IT systems · Practical guide

AVEVA PI System validation: data flows and GxP use

Collecting process data in the PI System makes them centrally available. Whether they can support a quality decision depends on capture, context and evaluation. Follow the intended use through the historian to the conclusion drawn from the data.

Table of contents

At a glance

How do you validate AVEVA PI System, formerly OSIsoft PI, for GxP use?

Define the intended process, data, calculations and outputs. Trace the relevant path from the source through transfer and the archive to the context model and report. Assess data reduction, timing, quality flags, permissions and historical access. Validate the configured application. The product name does not determine its GAMP category or evidence scope.

Define the PI System based on intended use

The AVEVA PI System is previously known as an OSIsoft product suite under various designations. For validation, the actual scope of implementation is decisive. AVEVA describes the PI Server, among other components, Data Archive and Asset Framework as elements for the storage of time-series data and their associated context. Additional components, data sources, and analytics are captured according to your architecture. A general product page does not constitute a complete description of your installation.

Determine which decision is supported. A energy report may have a different regulatory basis than a batch-specific temperature assessment. The same stored value may be used in both processes. Describe users, data, calculations, output, and required history. This forms the GxP scope. The system boundary must neither be reduced to a server name nor expanded to all technically accessible components without reference to the process.

Sources: [1] [2]

Make the data pathway and primary source visible

Draw the relevant path from the measurement point to the decision point. Capture the source, data acquisition or transmission component, stored signal identity, context mapping, and used analysis method. Describe how information is modified or enhanced at each stage. A clear data flow aids in assigning responsibilities and appropriate validation evidence. A line between two product logos is insufficient for this purpose.

Check the technical origin: Which system, which sensor and which process step belong to the signal? A correctly displayed numerical value can be wrong if it comes from a similar but different measuring point. Complete unit, time reference and quality marks. In the case of a replacement sensor or modified plant structure, the technical identifier can remain stable, although its meaning changes. Such changes belong to the controlled life cycle.

Sources: [2] [1]

Intentionally define data retention and data reduction policies

Evaluate data capture rate, transfer, and relevant compression mechanisms based on the required assertion. The PI vendor description outlines exceptions and compression as technical mechanisms along the data pathway. Specific parameters and components are assessed internally. Vendor recommendations serve as starting points and are not general GMP specifications. A trend must retain the information actually needed by the process.

Review known signal trends featuring short changes, constant phases, outages, and return-to-service. Compare source, recorded events, and displayed representation. A smooth line may mask a data gap or infrequently logged values. A high volume of archived points does not necessarily indicate correct data alignment. Data completeness and technical suitability require clearly defined criteria. The general contribution to historian data explains calculation discrepancies; here, the specific integration into the PI application is assessed.

Sources: [3] [2]

Treat the context model and calculations as a separate scope.

A context model can link time series to equipment, units, or additional metadata. Verify these associations and required calculations within the specific asset framework or application environment. Using a technically correct formula at the wrong equipment attribute yields an incorrect process statement. Therefore, both logic and data linkage must be validated. Templates can standardize repetitive structures, but require evaluation of their actual instances.

Define historical impact. If an attribute mapping is later modified, how can previous evaluations remain interpretable? What calculation and configuration states are required? A current system name may inadvertently mislead an older report. Assess the reconstruction needed for your specific purpose. Not every possible post-hoc calculation needs to be supported; however, the approved usage and its limitations must be clearly documented.

Sources: [1] [2]

Example: Evaluate batch temperature from PI

In this example, the temperature of a container is evaluated during a specific batch. Source T-101 falls under Section A, and the batch runs from a defined start to end time point. The expected profile includes a brief excursion and a data gap. The report must accurately reflect the process rule requirement and clearly identify the absence of evaluation basis. The numbers and rules apply only to this example; they are not universal process limits.

In addition, verify incorrect source assignment to Appendix B, modified time limits, and an invalid quality status. The technical reference is defined prior to execution. A screenshot of the final trend is insufficient. The following matrix links system level and inspection question. Its scope is adjusted to the actual architecture; it does not specify a particular PI-component combination for any regulated use case.

Own help – Example: Evaluate batch temperature from PI
LevelCritical questionReference case
Source Correct measurement point? Distinguishing Appendix A against Appendix B
Transfer Was is the correct value and time successfully transferred? Delay and interruption
Archive Have the required information been obtained? Sharp peak and constant phase
Context Correct batch and equipment? Altered assignment
Calculation Correct rule and limit? Independently calculated expected result
Output Are gaps and quality status detectable? Incomplete evaluation period

Sources: [2]

Control access and relevant changes

Determine who is authorized to read, write, correct, modify configurations, and maintain reports. The actual rights available must be verified at the system in use. A pure dashboard user requires different authorizations than an administrator. Test permitted and denied actions using defined roles. Consider technical identities and associated applications, not only interactive users.

Define the required traceability of relevant interventions. Data modifications, tag configuration, and report maintenance may originate from different protocol sources. Assign a verifiable evidence base to each required assertion. A general Windows event log does not automatically capture the prior and current professional data values. Where necessary history is not sufficiently available, the gap is assessed and the concept is adjusted. A broad manufacturer compliance statement does not substitute for this evaluation.

Sources: [2] [4]

Check failures, buffering, and recovery

Consider interruptions at the source, transmission, archive, and analysis stages separately. A subsequent arriving value must have the correct event context. Verify whether duplicate values, gaps, or inconsistent time sequences arise and how the system detects them. The specific buffering and recovery mechanism depends on the component in use. Do not assume automatic complete restoration solely based on a product name.

For quality assurance, required data and configurations are reviewed jointly. A time-series archive lacking necessary context may become difficult to interpret later. Verify restoration using a subject-matter reference case, not merely the initiation of a service. During long-term archiving or system replacement, the historical research pathway is evaluated. It must retain values, time, unit, and required configuration references to the extent available for release.

Sources: [2]

Traceability and ongoing operation justified

The closure encompasses intended use, architecture, data controls, calculations, test results, and operating rules. Define boundaries and responsibilities. The decision applies to specific usage. Validated temperature analysis does not automatically validate any additional parameter or new dashboard application. Furthermore, additional data sources may introduce different failure modes and require evaluation.

Manage reference cases and relevant configurations during changes. New sensors, tag assignments, compression parameters, calculations, or outputs may influence the conclusions. A periodic review evaluates actual usage and known events on a risk-based basis. Robust PI validation provides a clear and verifiable pathway from source to decision, establishing a credible boundary between potential data views and actually substantiated process information.

Sources: [2]

Project experience and collaboration

Trace the reported process result back to its source

A trend can look normal even when timestamps or quality flags are missing. With your process and data owners, we follow a specific analysis from source through transfer and archive to the report. Intended use determines the evidence needed for data reduction, calculations and context. We select controlled comparison cases and test interruption and restart. The validation scope and conclusions apply to your configured use of the PI System.

Frequently asked questions about AVEVA PI System / OSIsoft PI

Is the AVEVA PI System the same as OSIsoft PI?

OSIsoft PI was the previous name for the current product family now operated under the AVEVA PI System. However, the names do not fully convey the scope of a specific installation. For validation assessment, the actual components, versions, and functions being used are considered; older and current versions may differ.

Is the PI System automatically classified as GAMP Category 4?

A generic category based solely on the product name is not reliable. Assess components, configurations, calculations, and in-house developments. Classification supports the traceability strategy but does not replace consideration of the intended use and actual risks within the process.

Does Asset Framework fall within the scope of validation?

Insofar as relevant assignments, metadata or calculations influence the GxP statement, this scope belongs to the evaluation. Check the correct connection of signal, installation and process. Correct archive storage alone does not automatically prove the suitability of a context model based on it.

Sources: [1]

Must each visualization be individually fully verified?

Evaluate the specific statement and differences. Similar controlled templates may enable representative evidence. New calculations, filters, or data sources require additional consideration. A dashboard that provides only information may be assessed differently than an output used for a quality-relevant decision.

Is a backup of the archived files available?

Not automatically. Verify which configurations, mappings, and additional components are required for the required restoration. The evidence includes use in the process following the restore. The startup of a service or presence of files does not confirm the complete historical context.

Why must the PI data pathway and evaluation be jointly reviewed?

Proper archiving can lead to an inappropriate conclusion due to incorrect source assignment, calculation, or representation. Therefore, evaluate the entire relevant pathway up to the decision point. The specific intended use determines which components and relationships must be validated together.

Sources and review date

As of October 14, 2026. Regulatory classification and manufacturer responsibilities are separately documented. Examples, test matrices, and templates are standalone reference tools; they are customized for actual usage purposes.

  1. AVEVA: PI Server, Data Archive, and Asset FrameworkCurrent supplier overview for time-series storage, metadata, and context; reviewed on 14.10.2026. Specific license and version scope remains project-specific.
  2. European Commission: EU GMP Annex 11, Computerised Systems (2011)Current version as per EudraLex inventory as of 14.10.2026. Application, risks, operation, and data lifecycle; specific testing proposals are derived independently.
  3. AVEVA: Exception, Compression and their Impacts on PI System PerformanceOfficial manufacturer statement regarding Exception and Compression as data filters. Parameters and specific components must still be verified on a version- and application-specific basis; reviewed on 14.10.2026.
  4. PIC/S: PI 041-1, Good Practices for Data Management and Integrity (2021)Inspection Guidance on Data Integrity; in particular electronic data, transmission, access, and review.

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

How we can support you

Daniel Herrmann Consulting combines professional assessment with hands-on work on the specific system. During the first meeting, we clarify your intended use, existing documentation, and the next most appropriate action step.

  • Computer system validation →

    Bring your GxP systems into validated operation, from laboratory software, MES and eQMS to ERP. We define the scope, plan risk-based tests using GAMP 5 and hand the evidence to QA for release.

Send a project enquiry Customer testimonials →