The Influence of Crystal Size Distributions (CSD) on the Rheology of Magma: New Insights from Analogue Experiments

Knowing the flow properties, or rheology, of magma is of great importance for volcanological research. It is vital for understanding eruptive and depositional features, modelling magma flow rates and distances, interpreting pre-eruptive volcanic unrest and earthquakes, and ultimately predicting volcanic hazards related to magma motion. Despite its key role in governing volcanic processes, magma rheology is extremely difficult to constrain in time and space within a natural volcanic system, because it is dependent upon so many variables. Therefore, both analogue and experimental studies of permissible yet simplified scenarios are needed to isolate different rheological influences. Despite significant progress in understanding the rheological properties of silicate melts and two-phase mixtures (e.g. melt + crystals), as well as the impact of the volume fraction (e.g. Pinkerton & Stevenson, 1992; Caricchi et al., 2007; Mueller et al., 2010) and shape (Mueller et al., 2011) of crystals on magma rheology, the effect of the crystal size distribution (CSD) is still poorly constrained. A highly disperse CSD (i.e., a great variety of different crystal sizes) leads to a much more efficient packing of crystals in a flowing magma which predominantly controls the rheological behavior of magma in a sheared particle Accounting for, or neglecting, the size distribution of crystals can therefore make a considerable difference in magma flow models. We present the results of systematic rheometric experiments using multimodal analogue particle suspensions of well-defined size fractions of micrometer-sized glass beads in silicone oil as magma-analogue material. Starting with simple bimodal distributions (i.e. particles of two distinct sizes), the complexity of the samples' particle size distribution has been successively increased and evaluated towards tetramodal distributions (four distinct size fractions). Statistical values of the given suspensions have been calculated and controlled by image analyses either way to analyze multimodal Gaussian distributions with varying variance (`broadness') as well as with varying skewness (`tailed distributions'). This dataset of analogue experiments is the first of its kind and compared to theoretical and numerical approaches.