Geyser preplay and eruption in a laboratory model with a bubble trap

Geysers are springs that produce episodic eruptions of liquid water and vapor. Relatively short eruption cycles and accessibility of conduits make extensive observation of eruptive processes in geysers feasible. Along with field observations, laboratory models are useful for further describing geyser dynamics. Here we focus on the role of a 'bubble trap', a region in the geyser's plumbing system in which vapor can accumulate. We present measurements from a laboratory model. The model geyser consists of two reservoirs connected by a conduit with a central tight S-shaped bend to create a bubble trap. The conduit is thus divided into two sections: one extends into the upper reservoir and the other is connected to the lower reservoir. A second conduit returns erupted liquid to the lower reservoir. The apparatus is filled with water and heated below the lower reservoir. A period of quiescence follows each eruption. During this phase, a bubble is trapped in the lower S-bend. As the bubble grows, most of its volume remains in the bend while its edges oscillate and vapor is released into the upper conduit. Vapor occasionally reaches the top of the conduit and expels a small amount of liquid. This process may be analogous to geyser preplay. Eruption begins when the upper surface of the main bubble reaches the top of the conduit. We observe two modes of eruption: boiling occurs (1) in the entire system or (2) only in the conduit above the upper boundary of the trapped bubble. In the former case, the rapid hydrostatic pressure drop from filling the upper conduit with vapor results in boiling in the entire system. Eruption ends when enough cold erupted water has been recycled to the lower reservoir that the temperature drops below boiling. Though simpler than a natural geyser, our model provides insight into preplay and eruption styles in a conduit with a bubble trap, a feature that has been invoked to explain dynamics of geysers in Kamchatka and Yellowstone.