Task completion report for investigating why output signal-variable values differ from their output component-parameter values in test problem MST2 Page: 6 of 31
This report is part of the collection entitled: Office of Scientific & Technical Information Technical Reports and was provided to Digital Library by the UNT Libraries Government Documents Department.
The following text was automatically extracted from the image on this page using optical character recognition software:
values were shifted back one timestep in time. Their plotted values that should be
the same are slightly different because of round-off error in the component-data
values that are packed four values per storage word vs signal-variable values that
are stored one value per storage word in the TRCGRF file.
It is interesting to note that the magnitude of the Cray-evaluated oscillatory timestep
divergence evaluated at KAPL on Version 5.4.15 of TRAC-P is 25 times larger than
that which was evaluated on an Ultra SUN workstation with Version 5.4.28. Is this
due to machine dependence or to corrections and improvements made to TRAC-P
between these two versions? To determine if the oscillatory divergence is due to
recent changes to TRAC-P, the MST2 test problem, with added signal variables and
consecutive timestep edits, was evaluated on a Cray YMP computer with TRAC-P
Version 5.3.01 (dated 8/11/92). Figures 8 to 12 show the end-of-problem oscillatory
timestep divergence that was evaluated for the gas volume fraction, phasic mass
flows, and phasic velocities. Its oscillatory-divergence magnitude is 3.5 times greater
than the Cray-evaluated oscillatory divergence at KAPL. Thus, this oscillatory
timestep divergence evaluated at a material Courant number of 1000 in test problem
MST2 has existed since TRAC-PF1/MOD2 Version 5.3. Its magnitude is believed to
have decreased because of corrections and improvements made to TRAC.
A standard test problem should not have oscillatory divergence in its numerical
solution. The oscillatory divergence can be eliminated by changing the MST2 test
problem end time from 10,000 to 7000 s or less. It also can be eliminated by
evaluating TRAC-P with a maximum material Courant number <1000. The later
was demonstrated with the DPTCON update in TRAC-P Version 5.4.28 by inputting
a one-dimensional maximum material Courant number of 500 and evaluating the
MST2 test problem for 50,000 s. This reduced the timestep size from -100 to -50 s,
and the steady-state solution converged to a constant condition without oscillation
TRAC-P outputs signal variables and their component data to the TRCOUT and
TRCGRF files with beginning-of-timestep values for the signal variables and end-of-
timestep values for the component data at the end-of-timestep edit time. This shifts
the signal-variable values forward in time by one timestep making their values
inconsistent with their component-data values in a specific-time output edit. This
inconsistency was noted at KAPL and explained in this report by the above
statement. Further divergence of signal-variable values and their component-data
values after 9000 s in standard test problem MST2 also was observed at KAPL. This
was explained as being due to oscillatory timestep divergence of the numerical
solution evaluated with a material Courant number of 1000 when plotted at only
even- or odd-numbered timesteps. The oscillatory behavior in problem MST2 was
shown to exist in TRAC-PF1/MOD2 Version 5.3.01 as well as in TRAC-P Versions
5.4.15 and 5.4.28. The magnitude of the oscillatory divergence was 25 times greater
in Version 5.4.15, and 87 (25x3.5) times greater in Version 5.3.01 than in the current
Version 5.4.28 of TRAC-P. Reducing the maximum material Courant number from
1000 to 500 eliminated the oscillatory-divergence behavior of the numerical-
solution in the MST2 test problem.
Here’s what’s next.
This report can be searched. Note: Results may vary based on the legibility of text within the document.
Tools / Downloads
Get a copy of this page or view the extracted text.
Citing and Sharing
Basic information for referencing this web page. We also provide extended guidance on usage rights, references, copying or embedding.
Reference the current page of this Report.
Steinke, R.G. Task completion report for investigating why output signal-variable values differ from their output component-parameter values in test problem MST2, report, September 10, 1997; New Mexico. (https://digital.library.unt.edu/ark:/67531/metadc711430/m1/6/: accessed April 20, 2019), University of North Texas Libraries, Digital Library, https://digital.library.unt.edu; crediting UNT Libraries Government Documents Department.