I have been working with smart instruments and most of the come with a feature called "Damping Value for Process Variable Measurement".
The quick description of the parameter goes: The sampling interval used to smooth the output.

I would like to understand in which cases it would be recommended to set a damping for the process variable and how exactly does the damping work?

 

Damping is a slowing of response to achieve an effect similar to averaging. To give an idea how it works let's use the gas gauge on your car. The gas sloshes around with bumps and turns and is almost never level. If this were what showed on the gauge it wouldn't be very useful because the needle would be jumping all over. So the response is slowed way down to show about how much gas there is if it were level. The time constant can be very long because your tank really empties only slowly. Damping a variable is used the same way, to rate limit the change to what is really happening. It can be done either by direct rate limiting, that is any change from sample to sample over X is simply counted as X and some average is taken or some sort of averaging is used alone. Time constant or interval is selected with an eye towards how fast what you are interested in really changes.

Regards
cww

 

Curt has given you a pretty good example of where a long filter time is essential to get a stable and appropriate measurement. But consider that the time to empty your gas tank would be several hours of continuous running - so a 1 minute filter time is a very small fraction of the stabilising time of the measurement.

This rule applies to all measurements, the filter time needs to be a short as possible so that noise is reduced but meaningful process change information is not lost. It is unfortunately not uncommon to find excessive process damping on many signals in a misguided attempt to eliminate variability caused by a genuine measurement or instrument installation problem.

Process damping is typically done by applying a first order filter to a signal measurement. The longer the filter time ( or time constant ) the more the signal is damped. However, the wording of your description does not fit this. Specifically the use of the words "sampling interval". It is more likely that in your case the instrument is "over sampling". This is a technique that uses multiple measurements taken faster than the scan time of the PLC connected to the instrument. The instrument filters or averages these over sampled measurements and produces an updated and filtered measurement every PLC scan or Bus update. This is a very effective technique for removing high frequency noise from a signal while not over damping the process measurement. ( effectively the filter time is shorter than the PLC scan internal) without knowing what type of transmitter you have its not possible to give a more clear answer.

The "rule of thumb" is to avoid filtering if at all possible. If your measurement is noisy you need to understand if this is a process or installation issue and fix that. NEVER try to hide installation errors with a filter.

After you have verified those issues then you may apply a small or short filter to remove high frequency variation from the signal. Filter times of more than a few seconds in a typical process plant are a sign that you may be filtering too much.