Aligning PI System Data to 15-Minute Boundaries Using AF: Strategies and Solutions

In industrial data systems, aligning process data to consistent intervals (such as 15-minute boundaries) for reporting, analysis, and regulatory needs is a frequent requirement. The OSIsoft PI System and its Asset Framework (AF) provide powerful tools for data organization, but implementing precise time-alignment—especially with custom tolerances—can pose challenges. Below, we’ll discuss approaches to tackle this alignment in AF, including interpolation windows, marking gaps, and the role of custom Data References.

The Use Case: 15-Minute Data Alignment with Tolerance

Imagine an AF solution that processes and historizes calculations, but the customer mandates that both input and calculation results must fall exactly on quarter-hour marks (e.g., 12:00, 12:15, 12:30 …), with a 1-minute tolerance. Data outside this window should be marked as gaps, and only values within ±1 minute of the interval (or interpolated to it) are considered valid.

Common Approaches

1. PI ACE (Advanced Computing Engine)

PI ACE provides robust calculation capabilities and can be programmed to scan source data, align it to exact boundaries, and filter or interpolate per custom rules. However, using ACE often means moving business logic out of AF attributes/calculations and into separate code modules—contrary to modern best practices favoring AF-centric solutions for transparency and maintenance.

Pros:

• High flexibility (custom C# or VB code)
• Direct access to raw PI data and timestamps

Cons:

• Less integration with AF templates/calculations
• Additional deployment and management overhead

2. AFPI Data Reference Retrieval Options

The AF PI Point Data Reference offers basic retrieval options, including recorded and interpolated values. While you can request interpolated values at specific times (e.g., every 15 min via the AF SDK’s InterpolatedValues), custom logic around a tolerance window or marking gaps is not available out-of-the-box.

Limitation: Interpolated retrieval will always generate a value, even if a real reading is far from the target timestamp. There's no native setting to reject/interpolate only within a specified skew (like ±1 minute), or to mark when no real data exists in the window.

3. Rolling Your Own: Custom AF Data Reference

For advanced time-alignment requirements, a custom AF Data Reference (DR) can bridge the gap. Here’s how it works:

• The DR wraps an existing attribute (e.g., a PI Point or calculation).
• Configurable parameters define the output interval (e.g., every 15 min), a skew/tolerance window (e.g., ±1 min), and the data source.
• For each interval, the DR returns:
  • The actual value if a source reading falls within the window.
  • An interpolated value if calculated and allowed.
  • A configurable marker (e.g., "NO_DATA") if no valid data is available.

Tips for Implementation

• Override GetValue and GetValues in your DR class.
• Use Data.RecordedValue with AFRetrievalMode.AtOrBefore/After to find the closest point.
• Use Data.InterpolatedValue for calculating values in the window.

Example ConfigString:

<AF Attribute>;Function=TimeFilter;Period=15;MaxTimeSkew=1;MaxValues=1000

Considerations

• Custom DRs require C# coding and SDK familiarity.
• As of AF SDK 2.5+, use the OSIsoft.AF.Data namespace for rich data queries.
• For earlier SDKs, you’ll need PI SDK references.

4. AF Attribute Historizer

Some users have implemented custom wrappers or services that periodically query, calculate, and persist values back into AF attributes or PI tags on a set interval.

Caveat: Like ACE, this can lead to duplication of logic and non-centralized configuration.

Example Scenario Breakdown

Wrapping Up: What’s the Best Way?

• For pure AF-centric and maintainable solutions, a custom Data Reference is the recommended path.
• If project constraints dictate, use ACE or a separate historization utility.
• Always weigh maintainability and transparency—custom DRs integrate tightly with AF templates and keep configuration visible within AF.

Resources

• OSIsoft Learning—Custom Data Reference Development
• PI Square—Custom AF Data Reference Examples

Have you built your own time-alignment Data Reference, or faced similar alignment challenges in PI AF? Share your strategies in the comments!