- The Windows un-installer does not support multiple versions (v15.1,
v16.1) side by side. If you want to uninstall a previous version then do it
first, before installing a newer version. If an installation is broken,
consult the Installation section in the README file for manual installation
- One feature that is not treated well is the rolling direction parameter
CHI. It is generally advised to use CHI=0 or 180 or restrict CHI to at most
a few degrees.
- The results may contain a significant discretisation error when a small
number of elements (7 x 7, 15 x 15) is used. Particularly the frictional
work appears to be susceptible to this.
- The plot programs for visualizing results in Matlab don't work well with newer Matlab versions (R2014b or newer).
- When Fastsim is used for steady state rolling, the rolling step DQ must be equal to the grid size DX.
- Negative values for the approach aren't dealt with properly in Hertzian problems.
- The NormCG solver may fail in case the solution of the problem
consists of an empty contact area.
- The reading of numerical influence coefficients (C=3) is not fully
- The CONTACT GUI stopped working after updating Java in April 2013. It
worked ok with Java 7 update 15, but not with Java 7 update 22. The problem
occurs also with Java 6 update 45.
The resulting behavior is as follows: selecting an input-file works ok, but
no output is obtained from Check Input or Run Experiment anymore:
You may download a new version of the CONTACT GUI here. Store this file on your computer, and then use it to replace the original version in the bin-directory of the CONTACT installation.
- Long runs with large grids may fail due to the amount of memory that
is required. This is because some data that are used in the computations
are held in memory for a long time. These are cleaned up when the program
terminates. If you experience problems because of this ("forrtl: severe
(41): insufficient virtual memory"), you may want to upgrade to version
- The messages displayed by the CONTACT GUI can be confusing at first to
new users. The CONTACT executable says that it's done: "If you want to
continue, you must restart this program", then the Java GUI reports that
the executable is done: "Execution terminated". No error is involved
- There's a small bug in the guided input facility concerning Hertzian
inputs. Currently you cannot use the guided input for selecting method
IPOTCN=-1 (curvatures A1, B1).
- On Windows 7 (and probably Vista as well), write access is restricted
for the directory Program Files (x86). You should move the examples directory
to a different location before running the program.
- The results of visco-elastic problems should be treated with care. The
corresponding parts of CONTACT have not been validated for a long time. We
particularly expect inaccuracies when using different materials for the two
- The file-name for .mat-files is restricted to 72 characters. This
causes problems when using the CONTACT GUI, which uses full pathnames, when
using the default long installation path.
- When reading input for the subsurface stress calculation, some errors
in the user input are not well detected. Particularly "end-of-file". If you
provide less coordinate values than you said you would, the program seems
to hang indefinitively. Also, better not use z-coordinate exactly 0,
because this causes difficulties for the program as well.
- In non-quasiidentical problems, i.e. when using different material
constants for the contacting bodies, there was a serious error in
the interaction between normal and tangential parts. For instance
the theoretical results for the Spence compression problem
could not be reproduced with version v10.1. This error is corrected
in the current release.
- A problem existed with respect to the Carter2D example in the
previous release. The ConvexGS solver did not always converge when
run on a multi-core PC. This is resolved by using the SteadyGS
solver instead. It may also be circumvented by disabling parallel
computing on multi-core PC's, by setting the OMP_NUM_THREADS
environment variable to 1.
- In input option IBASE=3 (undeformed distance: "difference of
two sines", the signs of b_3 and b_6 are changed. There was
an inconsistency between the program and its User Guide.
- Small restrictions with respect to the amount of input data for options
IBASE=2 and E=2 have been removed.