Content uploaded by Charles Mader
Author content
All content in this area was uploaded by Charles Mader on Mar 25, 2015
Content may be subject to copyright.
Available via license: CC BY 4.0
Content may be subject to copyright.
NUMERICAL MODEL FOR THE KRAKATOA HYDROVOLCANIC
EXPLOSION AND TSUNAMI
Charles L. Mader
Mader Consulting Co.
Honolulu, HI 96825 U.S.A.
Michael L. Gittings
Science Applications International Corporation
Los Alamos, NM 87544 U.S.A.
ABSTRACT
Krakatoa exploded August 27, 1883 obliterating 5 square miles of land and leaving a
crater 3.5 miles across and 200-300 meters deep. Thirty three feet high tsunami waves
hit Anjer and Merak demolishing the towns and killing over 10,000 people. In Merak the
wave rose to 135 feet above sea level and moved 100 ton coral blocks up on the shore.
Tsunami waves swept over 300 coastal towns and villages killing 40,000 people. The
sea withdrew at Bombay, India and killed one person in Sri Lanka.
The tsunami was produced by a hydrovolcanic explosion and the associated shock
wave and pyroclastic flows.
A hydrovolcanic explosion is generated by the interaction of hot magma with ground
water. It is called Surtseyan after the 1963 explosive eruption off Iceland. The water
flashes to steam and expands explosively. Liquid water becoming water gas at constant
volume generates a pressure of 30,000 atmospheres.
The Krakatoa hydrovolcanic explosion was modeled using the full Navier-Stokes
AMR Eulerian compressible hydrodynamic code called SAGE which includes the high
pressure physics of explosions.
The water in the hydrovolcanic explosion was described as liquid water heated by the
magma to 1100 degree Kelvin or 19 kcal/mole. The high temperature water is an
explosive with the hot liquid water going to a water gas. The BKW steady state
detonation state has a peak pressure of 89 kilobars, a propagation velocity of 5900
meters/second and the water is compressed to 1.33 grams/cc.
The observed Krakatoa tsunami had a period of less than 5 minutes and wavelength of
less than 7 kilometers and thus rapidly decayed. The far field tsunami wave was
negligible. The air shock generated by the hydrovolcanic explosion propagated around
the world and coupled to the ocean resulting in the explosion being recorded on tide
gauges around the world.
Science of Tsunami Hazards, Vol. 24, No. 3, page 174 (2006)
INTRODUCTION
The Krakatoa volcanic explosion and its consequences are described in detail by
George Pararas-Carayannis in reference 1 and Simon Winchester in reference 2.
Krakatoa exploded August 27, 1883 obliterating 5 square miles of land and leaving a
crater 3.5 miles across and 200-300 meters deep. Thirty three feet high tsunami waves
hit Anjer and Merak, demolishing the towns and killing over 10,000 people. In Merak
the wave rose to 135 feet above sea level and moved 100 ton coral blocks up on the
shore. Tsunami waves swept over 300 coastal towns and villages killing 40,000 people.
The sea withdrew at Bombay, India and killed one person in Sri Lanka.
The tsunami was produced by a hydrovolcanic explosion and the associated shock
wave and pyroclastic flows.
A hydrovolcanic explosion is generated by the interaction of hot magma with ground
water. It is called Surtseyan after the 1963 explosive eruption off Iceland. The water
flashes to steam and expands explosively. Liquid water becoming water gas at constant
volume generates a pressure of 30,000 atmospheres.
The Krakatoa hydrovolcanic explosion was modeled using the full Navier-Stokes
AMR Eulerian compressible hydrodynamic code called SAGE with includes the high
pressure physics of explosions.
The observed Krakatoa tsunami had a period of less than 5 minutes and wavelength of
less than 7 kilometers and thus rapidly decayed. The far field tsunami wave was
negligible. The air shock generated by the hydrovolcanic explosion propagated around
the world and coupled to the ocean resulting in the explosion being recorded on tide
gauges around the world.
NUMERICAL MODELING
The compressible Navier-Stokes equations are described in reference 3 and 4 and
examples of many numerical solutions of complicated physical problems are described.
The compressible Navier-Stokes equations are solved by a high-resolution Godunov
differencing scheme using an adaptive grid technique described in reference 5.
The solution technique uses Continuous Adaptive Mesh Refinement (CAMR). The
decision to refine the grid is made cell-by-cell continuously throughout the calculation.
The computing is concentrated on the regions of the problem which require high
resolution.
Refinement occurs when gradients in physical properties (density, pressure,
temperature, material constitution) exceed defined limits, down to a specified minimum
cell size for each material. The mesh refinement is described in detail in reference 3.
Much larger computational volumes, times and differences in scale can be simulated
than possible using previous Eulerian techniques such as those described in reference 4.
The original code was called SAGE. A later version with radiation is called RAGE.
A recent version with the techniques for modeling reactive flow described in reference 3
is called NOBEL. It was used for the modeling of hydrovolcanic explosions described in
this paper.
Some of the remarkable advances in fluid physics using the SAGE code have been the
modeling of Richtmyer-Meshkov and shock induced instabilities described in references
Science of Tsunami Hazards, Vol. 24, No. 3, page 175 (2006)
6 and 7. It was used for modeling the Lituya Bay impact landslide generated tsunami
and water cavity generation described in references 8 and 9. NOBEL/SAGE/RAGE were
used to model the generation of water cavities by projectiles and explosions and the
resulting water waves in reference 10. The codes were used to model asteroid impacts
with the ocean and the resulting tsunami waves in references 11 and 12.
The codes can describe one-dimensional slab or spherical geometry, two-dimensional
slab or cylindrical geometry, and three-dimensional Cartesian geometry.
Because modern supercomputing is currently done on clusters of machines containing
many identical processors, the parallel implementation of the code is very important. For
portability and scalability, the codes use the Message Passing Interface (MPI). Load
leveling is accomplished through the use of an adaptive cell pointer list, in which newly
created daughter cells are placed immediately after the mother cells. Cells are
redistributed among processors at every time step, while keeping mothers and daughters
together. If there are a total of M cells and N processors, this technique gives nearly
(M / N) cells per processor. As neighbor cell variables are needed, the MPI gather/scatter
routines copy those neighbor variables into local scratch memory.
The calculations described in this paper were performed on IBM NetVista and
ThinkPad computers and did not require massive parallel computers.
The codes incorporate multiple material equations of state (analytical or SESAME
tabular). Every cell can in principle contain a mixture of all the materials in a problem
assuming that they are in pressure and temperature equilibrium.
As described in reference 4, pressure and temperature equilibrium is appropriate only
for materials mixed molecularly. The assumption of temperature equilibrium is
inappropriate for mixed cells with interfaces between different materials. The errors
increase with increasing density differences. While the mixture equations of state
described in reference 4 would be more realistic, the problem is minimized by using fine
numerical resolution at interfaces. The amount of mass in mixed cells is kept small
resulting in small errors being introduced by the temperature equilibrium assumption.
Very important for hydrovolcanic explosions, water cavity collapse and the resulting
water wave history is the capability to initialize gravity properly, which is included in the
code. This results in the initial density and initial pressure changing going from the
atmosphere at 2 kilometers altitude down to the ocean surface. Likewise the water
density and pressure changes correctly with ocean depth.
HYDROVOLCANIC MODEL
A hydrovolcanic explosion is generated by the interaction of hot magma with ground
water. It is called Surtseyan after the 1963 explosive eruption off Iceland. The water
flashes to steam and expands explosively. Liquid water becoming water gas at constant
volume generates a pressure of 30,000 atmospheres.
The Krakatoa hydrovolcanic explosion was modeled using the full Navier-Stokes
AMR Eulerian compressible hydrodynamic code called SAGE with includes the high
pressure physics of explosions.
Science of Tsunami Hazards, Vol. 24, No. 3, page 176 (2006)
The water in the hydrovolcanic explosion was described as liquid water heated by the
magma to 1100 degree Kelvin or 19 kcal/mole. The high temperature water is an
explosive with the hot liquid water going to a water gas. The BKW steady state
detonation state described in reference 4 has a peak pressure of 89 kilobars, a propagation
velocity of 5900 meters/second and the water is compressed to 1.33 grams/cc.
THE KRAKATOA MODEL
The island of Krakatoa today and before 1883 are shown in Figure 1.
Figure 1. Maps of Krakatoa today and before 1883.
It was modeled in two-dimensions as a spherical island 200 meters high above the
ocean level and 3 kilometers in radius tapering down to ocean level by 4 kilometers as
shown in Figure 2. The ocean was 100 meters deep and extended in the rock under the
island. The lava was initially assumed to interact with the water in the center of the
island in a 500 meter radius hot spot region. The propagating hydrovolcanic explosion
propagated outward at about 5900 meters per second and at a constant volume pressue of
about 30,000 atmospheres as shown in Figure 3.
Science of Tsunami Hazards, Vol. 24, No. 3, page 177 (2006)
Figure 2. The spherical model for the Krakatoa hydrovolcanic explosion.
Figure 3. The propagating hydrovolcanic explosion.
Science of Tsunami Hazards, Vol. 24, No. 3, page 178 (2006)
The expansion of the hydrovolcanic explosion is shown in Figures 4 and 5 at various
times up to 10 seconds as density picture plots.
Figure 4. The density profile at various times for the hydrovolcanic explosion of
Krakatoa.
Figure 5. The density profile at later times for the hydrovolcanic explosion
of Krakatoa.
Science of Tsunami Hazards, Vol. 24, No. 3, page 179 (2006)
The velocity contour picture plots in the X-direction are shown in Figure 6. The
propagation of the shock wave in the basalt below the island, the basalt above sea level,
the water and in the air is shown.
Figure 6. The velocity profiles in the horizontal or X-Direction at various times.
The water wave profiles at 4, 5, and 8 kilometers are shown in Figure 7. The wave
outside the hydrovolcanic explosion at 4 km is 130 meters high and decays to 48 meters
by 5 kilometers and to 7.5 meters at 8 kilometers.
Figure 7. The water wave profiles as a function of time at 4, 5 and 8 km.
Science of Tsunami Hazards, Vol. 24, No. 3, page 180 (2006)
The states of the water at 1.5 kilometers in the middle of the hydrovolcanic explosion of
the water reached pressures greater than 25,000 bars and expanded to altitudes greater
than 2 kilometers and drove the Krakatoa island basalt to altitudes greater than 2
kilometers. Perhaps the hydrovolcanic explosion looked something like 1946 Bikini
nuclear explosion shown in Figure 8 without the warships. The Baker shot was a 21
kiloton device fired at 27 meter depth in the ocean. The Krakatoa event released 150-200
megatons.
Figure 8. The 1946 Bikini Atomic Explosion.
CONCLUSIONS
A fully-compressible reactive hydrodynamic model for the process of hydrovolcanic
explosion of liquid water to steam at constant volume and pressures of 30,000
atmospheres has been applied to the explosion of Krakatoa in 1883. The idealized
spherical geometry exhibits the general characteristics observed including the destruction
of the island and the projection of the island into high velocity projectiles that travel into
the high upper atmosphere above 2 kilometers. A high wall of water is formed that is
initially higher than 100 meters driven by the shocked water, basalt and air. The initial
wave period of about 30 seconds and the rapid decay of the water wave suggests that the
hydrovolcanic explosion in the calculation was less than in the Krakatoa explosion. The
idealized 2-D geometry needs to be replaced with a realistic 3-D one. The
hydrovolcanic process needs to involve a more accurate description of the water filled
porous basalt layer where the hydrovolcanic explosion occurs.
Science of Tsunami Hazards, Vol. 24, No. 3, page 181 (2006)
The contributions of Dr. George Pararas-Carayannis are gratefully acknowledged.
REFERENCES
1. George Pararas-Carayannis, “Near and Far-Field Effects of Tsunamis Generated by the
Paroxysmal Eruptions, Explosions, Caldera Collapses and Massive Slope Failures of the
Krakatau Volcano in Indonesia on August 26-27, 1883,” Science of Tsunami Hazards,
Vol. 21, No. 4, pp 191-2l1 (2003).
2. Simon Winchester, Krakatoa, Harper Collins Publishers, New York, NY (2003)
3. Charles L. Mader, Numerical Modeling of Water Waves- Second Edition, CRC
Press, Boca Raton, Florida (2004).
4. Charles L. Mader, Numerical Modeling of Explosives and Propellants, CRC Press,
Boca Raton, Florida (1998).
5. M. L. Gittings, “1992 SAIC's Adaptive Grid Eulerian Code,” Defense Nuclear
Agency Numerical Methods Symposium, pp. 28-30 (1992).
6. R. L. Holmes, G. Dimonte, B. Fryxell, M. L. Gittings, J. W. Grove, M. Schneider, D.
H. Sharp, A. L. Velikovich, R. P. Weaver and Q. Zhang, “Richtmyer-Meshkov Instability
Growth: Experiment, Simulation and Theory,” Journal of Fluid Mechanics, Vol. 9, pp.
55-79 (1999).
7. R. M. Baltrusaitis, M. L. Gittings, R. P. Weaver, R. F. Benjamin and J. M. Budzinski,
“Simulation of Shock-Generated Instabilities,” Physics of Fluids, Vol. 8, pp. 2471-2483
(1996).
8. Charles L. Mader, “Modeling the 1958 Lituya Bay Tsunami,” Science of Tsunami
Hazards, Vol. 17, pp. 57-67 (1999).
9. Charles L. Mader, “Modeling the 1958 Lituya Bay Mega-Tsunami, II ,” Science of
Tsunami Hazards, Vol. 20, pp. 241-250 (2002).
10. Charles L. Mader and Michael L. Gittings, “Dynamics of Water Cavity Generation,”
Science of Tsunami Hazards, Vol. 21, pp. 91-118 (2003).
11. Galen Gisler, Robert Weaver, Michael L. Gittings and Charles Mader, “Two- and
Three-Dimensional Simulations of Asteroid Ocean Impacts,” Science of Tsunami
Hazards, Vol. 21, pp. 119-134 (2003).
12. Galen Gisler, Robert Weaver, Michael L. Gittings and Charles Mader, “Two- and
Three-Dimensional Asteroid Impact Simulations,” Computers in Science and
Engineering (2004).
Science of Tsunami Hazards, Vol. 24, No. 3, page 182 (2006)