Elsevier

Tectonophysics

Volumes 526–529, 10 March 2012, Pages 97-109
Tectonophysics

Numerical modelling of magma transport in dykes

https://doi.org/10.1016/j.tecto.2011.05.015Get rights and content

Abstract

The rheology and dynamics of an ascending pure melt in a dyke have been extensively studied in the past. From field observations, it is apparent that most dykes actually contain a crystalline load. The presence of a crystalline load modifies the effective rheology of such a system and thus the flow behaviour. Indeed, the higher density and viscosity of each crystal, compared to the melt, cause a decrease of the ascent velocity and modify the shape of the velocity profile, from a typical Poiseuille flow, to a Bingham-type flow. A common feature observed in the field is the arrangement of crystals parallel or at a very low angle to the edge of the dyke. Such a structural arrangement is often interpreted as the result of magma flow, which caused the crystals to rotate and align within the flow direction, but this process remains unclear. Another issue related to the introduction of a crystalline load concerns the possibility for crystals to be segregated from a viscous granitic melt phase during magma ascent. The implications of such a process on magmatic differentiation have not previously been considered, nor has such a process been previously investigated via numerical models. In this study, we examine the flow dynamics of a crystal bearing granitic melt ascending in a dyke via numerical models. In our models, both the crystal and melt phases are represented as highly viscous fluids in a Stokes regime. Our results reveal that the presence of crystals in the melt modifies the magma velocity profile across the dyke. Furthermore, we observe that whilst crystals continually rotate in the shear flow, over one period of revolution, their major axis has a high probability to be aligned parallel to the flow direction. Moreover, some experiments showed that the melt phase can effectively be squeezed out from a crystal-rich magma when subjected to a given pressure gradient range. This demonstrates that crystal-melt segregation in dykes during granitic magma ascent constitutes a viable mechanism for magmatic differentiation.

Research highlights

► We developed a new mechanical code to model the magmatic processes. ► Results explain why observation of frozen dykes show crystals aligned parallel to flow direction. ► Presence of crystals modifies the velocity profile from a typical Poiseuille flow shape to a Bingham-type shape. ► The segregation of granitic melt from an ascending crystal-rich magma is physically possible.

Introduction

The processes related to magma ascent from source zones towards emplacement sites of magmatic intrusions in the upper crust constitute a major subject of interest in Earth sciences, especially in terms of understanding intra-crustal differentiation. In particular, the mechanism leading to granitic melt migration towards the upper continental crust (represented by the “diapirism” and “dyking” end-members) has been controversially discussed throughout the 20th century (see reviews by Clemens, 1998, Petford, 2003). Nevertheless, since the study of Clemens and Mawer (1992), it is now largely agreed that the most viable mechanism for the migration of magma, from the deep partial melting zone where they form, to the upper crust where they emplace, is dyking (e.g., Clemens, 1998, Clemens, 2003, Clemens and Petford, 1999, Petford et al., 1994, Scaillet et al., 1998).

The rheology and the behaviour of a multiphase magma (i.e., composed of suspended crystals carried by a viscous medium) subject to a given pressure gradient are governed by the amount of crystals and their geometry (e.g., Bagdassarov and Dorfman, 1998). Here we do not consider the effect of gas bubbles because the processes that we study occur at a depth where volatiles are dissolved in the melt phase. When the amount of crystals is small, silicate melts are considered as Newtonian fluids and their behaviour follows the Einstein–Roscoe relations (Einstein, 1906, Roscoe, 1953). Above a critical solid fraction of suspended crystals, depending on their size, shape and distribution in the magma, the suspension can form a rigid skeleton (Philpotts et al., 1998), which introduces a yield stress in the magmatic suspension and thus results in an effective non-Newtonian rheology (e.g. Kerr and Lister, 1991). The volume of crystals at which the transition from a Newtonian to a non-Newtonian rheology occurs has been estimated somewhere between 15 and 50 vol.% (Champallier et al., 2008, Hallot et al., 1996, Kerr and Lister, 1991, Petford, 2003, Petford and Koenders, 1998, Walsh and Saar, 2008). When the magma contains ca. 55% of solid particles, only residual liquids can escape the rigid skeleton formed by crystals, an effect known as the “rigid percolation threshold” (Vigneresse et al., 1996). According to these authors, when ca. 75% of crystallisation occurs, the entire system becomes totally locked, preventing further mechanical melt percolation from occurring.

Field observations of crystal arrangement in frozen dykes reveal that both their repartitions and their orientations are not random (e.g., Chistyakova and Latypov, 2010, Paterson, 2009, Smith, 2002). In many instances, the crystals are found to be orientated with their major axis parallel, or at a low angle, to the edges of the dyke following the magma flow direction. However, the processes of crystal re-orientation and of their spatial organisation acting during magma transport cannot be directly observed and thus have to be modelled. The rotation of solids immersed in a deforming viscous medium has been addressed in numerous analogue experiments (e.g., Arbaret et al., 1996, Arbaret et al., 2001, Marques and Burlini, 2008, Marques and Coelho, 2001, Marques and Coelho, 2003, Van den Driessche and Brun, 1987, Willis, 1977) and numerical studies (Bons et al., 1996, Mandal et al., 2001, Marques et al., 2005a, Marques et al., 2005b, Samanta et al., 2002, Schmid, 2005). Also, observations of the crystal size distribution within dykes often show that the crystals are sorted by their size, increasing from the edges of the dyke to the centre (e.g., Nkono et al., 2006 and references therein). This can be attributed to their mechanical segregation during the magmatic transport, a phenomenon known as the “Bagnold effect” (e.g., Bagnold, 1954, Barrière, 1976, Bhattacharji, 1967, Komar, 1972a, Komar, 1972b). Besides the fact that these two phenomena should occur coevally during crystal-bearing magma transport in dykes, this complex mechanism remains poorly constrained. Numerous experimental (e.g., Bagdassarov and Dorfman, 1998) and numerical (e.g., Deubelbeiss et al., 2010) studies have been undertaken on magmatic suspension containing particles. However, they mainly addressed the issue of quantifying the effective viscosity of the crystal-melt system and the related rheological consequences, with an emphasis on volcanism (Caricchi et al., 2007, Costa et al., 2007, Dingwell, 1996, Melnik and Sparks, 1999, Papale, 1999, Taisne and Jaupart, 2011).

Finally, another fundamental aspect of magma dynamics is the capacity of crystal-melt segregation to occur during magma ascent in dykes, which has important consequences for magmatic differentiation processes. Crystal-melt fractionation is controlled by factors such as the density difference between the solid and liquid phases, the viscosity of the melt phase, the crystal size and the dynamics of the system. In granitic rocks, such a process is considered to be difficult to initiate, because of (i) the common belief that granitic melts are highly viscous and (ii) the lack of a sufficiently high density difference between the minerals and the residual liquid. Granitic melt viscosities are in the range 104–106 Pa.s (e.g., Clemens, 1998, Scaillet et al., 1998), and the density difference between the melt and common crystals is typically in the range of 200–400 kg.m 3.

To our knowledge, numerical modelling has not been used to study crystal-melt segregation processes that might take place in dykes at depth. In order to address this issue, we propose to use a fluid dynamic description of creeping flow (Stokes) to represent both the crystal and melt phases in one coupled system. We first verify that the numerical method developed can reproduce the known behaviour of highly viscous, or rigid, inclusion subject to both simple and pure shear boundary conditions. In these tests, we quantitatively compared our numerical results to analytical solutions. Then, we describe a model setup that can be used to understand field observations such as crystal orientation in dykes. Finally, we show that crystal-melt segregation is actually a viable mechanism during granitic magma ascent. In this paper, we present examples where all of the crystals introduced possess identical material properties and geometry.

Section snippets

Numerical model

In order to study the dynamics of a crystal suspension in an ascending magmatic flow, we define the system to be composed of highly viscous fluids. The equations governing creeping flow in two-dimensions are given by the Stokes equations (Eqs. (1), (2)) subject to the incompressibility constraint (Eq. (3)):Px+τxxx+τxzz=0,Pz+τzzz+τzxx=ρg,Vxx+Vzz=0,where P, τij, ρ and g correspond to the pressure, the deviatoric stress tensor, the density, and the gravitational acceleration,

Verification of the numerical scheme

Before using our code to study flows with randomly distributed crystals, we performed several experiments involving a viscous inclusion for which we had an analytic solution for the velocity and pressure field. These tests were conducted in order to understand the discretisation errors associated with the method, and to verify that these discretisation errors decreased at the appropriate rate as the numerical resolution in the model was increased. In the crystal-free case, the flow induced by a

Model setup

Our model setup is constructed in a manner to produce an effective pressure gradient between the base and the top of a channel (simulating the dyke), by using a rigid piston pushed in a fluid perforated by a hole (Fig. 5). The fluid corresponds to the melt phase of the magmatic material that can be filled with crystals. The term magmatic, as in a mineralogical point of view, means a mixture composed of both melt and crystals (as in Fig. 5). The two main advantages of such a model are (1) that a

Conclusion and perspectives

In this study we have presented a numerical technique to model magmatic flows within a dyke. The main results can be summarised as follow:

  • (1)

    Our code has been verified against several analytical solutions that possess characteristics similar to those found in crystal-melt systems. These tests indicate that the marker to node interpolation using 1-cell area is more accurate for problems that include rotating, non-circular rigid-bodies.

  • (2)

    The simulations presented here clearly illustrate that crystals

Acknowledgments

We first thank F.O. Marques and the GeoMod2010 team for the workshop organisation where multiple discussions and debates have initiated this work. Discussions with B. Kaus and M. Dabrowski were greatly appreciated. A special “cпacибo” is addressed to T. Gerya for his constant support and enthusiasm during coding using the M-I-C method. Multiple stimulating discussions with E. Hallot, K. Gallagher, P. Boulvais, M. Poujol, W. Husson and B. Cordonnier were also appreciated. Finally, we also thank

References (62)

  • S.K. Ghosh et al.

    Reorientation of inclusions by combination of pure shear and simple shear

    Tectonophysics

    (1976)
  • E. Hallot et al.

    Non-Newtonian effects during injection in partially crystallised magmas

    J. Volcanol. Geotherm. Res.

    (1996)
  • B. Ildefonse et al.

    Effect of mechanical interactions on the development of shape preferred orientation: a two-dimensional experimental approach

    J. Struct. Geol.

    (1992)
  • N. Mandal et al.

    Numerical modeling of heterogeneous flow fields around rigid objects with special reference to particle paths, strain shadows and foliation drag

    Tectonophysics

    (2001)
  • F.O. Marques et al.

    Rigid inclusions rotate in geologic materials as shown by torsion experiments

    J. Struct. Geol.

    (2008)
  • F.O. Marques et al.

    Rotation of rigid elliptical cylinders in viscous simple shear flow: analogue experiments

    J. Struct. Geol.

    (2001)
  • F.O. Marques et al.

    2-D shape preferred orientations of rigid particles in transtensional viscous flow

    J. Struct. Geol.

    (2003)
  • F.O. Marques et al.

    Influence of a low-viscosity layer between rigid inclusion and viscous matrix on inclusion rotation and matrix flow: a numerical study

    Tectonophysics

    (2005)
  • F.O. Marques et al.

    2D rotation of rigid inclusions in confined bulk simple shear flow: a numerical study

    J. Struct. Geol.

    (2005)
  • C. Nkono et al.

    Flowage differentiation in an andesitic dyke of the Motru Dyke Swarm (Southern Carpathians, Romania) inferred from AMS, CSD and geochemistry

    J. Volcanol. Geoth. Res.

    (2006)
  • N. Petford et al.

    The ascent of felsic magmas in dykes

    Lithos

    (1994)
  • S.K. Samanta et al.

    Development of structures under the influence of heterogeneous flow field around rigid inclusions; insights from theoretical and numerical models

    Earth Sci. Rev.

    (2002)
  • J.V. Smith

    Structural analysis of f low-related textures in lavas

    Earth Sci. Rev.

    (2002)
  • B. Taisne et al.

    Magma expansion and fragmentation in a propagating dyke

    Earth Planet. Sci. Lett.

    (2011)
  • R.H. Vernon et al.

    How late are K-feldspar megacrysts in granites?

    Lithos

    (2008)
  • S.D.C. Walsh et al.

    Numerical models of stiffness and yield stress growth in crystal-melt suspensions

    Earth Planet. Sci. Lett.

    (2008)
  • N. Bagdassarov et al.

    Granite rheology: magma flow and melt migration

    J. Geol. Soc.

    (1998)
  • R.A. Bagnold

    Experiments on a gravity-free dispersion of large solid spheres in a Newtonian fluid under shear

    R. Soc. Lond. Proc.

    (1954)
  • M. Barrière

    Flowage differentiation: limitation of the “Bagnold effect” to the narrow intrusions

    Contrib. Mineral. Petrol.

    (1976)
  • S. Bhattacharji

    Mechanics of flow differentiation in ultramafic and mafic sills

    J. Geol.

    (1967)
  • P.D. Bons et al.

    The development of δ-clasts in non-linear viscous materials: a numerical approach

    Tectonophysics

    (1996)
  • Cited by (35)

    • Deciphering syn- and post-emplacement processes in shallow mafic dykes using magnetic anisotropy

      2022, Journal of Volcanology and Geothermal Research
      Citation Excerpt :

      As dyke emplacement occurs in the subsurface it is very difficult to monitor in real-time, as only indirect signals such as ground deformation and seismicity can be measured during intrusion, e.g. during the eruption of Bardarbunga, Iceland in 2014 (e.g. Hjartardóttir et al., 2014; Sigmundsson et al., 2015). Numerical (e.g. Gudmundsson et al., 1999; Yamato et al., 2012) and analogue modelling techniques (e.g. Tibaldi et al., 2014; Kavanagh et al., 2018) are based on conceptual dyke emplacement models that are grounded in field and geophysical observations. Most of these studies investigate the mechanics of dyke propagation and are focused on the role of the host-rock deformation, however, understanding the processes occurring within the flowing magma, and how this is recorded during magma solidification, is also vital.

    • Emplacement of the Zindeng phonolitic lava flow (West-Cameroon) in the Cameroon volcanic line: Constraints from the anisotropy of magnetic susceptibility (AMS)

      2020, Journal of African Earth Sciences
      Citation Excerpt :

      It has been demonstrated that the magnetic fabric is representative of the whole mineral fabric in plutonic rocks (Archanjo et al., 1995; Grégoire et al., 1998), as well as in volcanic rocks, either of basaltic, trachytic or rhyolitic composition (Knight and Walker, 1988; Gountié Dedzo et al., 2011; Bella Nke et al., 2014). In volcanic rocks, AMS has been a very useful tool, used during the past three decades, to study the internal fabric and commonly, to access the ill-defined fabric of pyroclastic flows, tabular bodies such as sills, volcanic dykes or domes (Ernst and Baragar, 1992; Tarling and Hrouda, 1993; Raposo and Ernesto, 1995; Callot and Guichet, 2003; Borradaile and Jackson, 2004; Silva et al., 2004; Egydio-Silva et al., 2005; Gil-Imaz et al., 2006; Borradaile and Gauthier, 2006; Craddock et al., 2008; Cañón-Tapia et al., 2009; Creixell et al., 2009; Bolle et al., 2010; Eriksson et al., 2011; Yamato et al., 2012; Bella Nke et al., 2014; among others). It provides important informations about their flow mechanism and can also help in locating feeder magma chambers (Ray et al., 2007; Silva et al., 2010; Gountié Dedzo et al., 2011; Bella Nke et al., 2014; Dinni et al., 2016 amongst others).

    View all citing articles on Scopus
    View full text