5:15 PM - 7:15 PM
[SVC33-P14] Experimental study about the mechanism of Strombolian eruption
Keywords:Strombolian eruption, laboratory experiment, slug flow
Introduction
Jaupart & Vergniolle (1988) proposed a Strombolian eruption model (hereafter referred to as “JV model”) where small bubbles accumulate in the top of a magma reservoir, forming large bubbles that intermittently rise through the conduit and erupt at the surface, through laboratory experiments and other means. On the other hand, Ripepe et al. (2001) observed that slug ascent speeds, estimated from seismic and infrasonic data at Stromboli, were twice as fast as theoretical predictions. Even though Strombolian eruptions are still discussed based on the JV model, this discrepancy remains unexplained. This study reports a modified JV model to investigate the possibility that the observed slug flow upwelling velocity can exceed theoretical values.
Data & Methods
Following the JV model, the experimental setup consists of a water tank (magma reservoir) and a pipe (conduit). The tank has an inner diameter of 29 cm and a height of 20 cm, while the pipe has a 4 cm inner diameter and a 30 cm height, connecting to the tank’s upper center. Water simulates magma, and air is gradually supplied as volcanic gas from the tank’s side. A pressure sensor (Kistler Piezosmart 4045A2) was installed inside the pipe and in the tank, with data recorded at 100 Hz using a logger (Keisoku Giken HKS-9700). The experiment was conducted with the top of the pipe open (open conduit) and closed (closed conduit). The former is the same as in the JV model. The latter case represents the condition in which the uppermost part of the conduit is clogged by pyroclastic material, etc., and was closed before the eruption. In the case of the closed conduit, a thin plastic plate was glued to the top of the pipe with glue. In the open conduit, nothing is installed.
Results
For the open conduit, pressure sensors recorded a saw-tooth signal: a slow increase over minutes to tens of minutes followed by a rapid 1–3 second decrease, with amplitudes of 2–10 hPa. These characteristics are common to both sensors installed in the water tank and in the pipe. The slow increase in pressure is due to the accumulation of small bubbles at the top of the tank, which pushes out the water in the water tank and raises the water level in the pipe. The rapid decrease in pressure is due to the water level in the pipe decreasing as a result of the large bubbles, formed at the top of the tank by the accumulated small bubbles, rising in the pipe as a slug flow and flowing out of the top of the pipe.
In the closed conduit, the sensor installed in the pipe records a gradual increase in pressure due to the accumulation of small bubbles, but when the large bubble starts to intrude the pipe, it further records a sharp 2–16 hPa pressure spike lasting about 1 second, attributed to dynamic pressure from large bubble intrusion and restricted air outflow. Note that the sensor installed at the bottom of the tank does not record such a 1-second increase in pressure.
Discussion & Summary
A 1-second pressure increase was only observed in the closed conduit when large bubbles entered the pipe and may destroy the lid sealing the conduit. Since the destruction of the lid would initiate the eruption, there is no need for time for slug flow ascent, explaining why the time difference between seismic and airborne waves measured by Ripepe et al. (2001) is smaller than predicted by theoretical slug rise rates.
By analyzing the behavior and pressure fluctuations of the slug flow in open and closed conduit conditions using a water tank and a pipe, our results suggest that assuming a closed conduit in Strombolian eruptions can explain the discrepancy in the apparent velocity of the slug flow, which has not been examined for many years.
Jaupart & Vergniolle (1988) proposed a Strombolian eruption model (hereafter referred to as “JV model”) where small bubbles accumulate in the top of a magma reservoir, forming large bubbles that intermittently rise through the conduit and erupt at the surface, through laboratory experiments and other means. On the other hand, Ripepe et al. (2001) observed that slug ascent speeds, estimated from seismic and infrasonic data at Stromboli, were twice as fast as theoretical predictions. Even though Strombolian eruptions are still discussed based on the JV model, this discrepancy remains unexplained. This study reports a modified JV model to investigate the possibility that the observed slug flow upwelling velocity can exceed theoretical values.
Data & Methods
Following the JV model, the experimental setup consists of a water tank (magma reservoir) and a pipe (conduit). The tank has an inner diameter of 29 cm and a height of 20 cm, while the pipe has a 4 cm inner diameter and a 30 cm height, connecting to the tank’s upper center. Water simulates magma, and air is gradually supplied as volcanic gas from the tank’s side. A pressure sensor (Kistler Piezosmart 4045A2) was installed inside the pipe and in the tank, with data recorded at 100 Hz using a logger (Keisoku Giken HKS-9700). The experiment was conducted with the top of the pipe open (open conduit) and closed (closed conduit). The former is the same as in the JV model. The latter case represents the condition in which the uppermost part of the conduit is clogged by pyroclastic material, etc., and was closed before the eruption. In the case of the closed conduit, a thin plastic plate was glued to the top of the pipe with glue. In the open conduit, nothing is installed.
Results
For the open conduit, pressure sensors recorded a saw-tooth signal: a slow increase over minutes to tens of minutes followed by a rapid 1–3 second decrease, with amplitudes of 2–10 hPa. These characteristics are common to both sensors installed in the water tank and in the pipe. The slow increase in pressure is due to the accumulation of small bubbles at the top of the tank, which pushes out the water in the water tank and raises the water level in the pipe. The rapid decrease in pressure is due to the water level in the pipe decreasing as a result of the large bubbles, formed at the top of the tank by the accumulated small bubbles, rising in the pipe as a slug flow and flowing out of the top of the pipe.
In the closed conduit, the sensor installed in the pipe records a gradual increase in pressure due to the accumulation of small bubbles, but when the large bubble starts to intrude the pipe, it further records a sharp 2–16 hPa pressure spike lasting about 1 second, attributed to dynamic pressure from large bubble intrusion and restricted air outflow. Note that the sensor installed at the bottom of the tank does not record such a 1-second increase in pressure.
Discussion & Summary
A 1-second pressure increase was only observed in the closed conduit when large bubbles entered the pipe and may destroy the lid sealing the conduit. Since the destruction of the lid would initiate the eruption, there is no need for time for slug flow ascent, explaining why the time difference between seismic and airborne waves measured by Ripepe et al. (2001) is smaller than predicted by theoretical slug rise rates.
By analyzing the behavior and pressure fluctuations of the slug flow in open and closed conduit conditions using a water tank and a pipe, our results suggest that assuming a closed conduit in Strombolian eruptions can explain the discrepancy in the apparent velocity of the slug flow, which has not been examined for many years.