Efficient Plant July 2018 - 20
feature | maintenance strategies
TRIAL AND ERROR
Use of trial and error to solve a control-system problem can be very helpful.
The best approach is to change only one variable at a time and observe the
effect it has on the issue. When using this method, be sure to document the change
and result with each trial. Otherwise, you could find yourself after, say, performing
15 trials, trying to remember the result of the third one-to no avail.
The trial-and-error method works best for issues that can be repeated rapidly.
For slower issues, such as those that are time dependent, it's possible to
attempt several changes with each trial and record the result. Minimize
the number of changes in each trial and make no more than two or
three changes at a time. If a trial with multiple changes resolves the
issue, then you can independently test the two or three changes to
determine which is the root cause. Keep in mind when changing
multiple variables at a time, they could be dependent upon each
other and further complicate troubleshooting attempts.
starts with a good grasp of the issue.
Computer log files are a good place to begin troubleshooting
if a control-system issue involves any servers, Windows-based
HMIs, or other devices that keep log files. Every Windows computer has a tool
called Event Viewer that can be found in the Administrative Tools of the control
panel. While the Event Viewer contains a substantial amount of information that is
likely irrelevant to the issue, it's necessary to sort and filter. Often the "System" and
"Application" logs provide the most value. Certain applications may also maintain their
own log files, which can provide invaluable troubleshooting insight.
Many PLCs also have the ability to log data points and create trend charts on the
fly. This method is effective for any timing or PLC-code-related issues. Set up a trend
to capture relevant I/O (input/output) points and internal PLC tags that are used in
the logic, then reproduce the issue and see what the trend looks like. When using
this method, pay attention to the rate at which data is being collected as it could alter
results or provide a false view of the situation.
For intermittent issues, setting up some type of a data logger may also help to capture the issue when you're not physically present. If the operators are able to observe
the issue, have them note the exact time when the issue occurs so you can go back and
examine the log files.