Explosive volcanic eruption processes: from mesoscopic simulations to constitutive laws

When volcanoes erupt explosively, they are amongst the deadliest of all natural hazards. An explosive eruption, like that at Mt St Helens in 1980, occurs when the molten rock, or magma, erupts so violently that the magma is broken into small fragments. These fragments rise into the air as volcanic ash and flow down the sides of the volcano as deadly pyroclastic flows. Not all volcanic eruptions are explosive, however. Kilauea in Hawaii has been erupting continuously for over twenty years, but has not exploded in this time. Rather, lava has flowed gently down the volcano and into the sea, causing little hazard to the local population; this is an effusive eruption. Some volcanoes can show both types of behaviour: sometimes erupting explosively, sometimes effusively. These volcanoes, of which Mt St Helens is an example, are particularly dangerous as it is very hard to predict when the eruption will switch from effusive to explosive. A main aim of volcanology is to understand what controls whether an eruption is explosive or effusive, and what makes it switch between the two. My research will help to answer this question. Magma contains many gas bubbles. Volcanologists believe that explosive eruptions occur when the pressure in the bubbles gets high enough to make the surrounding magma break into fragments. If the bubbles get big enough, they can merge and form networks along which gas can flow, allowing it to escape from the bubbles. This means that the pressure can't build up and the eruption is more likely to be effusive. It is vital for volcanologists to know how easily gas can flow along these bubble networks. I am developing a computer programme, LBFLOW that simulates gas flow through the bubble networks in volcanic rocks. I am using this programme to calculate how rapidly gas can escape from bubbles for different types of bubble network. This information will allow me to work out how rapidly the pressure can escape from magma at different volcanoes. I have already used LBFLOW to look at gas flow through small networks of bubbles. In order to make the result as useful as possible, I want to adapt LBFLOW to run on a 'parallel computer', a type of supercomputer that we have at the BP Institute at Cambridge University, where I will carry out this work. Running LBFLOW on this computer will allow me to simulate much larger networks, making the results more reliable. When using computer models, it is vital to check the results against experiments to ensure that the model gives the correct answer. I am going to use an MRI machine, similar to ones used in hospitals to look inside a patient's body, to look at fluid flowing through networks of bubbles in rock. The MRI scan will show how fast the fluid is flowing in different parts of the network. By running LBFLOW on the same bubble network, I can compare the programme's results with the MRI results and ensure that they are the same. There are a couple of major advantages to using LBFLOW for this research. Firstly, because computers are so fast, I can examine many different types of bubble network rapidly. The other advantage is that I can control exactly what the bubble networks are like: i. e. how big the bubbles are and how much they overlap. This will make my results applicable to a wide range of volcanoes. It is also important for volcanologists to know how sticky, or viscous magma is. If the magma is sticky, it is much more likely to erupt explosively than it if is runny. LBFLOW will also show how bubbles affect the viscosity of the magma. I have already performed experiments in the laboratory to look at this. I will use LBFLOW to look at how crystals affect magma's viscosity. The results of my work will help volcanologists to understand the way volcanoes erupt. It will help us to work out what type of eruption is likely to happen at a volcano, and what would make it switch from effusive to explosive activity.