root/trunk/EBaporate/README.tex

\documentclass{article}
\usepackage{fullpage,fonts,page,cxref}
\begin{document}
\title{\tt EBaporate}
\author{Copyright 2002, 2003 by Adam Powell}
\date{February 18, 2003}
\maketitle

\section{Introduction}

{\tt EBaporate} is a new simulation code designed to intelligently estimate the
rate of evaporation of metal and alloying elements from an electron beam
melting furnace.  It is based on the graduate work of Adam Powell in
cooperation with the Sandia National Laboratories Liquid Metal Processing
Laboratory, as described in
\begin{quote}
  A. Powell, J. Van Den Avyle, B. Damkroger, J. Szekely and U. Pal ``Analysis
  of Multicomponent Evaporation in Electron Beam Melting and Refining of
  Titanium Alloys,'' {\em Metall. Trans.} {\bf 38B}, 1227-1239 (1997).
\end{quote}

{\tt EBaporate} takes an image of a liquid metal surface, converts that into a
temperature map, and then gives two estimates of the evaporation rates.  The
first estimate is based on a Clausius-Clapeyron/Langmuir analysis of
evaporation at the surface temperatures.  Then, recognizing that evaporation
rate is a strongly nonlinear function of temperature, and that the beam
produces significant local temperature fluctuations as it scans across the
surface, this program estimates the magnitude of those fluctuations using a 1-D
finite element model of local heat conduction down through a thin surface layer
at a point on the melt during a beam scan cycle (it runs over several cycles
until a ``steady state'' is reached), and then calculates the average
temperature and the average evaporation rate in that cycle.  This is repeated
for several average surface temperatures, resulting in a new curve of
evaporation rate vs. temperature, which is used for the second estimate of
evaporation rate over the melt surface.

For titanium melting under Sandia furnace conditions, the calculated average
evaporation rate is significantly above the evaporation rate at the average
temperature at frequencies up to about 200 Hz, making that frequency the
approximate threshold between regimes I and II as described in the paper.  The
program also estimates the transient velocities generated by the beam spot, and
the resulting Peclet number, which indicates the relative importance of
convection and conduction, and thus the transition to regime III.  In regimes
III and IV, this heat conduction model overestimates the temperature
fluctuations and the average evaporation rate.

\section{The Code}

This code has two front-ends, {\tt ebsurfweb} and {\tt ebaporate}, and the
simulation code is mostly in {\tt ebsurf.c} and {\tt shorts.c}.  Source code
documentation is extracted from the code automatically using the tool {\tt
  cxref} ({\tt http://www.gedanken.demon.co.uk/cxref/}), and is in the
appendix.

\subsection{\tt ebsurfweb}

Although is the original web CGI front end written in 1997, I cannot recommend
using {\tt ebsurfweb} on a web server, as there are many potential buffer
overflows which, while harmless in a personal application, can allow a cracker
an easy target if accessible in a server exposed to the net.

That said, the way to use it on the command line is to set its parameters in
the environment variable {\tt QUERY\_STRING}.  The program does one simulation
of surface layer heat conduction at a given surface (or layer bottom)
temperature, and outputs the results in HTML to standard output, along with
several graphs.  The temperature at the bottom of the layer is held constant
during each layer cycle.  The input parameters are fully explained at {\tt
  http://lyre.mit.edu/~powell/Software/ebparams.html}.

This frontend outputs its results in HTML to the standard output, using the
file {\tt ebtemp.html} as its template, including the input parameters used to
generate those results.  Each input and output parameter name is linked to a
target in {\tt ebparams.html} or {\tt outputs.html} explaining that parameter
in detail.  It also uses templates {\tt onedim.c} for 2-D and {\tt onedim3.c}
for 3-D to generate postscript graphs with the following filenames (\%p is the
process id number, used so that multiple runs don't clobber the results):
\begin{itemize}
\item {\tt temp\%p.ps} shows a 3-D graph of calculated temperature vs. time and
  depth, with one curve for each timestep, over multiple cycles of beam
  scanning, illustrating the convergence to ``steady state''.  It is only
  produced if the number of cycles is specified.

\item {\tt last\%p.ps} shows a 3-D graph of calculated temperature vs. time and
  depth, with one curve for each timestep, over the last converged beam scan
  cycle.

\item {\tt surf\%p.ps} shows the surface temperature as a function of time.

\item {\tt hist\%p.ps} shows the beam power input to the surface point over the
  cycle.

\item {\tt gen\%p.ps} shows the beam heat input as a function of depth into the
  melt.  The graph in Schiller's {\em Electron Beam Technology}, p. 38, is used
  to estimate penetration depth from the voltage, and the volumetric heating
  profile is integrated over each finite element shapefunction to give this
  discretized heat distribution.

\item {\tt flux\%p.ps} shows the flux at the bottom of the surface layer.  If
  it fluctuates by more than 20-30\%, then the surface layer is not thick
  enough, and the {\tt aleph} parameter should be increased to make it thicker.

\item {\tt patt\%p.ps} shows the beam pattern on the surface, if one is
  specified (default is one beam strike at the beginning of each cycle, but
  more complex patterns can be used).
\end{itemize}

Input parameters are contained in the {\tt QUERY\_STRING} environment variable,
following the CGI convention of the NCSA httpd web server, the ``first
generation'' server which was in use when this program was written.  By that
convention, the arguments are separated by the \& character, e.g.:
\begin{quote}
  \tt alpha=0.1\&cyc=5\&pattype=1
\end{quote}
Those arguments are parsed in the {\tt paraminp()} function in {\tt ebsurf.c},
currently at lines 371-511.  Comments for each argument type, and their default
values, are also given there.

The one known bug is that when using the cyc parameter, the postscript graphs
don't quite work.

\subsection{\tt ebaporate}

{\tt ebaporate} calculates the total evaporation rate or (currently) average
aluminum activity coefficient based on a PPM image pixmap read from stdin.  To
run it in Unix, one must redirect stdin to a file such as:
\begin{quote}
  {\tt ebaporate} $<$ map.ppm
\end{quote}
Right now, {\tt ebaporate} can only handle monochromatic images.  The final
version will be able to pull one color from a multicolor image.

The current version also has just one function (called ``func()'') to calculate
evaporation rate or Al activity coefficient from color.  A function which uses
{\tt ebsurf.c} functions to calculate the ``enhanced'' regime III evaporation
rate due to the beam is nearing completion, but having problems which are
taking more time than anticipated to resolve, so it is not being shipped at
this time.

\section{Status, Version, Copyright}

This version, 0.9, is a beta release.  Following testing with real EB data, the
code will be modified and finally released as 1.0.

Copyright 2002, 2003 by Adam Powell, Licensed for unlimited use by members of
the Specialty Metals Producers' Consortium (SMPC).

\appendix

% Begin-Of-Source-Files

\input{ebsurfweb.c.tex}

\input{ebsurf.c.tex}

\input{ebsurf.h.tex}

\input{shorts.c.tex}

\input{ebaporate.c.tex}

% End-Of-Source-Files

\end{document}