To: SEAonc seaint <seaint(--nospam--at)seaint.org>
Subject: Computer Program Errors
From: merrick group <merrickgroup(--nospam--at)compuserve.com>
Date: Tue, 18 May 1999 12:30:45 -0400
Computer Program Errors
I have found errors in computer programs. I have
reported those errors to the PROGRAMMER.
I get responses like "Send a mathematical proof",
or "Oh OK". The new version is then published with
the same error. Other USERS report that they were never
notified of the errors I reported to the programmer.
SOME POSSIBLE ERRORS TO LOOK FOR:
PLEASE ADD TO THIS LIST.
Diaphragm openings acting as "black holes"
devouring forces that then never make it to
the base shear. The total base shear is difficult
to add up from the printout, but if accomplished,
it does not equal the intended base shear. The
fix is to not have an opening with edges not in
line with shear walls. The vaccination is to
always add up elements in a program to check the
Beams deflections less than hand calculated
deflections, for some choices of materials.
Every beam result must be hand checked. (I opted
to not use the program)
Partition weights not distributed to the roof for
seismic analysis. The remedy is to calculate the
system by hand or with your own program.
Rigid diaphragms not considered.
Shear wall programs where Eschewed shear wall
components being divided into the given axis, and
not recognizing that the principle axis is rotated
from the given axis. Solution is to iterate angles
to find the principle axis. Other solutions are to
use other programs, or implement the x and y
deflections of the x and/or y force to calculate
the principle axis. Recall that principle axis is
where the deflection vector is parallel to the
If columns don't line up with the next support
below one program transfers the load to an
imaginary pier footing that is not indicated in
results. There is no simple way to add up
all known footing loads to check the total. I
found a way to list column lengths to
discover such errors.
Programs that don't print out a full input, list
of calculation methods and a full out-put.
The only way to check such is to make your own
program and check it. Plan check can cause a
problem even if the results are correct.
Beam equations developing beam torsion stability
for narrow "I" shapes that are incomplete.
No warning of known errors. How about that! If a
programmer is aware of this, how can there be no
liability FOR NOT TELLING HIS USERS?
The above are only for discussion and must not be
considered or used as opinion. The list is
incomplete and is intended only to understand that
a remedy may be needed for commercial programs.