See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/243220294

The “Small-World” topology of rock fracture

networks

Article in Physica A: Statistical Mechanics and its Applications · April 2007

DOI: 10.1016/j.physa.2006.11.025

CITATIONS READS

30 49

3 authors:

Luca Valentini Diego Perugini

University of Padova Università degli Studi di Perugia
28 PUBLICATIONS 226 CITATIONS 188 PUBLICATIONS 2,033 CITATIONS

SEE PROFILE SEE PROFILE

Giampiero Poli
Università degli Studi di Perugia
134 PUBLICATIONS 3,451 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

COST TU1404 WG2 benchmark on numerical modelling of CBM at early age - stage 1 View project

Alkali activated calcined clays View project

All content following this page was uploaded by Giampiero Poli on 15 December 2017.

The user has requested enhancement of the downloaded file.

 ARTICLE IN PRESS

Physica A 377 (2007) 323–328

www.elsevier.com/locate/physa

The ‘‘small-world’’ topology of rock fracture networks

Luca Valentinia, Diego Peruginib,, Giampiero Polib
a
Department of Earth & Ocean Sciences, National University of Ireland, Galway, Ireland
b
Department of Earth Sciences, University of Perugia, Perugia, Italy
Received 3 October 2006; received in revised form 20 October 2006
Available online 4 December 2006

Abstract

In this contribution we show that natural rock fracture networks, acting as media for the delivery of a variety of
geological fluids, can be studied by using the principles of complex systems. Natural networks at different length scales
(from the metre to the micrometer) have been analysed by evaluating their connectivity at global and local scale and results
show that they share topological properties of ‘‘small-worlds’’, a class of networks characterised by high global and local
transport efficiency. This may have important geological implications for a variety of geological processes related to the
transfer of fluids within the Earth, from the delivery of magmas to the Earth surface to the dispersion of pollutants in
shallow aquifers.
r 2006 Elsevier B.V. All rights reserved.

Keywords: Rocks; Fracture networks; Graph theory; ‘‘Small-world’’ networks

1. Introduction

Fracture networks occur in most geological environments. Their development depends on several processes
such as tectonic stresses acting on brittle rocks, volume shrinkage due to solidification of magmatic masses, or
reactive processes generating porous channels. Fracture networks play a fundamental role in a variety of
geological processes such as the migration of magmas within the Earth mantle and/or crust, the flow of water
in aquifers, the dispersion of pollutants in hydrologic systems, and the extraction of oil and gases from
hydrocarbon reservoirs.
The migration of fluids through the Earth’s interior may occur as both pervasive and/or focussed porous
flow. In the former case the flow occurs into an interconnected system of fractures which propagate through
the brittle country rocks (e.g. Ref. [1]), whereas in the latter the fluids flow through intercrystalline porous
spaces and porosity generated by the dissolution of surrounding rocks (e.g. Refs. [2–4]). Such geological
networks have been extensively studied by means of fractal techniques (e.g. Refs. [5–7]) and results have
shown that they can be regarded as the result of a hierarchical fragmentation mechanism propagating from
large to small scales. The fractal nature of fracture networks suggests that they might be regarded as complex

Corresponding author. Tel.: +39 75 5852652; fax: +39 75 5852603.

E-mail address: [email protected] (D. Perugini).

0378-4371/$ - see front matter r 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.physa.2006.11.025
 ARTICLE IN PRESS
324 L. Valentini et al. / Physica A 377 (2007) 323–328

systems, i.e., systems formed by a series of non-linearly interacting elements whose behaviour cannot be fully
characterised by studying the system at the scale of single components (e.g. Ref. [8]). For instance, in the case
of magma migration, it is known that melts are formed at the scale of single mineral grains at the edges and
corners of crystals and flow is attained once these microscopic channels join into larger ones, forming an
interconnected network (e.g. Ref. [9]). Such an inter-granular network will subsequently develop into a system
of metre scale fractures at shallow depth, as part of a branching multi-scale system (e.g. Ref. [5]). Thus, the
development of an efficient system for magma transport stems from the interplay of sets of fractures/conduits
at several length scales.
In this contribution natural fracture networks occurring at different length scales are studied by using
methods of complex networks with the aim of characterizing their topological features. In particular, fracture/
conduit networks are represented as graphs and analysed according to the principles of Graph Theory. It is
shown for the first time that rock fracture networks have typical features of ‘‘small-worlds’’, i.e., a special class
of networks characterised by high global and local transport efficiency. The results presented in this
contribution may give new insights into the structural properties of fracture networks and provide a new tool
for understanding the transport of fluids within the Earth’s interior.

2. ‘‘Small-world’’ networks

Many recent studies on transport phenomena through complex networks have been carried out by using
principles of Graph Theory, which allow the representation of a network system as a set of N  N nodes that
interact through a set of k connections (e.g. Ref. [10]). Such a configuration can be represented as an adjacency
matrix whose entries aij are either 1, if two generic nodes i, j share a connection, or 0 if they do not.
Watts and Strogatz [11] introduced the concept of ‘‘small-world’’ referring to a special class of networks
sharing topological features of both regular graphs (i.e., high connectivity among neighbour nodes) and
random graphs (i.e., short average path length between nodes). These authors have shown that small-world
networks are characterised by a highly efficient connectivity and that such a configuration can be attained by
randomly rewiring a fraction of nodes belonging to an initial regular lattice [11]. This procedure creates a
series of ‘‘shortcuts’’ whose non-linear effect on interacting nodes drastically reduces the distance between any
pair of nodes, though the high clustering, which is typical of regular networks, is preserved.
Latora and Marchiori [12] have proposed a refined version of the small-world theory for applications to
weighted networks, i.e., those networks in which the Euclidean length of each connection is also taken into
account. In particular, they defined the ‘‘efficiency’’ (ij ) of a connection as
1
ij ¼ ,
d ij
where d ij represents the shortest path length between any two nodes in a given network.
The global efficiency E globðeÞ of the entire network is then given by
P
iaj ij 1 X 1
E globðeÞ ¼ ¼ .
NðN  1Þ NðN  1Þ iaj d ij

The value of E globðeÞ is normalised to the ideal case in which the network has all the possible connections
among nodes, the length of connections being equal to the Euclidean distance l ij between each pair of nodes,
so that the global efficiency attains its maximum value
1 X1
E globðmaxÞ ¼
NðN  1Þ iaj l ij

such that 0oE glob p1.

The other fundamental quantity is the ‘‘local efficiency’’ (E loc ) which can be calculated by evaluating
the efficiency of G i (i.e., the sub-graph of the neighbours of i) for each node i. A sub-graph G i , relative to a
generic node i, is defined by all the nodes connected to i, with i and all the connections departing from it
removed [12]. The local efficiency E loc is defined as the average efficiency of the local sub-graphs.
 ARTICLE IN PRESS
L. Valentini et al. / Physica A 377 (2007) 323–328 325

E loc quantifies the extent to which the system is ‘‘fault-tolerant’’ (e.g. Refs. [13,14]), i.e., how much the
efficiency of the transport between the first neighbours of a generic node i is preserved, when i is removed.
According to this formalism, small-world networks are defined as those characterised by large values of both
global (E glob ) and local efficiency (E loc ).

3. Analysis of natural fracture networks

Values of E glob and E loc for a number of natural fracture/conduit networks have been measured with the aim
of defining their topological features in terms of global and local efficiency, respectively. In particular,
fractures occurring in rock samples at several length scales (from the metre to the micrometer) have been
considered. Some examples of the analysed rock fracture networks are given in Figs. 1–3. Natural networks
have been represented as graphs whose nodes correspond to the tips and intersections of fractures (Figs. 1–3).
These graphs have been converted into adjacency matrices, weighted by the normalised length of each
fracture, and values of global efficiency (E glob ) and local efficiency (E loc ) have been calculated. For
comparison, theoretical regular, random, and ‘‘small-world’’ networks of the same size (i.e., number of nodes)
of natural ones have been also analysed. Results of calculations for natural and theoretical networks are given
in Table 1. Regular and random networks are typically characterised by low E glob ðo0:1Þ and high E loc ð40:1Þ
and high E glob ð40:1Þ and low E loc ðo0:1Þ, respectively. ‘‘Small-world’’ networks, on the contrary, have large
values of both E glob and E loc ð40:1Þ. As regards natural networks they all have large values of both global

Fig. 1. Metre size fracture trace map from Ref. [6] (p. 157; BT-C in Table 1). The left side of (a) shows the natural fracture pattern, and the
corresponding graph is presented on the right side of (b).
 ARTICLE IN PRESS
326 L. Valentini et al. / Physica A 377 (2007) 323–328

Fig. 2. Centimetre size aplitic veinlets in a granitoid rock from the Aztec Wash Pluton (USA). The left side of (a) shows the natural
fracture pattern, and the corresponding graph is presented on the right side of (b).

(E glob ) and local (E loc ) efficiency. Therefore, from presented results it is clear that structural properties of
natural networks are consistent with a ‘‘small-world’’ topology.

4. Summary and conclusions

In this contribution we have shown that rock fracture networks can be studied by representing them as
weighted networks and by applying the principles of Graph Theory. Results have shown that fracture
networks have properties of ‘‘small-worlds’’, i.e., they are globally and locally efficient networks. Noteworthy
is the fact that the ‘‘small-world’’ topology is retained by fracture patterns at several length scale, from the
metre to the micrometer.
These results may have important implications for a variety of geological processes concerning the
migration of fluids through geological media. For instance, in the field of igneous petrology, one of the likely
outcomes is that, by travelling through such efficient networks, magmas formed deep in the Earth crust or
mantle can be efficiently delivered to the Earth surface with little modification, thus potentially providing a
great deal of information about the source rocks and the process of magma genesis. A less enthusiastic
scenario is the dispersion of polluted waters in shallow aquifers: the high efficiency of fracture networks may
imply a very fast dispersion of dangerous components thus opening a significant problem of hydrologic
hazard.
Research is in progress with the aim of developing numerical models that couple topological parameters of
the network, such as global and local efficiency, to hydraulic properties of the geological media, such as
permeability. It has been shown, in fact, that the ratio between the permeability resulting from fracture
aperture and matrix permeability (i.e., permeability ratio) may influence the distribution of flow as well as
 ARTICLE IN PRESS
L. Valentini et al. / Physica A 377 (2007) 323–328 327

Fig. 3. (a) Fracture map of an olivine crystal occurring in a basaltic lava flow on the island of Vulcano (Aeolian Islands, Italy). The left
side of (a) shows the natural fracture pattern, and the corresponding graph is presented on the right side of (b).

Table 1
E glob and E loc calculated for some representative natural fracture/conduit networks and corresponding (i.e., same number of nodes and
connections as natural ones) theoretical regular, random, and ‘‘small-world’’ networks

Natural networks Theoretical networks

Regular Random Small-world

E glob E loc E glob E loc E glob E loc E glob E loc

BT-C 0.22 0.24 0.06 0.72 0.16 0.007 0.14 0.66

BT-G 0.21 0.41 0.02 0.72 0.15 0.003 0.21 0.13
OL-1 0.17 0.27 0.03 0.72 0.15 0.004 0.23 0.13
OL-2 0.23 0.49 0.09 0.72 0.18 0.014 0.29 0.28
AP-1 0.25 0.34 0.12 0.72 0.21 0.030 0.27 0.46
AP-2 0.14 0.18 0.08 0.72 0.18 0.010 0.30 0.11
DUN 0.18 0.32 0.06 0.72 0.17 0.007 0.24 0.37

BT-C: fracture trace map from Ref. [6] (p. 157); BT-G: fracture trace map from Ref. [6] (p. 161); OL-1 and OL-2: two different olivine
crystals from a basaltic lava flow on the island of Vulcano (Aeolian Islands, Italy); AP-1 and AP-2: aplitic veinlets in granitoid rocks from
the Aztec Wash Pluton (USA); DUN: dunite conduits exposure from Ref. [4].

fracture connectivity [15]. Detailed studies on the interplay between topological and hydraulic characteristics
of geologic networks may provide a better understanding of transport phenomena through the Earth’s
interior.
 ARTICLE IN PRESS
328 L. Valentini et al. / Physica A 377 (2007) 323–328

Acknowledgements

We thank K.R. Moore for critical reading of the manuscript and for useful suggestions. This work was
funded by MIUR (Ministero dell’Istruzione, dell’Università e della Ricerca) and Università degli Studi di
Perugia grants. Comments of anonymous referees are gratefully acknowledged.

References

[1] A. Nicolas, A melt extraction model based on structural studies in mantle peridotites, J. Petrol. 27 (1986) 999–1022.
[2] P. Kelemen, N. Shimizu, V. Salters, Extraction of mid-ocean ridge basalt from the upwelling mantle by focused flow of melt in dunite
channels, Nature 375 (1995) 747–753.
[3] P. Kelemen, G. Hirth, N. Shimizu, M. Spiegelman, H. Dick, A review of melt migration processes in the adiabatic upwelling mantle
beneath ocean spreading ridges, Philos. Trans. R. Soc. London, Ser. A 355 (1997) 283–318.
[4] P.B. Kelemen, M. Braun, G. Hirth, Spatial distribution of melt conduits in the mantle beneath oceanic spreading ridges: observations
from the Ingalls and Oman Ophiolites, Geochem. Geophys. Geosyst. 1 (2000) doi: 1999GC000012.
[5] S. Hart, Equilibration during mantle melting: a fractal tree model, Proc. Nat. Acad. Sci. 90 (1993) 11914–11918.
[6] C.C. Barton, Fractal analysis of the scaling and spatial clustering of fractures in rock, in: Fractal in the Earth Sciences, Plenum Press,
New York, 1995.
[7] D.L. Turcotte, Fractals and Chaos in Geology and Geophysics, Cambridge University Press, Cambridge, 1992.
[8] Y. Bar-Yam, Dynamics of Complex Systems, Addison-Wesley, Reading, MA, 1997.
[9] S. Maaløe, A. Scheie, The permeability-controlled accumulation of primary magma in a planetary mantle, Phys. Earth Planet. Inter.
29 (1982) 344–353.
[10] F. Harary, Graph Theory, Addison-Wesley Publishing Company, 1969.
[11] D. Watts, S. Strogatz, Collective dynamics of ‘small-world’ networks, Nature 393 (1998) 440–442.
[12] V. Latora, M. Marchiori, Efficient behavior of small-world networks, Phys. Rev. Lett. 87 (2001) 1987011–1987014.
[13] R. Albert, H. Jeong, A. Barabasi, Error and attack tolerance of complex networks, Nature 406 (2000) 378–382.
[14] P. Crucitti, V. Latora, M. Marchiori, A. Rapisarda, Error and attack tolerance of complex networks, Physica A 340 (2004) 388–394.
[15] W. Taylor, D. Pollard, A. Aydin, Fluid flow in discrete joints sets: field observations and numerical simulations, J. Geophys. Res. 104
(B12) (1999) 28983–29006.

View publication stats