Troubleshooting PI AF Analysis: Common Pitfalls and How to Overcome Them

PI AF (Asset Framework) Analysis is a powerful feature of the OSIsoft PI System that enables users to perform real-time calculations and derive valuable insights from process data. However, users often encounter issues with analyses that can be puzzling and difficult to troubleshoot. Let’s discuss a common scenario and explore solutions to improve the reliability of your PI AF analyses.

The Scenario

In this case, an error emerges during the evaluation of a PI AF analysis. The Salt_ValueOutput, derived from another analysis, has a correct integer value. However, during evaluation, the analysis throws an error related to invalid operations. Additionally, attributes Salt_Yellow and Salt_Red with a "none" data reference are manually entered values. Surprisingly, the same analysis works correctly in the majority of templates.

Common Causes and Solutions

1. Handling Questionable Values

One reason for errors in PI AF analyses is questionable data values, which the system treats as bad values. Questionable values are flagged due to concerns over data reliability. Within Asset Analytics, these values are often treated the same as bad data, causing calculations to fail.

To address this:

• Utilize the IsSet() function to determine if an attribute is flagged as questionable.
• The BadVal() function can be employed to check if a tag value is not reliable and avoid processing such data.

2. Enum to Number Conversion Errors

Errors converting enums to numbers typically arise when attributes inadvertently become linked with enumeration sets. A common misconception is treating these attributes as numeric when they actually reference a state code or condition.

Solution:

• Use the StateNo('AttributeName') function to extract the integer state code from enumeration attributes.
• Ensure that attributes like Salt_ValueOutput do not inadvertently have enumeration values, especially when expecting numeric data.

3. Digital State Misinterpretations

An analysis error may also occur due to Digital State values such as "NoData" or "Arc Offline" appearing in PI points. These are special conditions that render numerical operations invalid.

Resolution:

• Verify the source tag's status and address Digital State issues that can cause erroneous values.
• Configure analyses to handle or bypass Digital State cases appropriately, ensuring smooth operation.

Mitigating Future Errors

To avoid similar issues in the future:

• Regularly review PI point configurations and quality status values to preemptively handle questionable conditions.
• Incorporate conditional logic within your analyses to manage questionable or bad data gracefully.
• Engage in proactive monitoring and consistent refinement of analysis conditions to align with changes in data patterns and systems.

Understanding and troubleshooting common errors in PI AF analysis ensures robust and reliable analytics that underpin crucial business decisions. By implementing these strategies, users can significantly reduce the frequency of analysis disruptions, enjoying a stable and insightful PI System experience.