Pablo Benvenuto

PI 101, Module 10: Compression Algorithm

Discussion created by Pablo Benvenuto Champion on Oct 2, 2012

The Compression Algorithm

The PI Server uses a Compression Algorithm to determine what events need to be saved in the PI Server archives, or what is meaningful.

Meaningful: The data required to accurately reproduce the original data from the data source.

But isn’t everything meaningful? Not necessarily. For example, consider the following illustration. Which values would you require to accurately represent the movement of the value at a later date?


In the above illustration, the following values will be sent to the PI Server: A, E and F

Effects of Exception and Compression on displayed data

Often you will experience a situation where one of the following conditions will be true:

1. You are watching a trend and you can see many values, but when the trend is refreshed most of them disappear.
2. You query for historical data in Microsoft Excel. You are sure the device is being scanned every ten seconds, yet you see an uneven recording of data (what appear to be gaps in timestamps).


This is completely normal behavior and what you are witnessing is the results of Exception and Compression being applied.

Default Values for Exception and Compression

The default values for exception and compression are as follows:

ExcDevPercent = 1 (% of span);
ExcMax = 600 seconds (10 minutes);
CompDevPercent = 2 (% of span);
CompMax = 28800 seconds (8 hours);
Zero = 0;
Span = 100.

Why are the default values important to me? Because a deadband that is too wide will filter too much of your data and a deadband that is too narrow may give you a lot of unnecessary data.

On the other hand there are instances where you would want to capture everything you collect with no Exception or Compression. You might be performing a calculation and want to capture every result, or you may have regulatory requirements that require you to store every reading. Every situation is unique and you are the best qualified to make these decisions.

Review Questions


Select between five and ten tags associated with processes or equipment that you oversee or are familiar with. Search for the tags and inspect the ExcDevPercent and CompDevPercent. Find tags where defaults values are being used in these attributes. Are the default values good selections for these tag attributes? Why? Feel free to share your findings in this thread.

Congratulations! You have completed the PI System Basics Workshop. Nicely done. You can go back to the Workshop Outline.


Watch these two videos for additional comments.