By Leslie Shields, Lead Systems Engineer at ITI Group.
PI Asset Framework (PI AF) is a powerful part of the AVEVA PI System, which allows your operational data to be contextualised in a more meaningful and consistent way. This means that your data processing and analysis can be applied more consistently, and therefore more usefully.
At the heart of PI AF is the concept of Templatisation: allowing you to create wonderfully standardised models and calculations. Build it once, scale it out, maximise your ROI. When done effectively, we have seen customers achieve seven and eight figure returns on investment in PI AF within a few years – but getting to that point can be challenging.
At first, building an asset hierarchy seems like a simple if somewhat time-consuming task. We are often told that the template for each asset can be built for one unit and then rolled out to cover others, as the process is the same across all assets. However, once the build is started it often becomes clear that the complexities and differences of actual operational assets means that this assumption is very much misguided. This often leads to “regret work” to manage the differences and maintain the templated approach to the asset hierarchy.
The potential pitfalls and difficulties can however be managed and minimised. Thus, creating a successful Asset Hierarchy is possible if sufficient planning and design is implemented prior to starting the AF build.
By considering the following steps, tips and traps, you can build an effective asset hierarchy in AF that enhances data accessibility and supports robust, performant analytics for your organisation’s objectives…
Getting Started
1. Define Your Objectives:
Clarify the purpose of your asset hierarchy and analytics. Are you focusing on operational efficiency, maintenance tracking, or data analysis? This will guide your structure. Note that it is sometimes better to have several smaller focused structures than one huge (and potentially unfinishable) all-encompassing structure even if there is some cross over of data between hierarchies.
2. Gather Your Data Upfront:
Collect relevant data about your assets, including types, functions, and relationships, as well as key data points you want to record and analyse. The more data you can gather upfront will help reduce regret work to fix assumptions that the process / instrumentation will be consistent across all assets.
3. Choose a Hierarchical Structure:
Decide on a structure that suits your requirements, perhaps functional, geographical, or process-based hierarchies. Choosing the right hierarchy type enhances clarity, scalability and usability. (Examples of Hierarchical Structures and the benefits of each are given at the end of this article.)
4. Focus Your Analytics:
Tailoring analytics to align with the chosen hierarchy allows for more relevant insights and actionable data.
Top Tips for Successful Implementation
1. Engage Your Stakeholders
Involve key users from relevant departments to gather insights and foster buy-in, enhancing the design and ensuring it meets practical needs. Which specific tasks and activities is your solution supporting, and what do they need?
2. Use Templates and Libraries
AF templates speed up development. To minimise regret work, don’t rely on a single example to base the template on. Assess what is common and what the differences are across similar assets, so that you can create a base template with common attributes and use derived templates to apply differences.
3. Use Categories
Utilise categories to organise the data in the AF database and allow for easier search and identification of data.
4. Documentation
Maintain thorough documentation of your asset hierarchy and analytics processes, thus aiding training and any future updates.
5. Iterate and Improve
Don’t shoot for the moon. Start with a pilot project and refine the hierarchy and analytics based on feedback. Continuous improvement will help maintain relevance as the hierarchy grows.
6. Training and Support
Provide training for users on both the hierarchy and AF to maximize effectiveness and user engagement. Engage trained users regularly to ensure they are successful and have a PI expert review their products before deploying.
Don’t Forget!
Simplicity vs. Complexity
Strike a balance between a user-friendly hierarchy and the complexity needed to capture detailed relationships and data flows.
Consistency
Use consistent naming conventions and classifications to avoid confusion, aiding understanding across the organisation.
Performance
Avoid overly deep structures and ensure efficient data retrieval by organizing data as logically as possible. Overly deep structures can cause potential issues when searching for attributes or when referencing attributes in formulae or analytics.
Integration
If you have a requirement to integrate your hierarchy into existing systems then consider reusing existing structures! Note that this can come with some downsides with divergence over time due to changes made in the source structure. Avoid creating technical debt for yourself wherever possible.
Avoid These Common Pitfalls:
Data Quality: Ensure data feeding into your analytics is accurate and up to date. If your data foundations are weak, consider investing here – there’s no point in building your own private jet if you only have vegetable oil to run it on.
Error Handling: Ensure good error handling is implemented. Check data quality of input values, handle error states gracefully and use specific outputs to indicate what the issue is.
Aggregation Levels: Design analytics around appropriate aggregation levels to balance detail with performance. Consider using summary data for high-level insights and detailed data for specific analyses.
Leverage PI Tags: Use PI tags where possible for analytic results as this optimizes performance and minimizes data retrieval requests.
PI Tag Attributes: Agree on what the PI Tag attribute defaults should be for PI Tags created within AF – don’t forget default security, point source(s) and compression!
Scheduling: Consider triggers carefully for any analytics. It is easy to set all analytics to be naturally scheduled on any input (default), but this has a detrimental impact on the performance of the analytics service and causes runaway performance issues.
Backfilling Historic Data: If there is a requirement to backfill historic data it’s important for analytics to be backfilled in order to ensure that calculations run with correct historic data. Consider creating logically numbered analytic names or descriptions to aid in the backfilling order.
Analytic Service Settings: Assess the performance of the analytics server and optimise Analysis Service configuration settings to better reflect the requirements of the analyses that are running. If you don’t know how to do this, our team are happy to help.
AF Attribute Path: When using PI Tags as outputs for analytics you can store the AF path of the attribute in the PI Tag’s extended descriptor (exdesc) tag attribute. This allows easier identification of the element path for the output PI Tag.
Event Frames: Consider carefully what Event Frames are required and the trigger status for these. Whilst Event Frames in themselves are not particularly burdensome on the analytics service, having a high number of Event Frames can drastically affect the backfilling performance of the service, particularly at service start-up.
Examples of Hierarchical Structures
Functional Hierarchy
Assets are organised by function, such as “Production,” “Maintenance,” “Quality Control,” and “Logistics.” Each functional area then includes subcategories for specific equipment and processes (e.g., under Production: mixers, conveyors, and packaging machines).
Benefits: This enables teams to focus on specific functional areas, facilitating collaboration and ensuring that all related assets are grouped logically. Makes it easier to set and measure KPIs by function.
Geographical Hierarchy
Assets are organised by location, starting from the corporate level down to regions, facilities, and finally to individual pieces of equipment (e.g., Corporate > Region > Facility > Equipment).
Benefits: This structure simplifies reporting and compliance tracking across multiple locations, making it easier to analyse performance on a regional basis.
Process-Based Hierarchy
Organised around the stages of a specific process, such as “Raw Material Handling,” “Processing,” “Packaging,” and “Distribution.” Each stage includes relevant equipment and systems.
Benefits: This clear flow representation can help identify bottlenecks and inefficiencies in the overall process.
Asset Type Hierarchy
Assets categorized by type, such as “Electrical,” “Mechanical,” “Instrumentation,” and “Utilities.” Each category can have further subdivisions (e.g., under Electrical: generators, transformers, circuit breakers).
Benefits: This hierarchy can help support maintenance and inventory management by grouping similar assets. Assets can be geared around specific tasks and workflows, creating clearer value cases. This adds risks of duplication where many a single asset has diverging versions, which needs to be managed. These prototype hierarchies are not strictly required to be so fixed: we often find that many structures require implementation as a hybrid, so you may have a mix of geographical and asset type hierarchies where asset substructures are created under a geographical base structure.
Get in touch
If you’re looking for a Partner who can help you make the most of your AVEVA PI System, ITI Group is an AVEVA Endorsed Operations Systems Integrator with more Accredited PI System Engineers than any other partner in the UK.
Send us a message: our team are always happy to help
"*" indicates required fields