Vulcanian fountain collapse mechanisms revealed by multiphase numerical simulations:

1
Vulcanian fountain collapse mechanisms revealed
by multiphase numerical simulations

• Influence of volatile leakage on eruptive style
  and particle-size segregation
• by A.B. Clarke, B. Voight, A. Neri G. Macedonio

2
Outline

• Montserrat Vulcanian explosions
• Here we test the effect of volatile leakage on
  Vulcanian explosions using a first-order leakage
  model to supply initial conditions for an
  axisymmetric, multiphase numerical model
• Volatile loss can cause change in eruptive style
  from explosive to effusive (Jaupart and Allegre,
  1991 Jaupart, 1998)
• Comparison of models to real events

3
Soufrière Hills volcano, Montserrat, BWI

• Andesite dome-building eruption
• Ongoing since 1995
• 1997 was a very active year, including 88
  Vulcanian explosions

4
Events preceding Vulcanian explosions on
Montserrat
5
(No Transcript)
6

• Duration lt 1 minute
• Plume height 10 km (3 15)
• Magma ejected 0.8 x 109 kg
• Exit velocity 40 140 m s-1

• Fountain collapse height 300 650 m
• Ash-cloud surge velocity 30 60 m s-1
• Pumice flow runout 3 6 km
• Explosion interval 10 hours

7
Numerical model

• Solves Mass, Momentum and Energy for 3 particle
  sizes and a gas phase
• Unsteady vent parameters (mass flux of each
  phase) calculated by model
• Initial conditions and geometric parameters
  obtained from field data (Geometry topography
  OP 10MPa from pumice 3 particle sizes from
  deposits)
• Results of pyroclastic dispersal compared to
  field observations

8
Radial Volatile Leakage
Begin with reference simulation and apply the
leakage model 10 MPa OP 3 particle sizes 20 m
cap 4.3 wt.H20 65vol crystals

• q is mass flow rate of gas per unit area
• ?g ?g are gas density and viscosity
• ? is gas volume fraction
• Pc is gas pressure in the conduit
• Pl is lithostatic pressure
• K is permeability of country rock

• From Jaupart Allegre, (1991)

9
(No Transcript)
10
Results effects of volatile leakage

• SimB (volatile loss)
• more energetic plume
• overhang style
• less mass to flows-68
• higher later fountain collapse
• ------------------------------
• elutriation of fines from pyroclastic current

• SimC (3x SimB loss)
• less energetic plume
• boil-over style
• more mass to flows-82
• lower earlier fountain collapse
• -------------------------------------
• elutriation of fines from pyroclastic current

11
Overhang style
Boil-over style
12
Overhang style
Boil-over style
13

• Elutriation of fines
• Occurred for all simulations was observed in
  real events
• Elutriation was more dramatic for overhang-style
• SimB at 80 s 50 of fines were part of pf, but
  by150 s only 12 of fines remained part of the pf

14
(No Transcript)
15
(No Transcript)
16
(No Transcript)
17
(No Transcript)
18
(No Transcript)
19
Conclusions

• Duplication of real explosions requires some
  volatile leakage and/or delayed exsolution
• Lateral volatile leakage plays an important role
  in explosion style (as well as strength)
• Simulations revealed important mechanisms of
  fountain collapse and particle size segregation

20
(No Transcript)
21
(No Transcript)
22
(No Transcript)
23
Conduit model assumptions flow has
stagnated no viscosity changes with
depth equilibrium degassing constant
crystal volume fraction with depth constant
overpressure with depth Reasonably duplicated
real behavior --- however permeability (or
anything that would reduce gas volume fraction,
such as non-equilibrium degassing) proved to be
significant in overall plume development
24
How should we improve the conduit model?
Results from Melnik 1999 suggest a few
things Still assume equilibrium degassing
Allow for viscosity changes due to crystal
growth degassing Resulting in a non-constant
overpressure with depth and corresponding
vesicularities How do these changes affect
explosion results?
25
Accounting for viscosity changes during
ascent had little affect on plume
ascent rate changed qualitative
behavior of plume changed pyroclastic
flow runout distance
26
How do we test which conduit model best
represents reality? Pumice samples from a
single event assume pumice records
pre-fragmentation conditions Does pumice
record pressure, temperature, and vesicularity
variations with depth? If so, how do we
measure these parameters?
27
Methods Comparison against experiments on
the same magma Matrix glass K2O
composition (varies as the inverse of P and
T) An content (increases with increasing P and
T) Measure matrix glass water content
In conjunction with density to better
understand gas lost from system