We explore an alternative interpretation of the centrifuge data in which the low melt fraction (< 0.3) centrifuge results are assumed anomalous compared to deformation experiments on partially molten rocks. Under this assumption, we suggest that the compaction in the centrifuge experiments at intermediate melt fractions (> 0.3) is accommodated primarily by particle rearrangements (repacking) with minor GBD contribution. This approach allows us to test whether the discrepancy between the centrifuge data at intermediate melt fractions and earlier rock mechanics experiments at lower melt fraction can be caused by a transition from repacking to GBD as the rate-limiting compaction mechanism with decreasing melt fraction. It is important to note that while this interpretation would provide a framework to characterize compaction over a wide range of melt fraction, it does not reconcile the discrepancy between the low melt fraction centrifuge data and previous rock mechanics experiments.
The distinction between GBD and repacking is important as upscaling the rheology to natural conditions in silicic, crustal magma systems involves different grain size dependence and therefore will yield significantly different melt extraction timescales. At melt fractions below 0.3, effective compaction rates (or compaction viscosities) inferred from the centrifuge experiments (Bagdassarov et al. , 2009, Connolly et al. , 2009) and measured by deformation experiments (Allan et al. , 2013, Hirth & Kohlstedt, 1995, Meiet al. , 2002, Renner et al. , 2003, Scott & Kohlstedt, 2006, Zimmerman & Kohlstedt, 2004) display several orders of magnitude of difference. While we do not directly address compaction at melt fractions below 0.3, given the agreement between the experiments designed to directly measure compaction rate (Allan et al. , 2013, Hirth & Kohlstedt, 1995, Mei et al. , 2002, Renner et al. , 2003, Scott & Kohlstedt, 2006, Zimmerman & Kohlstedt, 2004), we make the assumption here that they provide reliable estimates of compaction rates at low melt fraction (Fig. 1 ). Because of the limited amount of experimental data published for compaction rates at melt fraction > 0.3, we use both the olivine centrifuge experiments at intermediate melt fractions and the repacking experiments of Hoyos et al. (2022) to assess the role of repacking on compaction in this regime. This suite of experiments consists of olivine-MORB melt aggregates that are compacted in a centrifuge apparatus. Crystal-melt separation initially occurs by sedimentation followed by crystal matrix compaction. Interestingly, the melt fractions in this suite of experiments remains above the minimum melt fraction of ca. 0.3, consistent with the minimum melt fraction obtained in natural silicic systems (Lee & Morton, 2015). These observations suggest that the physics governing the minimum amount of trapped melt in the centrifuge experiments and in silicic, crustal magma systems reflect the same processes.
The hypothesis that compaction in the centrifuge experiments may be accommodated by repacking is inspired by a suite of analog experiments on suspensions performed at or near ambient pressures and temperatures (Boyer et al. , 2011, Cassar et al. , 2005, Costa et al. , 2009, Faroughi & Huber, 2015, Hoyos et al. , 2022, Jopet al. , 2006). For instance, the experiments of Boyer et al. (2011) included applied shear stresses to suspensions housed in a shear cell and revealed that at intermediate fractions (ca. 0.4 – 0.6), phase separation is dominated by a combination of hydrodynamic and particle-particle interactions (repacking).
The purpose of this paper is first to establish whether continuum models of compaction can be used to model repacking phenomena (section 2.2). The second objective is to constrain the rheology associated with repacking by developing a compaction model and applying it to the high T + P centrifuge experiments and to the low T + P experiments of Hoyoset al. (2022) (section 3). The experiments of Hoyos et al.(2022), unlike those of Boyer et al. (2011), were performed under conditions of uniaxial compression with no macroscopic shear component. We demonstrate that a transition between compaction by repacking to grain boundary-controlled diffusion near the maximum packing of the mush can explain the sharp increase in effective resistance to compaction between the centrifuge and partially molten rock experiments and is consistent with petrologic observations in natural magma chamber systems. We then simulate a composite grain boundary-controlled diffusion and repacking composite rheology and apply the composite rheology to high strain rate transients measured in Renner et al.(2003). Finally, we extrapolate our results to conditions relevant to silicic, crustal magma systems in the continental crust.