Subject: Incode Proceedings 6/4/02
Date: Thu, 06 Jun 2002 16:39:56 -0500
From: "Robert A. Fiedler" 

Note that EnCoDe has been changed to Incode for Integrated
Code Development team.

ARC ROCKETS

We recently received extensive data on some small attitude control
rockets manufactured by Atlantic Research Corp., including detailed
blueprints and pressure histories.

We looked at one axially symmetric rocket this morning as a
geometry for Damrong to make a CAD model of in Pro/E.  The
main difficulty we see is that the propellant tapers to zero
thickness towards both the head and aft ends, so the topology will
change immediately after it begins to burn back.  In the lab scale
rocket (Motor 13), the propellant tapers towards the aft end, but
it still has a substantial thickness there.

We have much work to do to prepare our papers for the AIAA meeting
in July, so we may have to put this project on hold until then,
except for CAD model and possibly mesh generation.

ARTERY

Andreas sees a pressure wave that moves at approximately the sound
speed in the gas, no matter what Young's modulus is in the artery
wall.  The theory says that the propagation speed should depend
on both the sound speed in the gas and the sound speed in the
solid.

The system should support 3 modes of wave propagation --
  1) Pure sound waves in the gas
  2) Pure sound waves in the solid
  3) Coupled gas/solid wave

Scot will run a solids-only simulation with GEN2 to make sure
Rocfrac can propagate at least a pure solids sound wave given
initial conditions like the ones we are using.  The wave speed
had better depend on Young's modulus.

The initial conditions must excite at least the coupled mode.
Bob suggested that Andreas may be detecting only pure sound
waves in the gas, and the coupled mode could be hard to see.

These waves may not be easily visible for long, due to the
initial conditions and numerical dispersion.

FLEXIBLE INHIBITOR

Scot made a CAD drawing that can be read by Gridgen and PATRAN.
Mark made the fluids mesh, and Scot is generating the solids
mesh.  We need to complete is simulation for the AIAA meeting.
The geometry is a section of a "typical large booster".

TITAN

The simulation has run to 651 ms, but the failure occurs at
about 1600 ms, so we are asking for more time on one of the IBM
SPs.  The run stopped before 652 ms with a bad fluids solution
("Not a Number" for some quantities).

Bob has been working on improving the robustness of Rocflo, for
example by forcing the density and thermal pressure to remain
above prescribed floor values, rather than allowing them to go
negative.  The rationale is that bad values of these quantities
are due to numerical errors and they occur mostly in the nozzle
and plume regions, which do not affect the solution upstream of
the nozzle throat, since the flow is supersonic there.  The bad
values are associated with strong transients and shock fronts,
not with low speed flows or boundary layers.

We have already tried every option in Rocflo relevant to the
algorithm, such as first and second order upwind schemes, central
differencing scheme with artificial dissipation, different flux
limiters and averaging nuances, etc.  It just cannot handle very
strong shocks hitting higher density regions very well.  The ratio
of the internal and kinetic energies seems to require some
additional control to prevent unphysically high temperatures and
low densities from developing occasionally in small regions.
Some people use the internal energy density as the integration
variable instead of the total energy density in order to avoid
this problem, although that allows the scheme to violate total
energy conservation.  Bob is refining his "training wheels" to
prevent these high temperatures while conserving total energy.
The density seems to remain well above the floor value on its
own as long as the temperature stays under control.