B.V., Kesslerpark 1, 2288 GS Rijswijk ZH, The Netherlands d Shell Exploration & Production, c/o NAM, PO Box 28000, 9400 HH Assen, The Netherlands abstract article info Article history: Received 6 October 2010 Received in revised form 6 February 2011 Accepted 8 February 2011 Available online 8 March 2011 Editor: T.M. Harrison Keywords: fracture network 4D analog experiments fractal Dilatant fractures in normal fault zones are widely recognized as major pathways of fluid flow in the upper crust where the ratio of rock strength and effective stress is suitable for their formation, but the structure of these fracture networks in 3D, their connectivity and their temporal evolution is poorly known. Here we build on 2D studies of scaled models of fracture networks in dilatant normal fault zones, using a series of X-ray computer tomographic scans of a physical model. We show how the dilatant fracture network evolves in 3D, as a complex self-organizing system with self-similar geometry. We processed the CT-scan data using a threshold filter to identify the open fracture volume, to allow visual and quantitative analysis of the evolving fracture system in 3D. Dilatant jogs initiated along the evolving fault plane coalesce into a self-similar percolating volume (Fd = 1.91). The fracture volume increases non-linearly with progressive displacement as the velocity of the fault blocks diverges from the master fault orientation and we infer that the normal stress on the fault decreases correspondingly. This process continues until the system triggers the formation of antithetic faults, with a corresponding increase in normal stress on the master fault and a decrease in the rate of fracture volume creation. We infer that although parameters like the width of the fractures are not scaled with the same ratio as length and stress, the processes and evolution of fracture geometries in our model are robust and apply to a wide range of normal fault zones in nature. Since our physical model does not involve chemical processes such as cementation or fault healing, the experiment suggests that fault systems can show a non-linear change of fracture network properties caused by a geometric evolution only. © 2011 Elsevier B.V. All rights reserved. 1. Introduction Fault zones occur at a wide range of length scales; they are first order mechanical discontinuities in the Earth's crust and form major systems of barriers and/or conduits for fluid flow (e.g. Ben-Zion and Sammis, 2003; Bonnet et al., 2001; Caine et al., 1996; Cowie et al., 1995; Crider and Peacock, 2004; Hallgass et al., 1997; Handy et al., 2007; Petit et al., 1999; Roberts et al., 1991; Sibson, 2000, e.g. Sornette et al., 1990; Townend and Zoback, 2000). In those parts of the upper crust where the compressive strength of rock is high compared to the mean effective stress, deformation in releasing sections of normal faults generates prominent open fractures, thus creating major pathways controlling the fluid flow in the rock mass (Acocella et al., 2003; Clifton and Schlische, 2003; Clifton et al., 2003; Crider and Peacock, 2004; Doubre, 2004; Ferrill and Morris, 2003). Examples of such systems are found in basalt (e.g. at Mid-Ocean Ridges) or brittle crystalline rocks (Acocella et al., 2000, 2003; Caine and Forster, 1999; Clifton and Schlische, 2003; Clifton et al., 2003; Duffield, 1975; Gudmundsson, 1992, 2000, 2001; Gudmundsson and Bäckström, 1991; Holland et al., 2006; Le Gall et al., 2000). Other examples are found in cemented carbonates and mudrocks in parts of sedimentary basins in which diagenetic processes have led to a strong increase of the compressive strength (Barton et al., 1998; van Gent et al., 2010). In this paper we only address systems without elevated pore pressures (Townend and Zoback, 2000) and do not consider fault- valve processes and invasion percolation (Cox, 1995, 2005, 2007; Micklethwaite, 2009; Miller, 2000; Miller et al., 1999; Petit et al., 1999). Earth and Planetary Science Letters 304 (2011) 399–406 ⁎ Corresponding author at: Baker Hughes, RDS, Emmerich-Josef Str. 5, D-55116 Mainz, Germany. E-mail address: m.holland@ged.rwth-aachen.de (M. Holland). URL: http://www.ged.rwth-aachen.de (M. Holland, H. van Gent, J.L. Urai). 0012-821X/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.epsl.2011.02.017 Contents lists available at ScienceDirect Earth and Planetary Science Letters journal homepage: www.elsevier.com/locate/epsl