Supercharging PI System for Data Sharing & Advanced Analytics: Part One

Sean Ely
Sean Ely

In the age of IoT, effective data management, including accessibility, is critical. The proliferation of connected sensors generates a massive volume of timestamped data that needs to be shared, centralized, and analyzed. This three-part blog series reviews the value of collaborating on time series data and the challenges associated with it. Specifically, it explores the capabilities and limitations of PI System, a popular data historian used throughout manufacturing, and illustrates how TDengine can be used to extend this system to enable effective data centralizing and sharing.

For more on extending PI System’s capabilities for data sharing and advanced analytics, download the full whitepaper.

Managing & Sharing Data in the IoT Era

The Proliferation of IoT Data

Enabled by the internet of things (IoT) and the digitization of manufacturing, the industrial internet of things (IIoT) is the greatest advancement in manufacturing since the electrification of factories in the early 20th century.

In practice, IIoT applies a connected network of devices and sensors that automatically monitor equipment and processes in real-time, vastly improving efficiency, reducing downtime, and supporting advanced business intelligence. Naturally, this approach generates a massive volume of data, virtually all of which is timestamped, that needs to be collected and analyzed.

With the number of connected sensors increasing exponentially, it’s critical for businesses to implement effective strategies, processes, and technology to ingest and manage this data at scale.

Global Active IoT Connections (in Billions)
Although growth slowed during the height of the pandemic, the number of IoT connections is expected to double every three years.
Source: IoT Analytics Research, State of IoT 2022. May 2022

Adapting to IIOT: The Value of Centralization & Sharing

Adapting to this industrial revolution requires introducing new technologies, systems, and approaches.

For example, with IIoT bridging the gap between physical devices and digital applications, legacy embedded systems are being replaced in smart factories by cyber-physical systems (CPS) which combine computing, communication, and control capabilities with physical procedures and elements.

To maximize the effectiveness of these systems, and the business benefits they can provide, data from individual sites must be centralized and shared. This enables an organization to benefit from a global view of an entire network of connected sensors and devices, and develop insights that lead to company-wide efficiencies. Operating in silos doesn’t cut it when it comes to IIoT.

Benefits of Centralizing IIoT data:

  • Unified Worldview: Instead of analyzing fragmented, siloed data, centralizing allows users to view a single dashboard.Users can review global operations while retaining the ability to zero in on individual sites.
  • Seamless Benchmarking: By pulling together data from multiple sites, users have greater context of the performance and outliers in operational data.It’s easier for users to see which sites need support, and allocate staffing and other resources accordingly.
  • Controlled Sharing: Ensure datasets or limited segments can be shared throughout the organization with fine-tuned privileges.Data segments can be shared with external partners, vendors or regulators.

Centralization & Sharing Challenges

Many large manufacturing companies, especially in mature industries such as utilities and automotive, have developed or acquired unique data collection and operation systems within many of their sites.

Aveva PI System, a data historian that collects and stores time-series data, has been widely adopted in manufacturing to monitor and report on individual sites. To ensure adequate reliability and low latency, a PI instance is deployed in each site, often physically close to the data it’s collecting. This architecture often requires a significant investment in software, hardware, and time for deployment and administration.

Although PI System excels at collecting data for individual sites, it has considerable limitations when it comes to centralizing data from multiple locations. Having made such a significant investment in this architecture however, ripping and replacing the solution may not be feasible, despite the benefits of a centralized approach. This issue is compounded when businesses use varying data historians across sites or protocols such as MQTT.

Vestas North America: Company Sites & Turbine Installations
Like many IIoT companies, Vestas has to centralize data across tens of thousands of devices and multiple factories – in North America alone.

Centralizing & Sharing Data with PI System

PI System for Time Series Analytics

The PI System, launched in 1980 by OSIsoft, was developed in response to the increased automation of industrial processes, where devices gather large amounts of data that can be valuable when analyzed and processed.

In addition to timestamped data, legacy databases struggled to collect and store vast amounts of metadata including time, location, device type, etc. Instead, the PI System is specifically designed to handle time series data by facilitating quick data ingestion and querying, while maintaining the full context of each data point, which is essential for developing meaningful insights.

Centralization and Sharing Limitations

Although PI System is well-suited for time series data, it has significant limitations when it comes to centralizing data from multiple sources. It usually requires either complicated IT implementations or exporting and categorizing vast amounts of information manually, which prohibits any form of real-time analytics.

PI-PI Data Centralization & Sharing
Moving data from individual control networks into a business network is often achieved with the PI to PI Interface, with sharing to external users via Aveva Data Hub.

Unfortunately, PI System also lacks easy-to-use tools for sharing data, such as creating specific privileges for access to a segment or filtered data. This creates difficulties when businesses need to share data, including sensitive information, with customers, partners, and regulators.

Visit TDengine Cloud – Data Sharing to see how sharing time series data can be as easy as sharing a Google Doc.

In some cases, the limitations of PI System have resulted in further fragmentation of a business’s data stack, as rather than finding an agile solution that supplements PI System’s capabilities, additional solutions are brought in to manage each scenario that PI cannot handle.

Example: High Availability Pi System Architecture Across Two Data Centers
As this diagram shows, even a carefully planned system for PI integration requires complex architecture, which is expensive to run and maintain.

Visit TDengine for PI System for information on how to easily centralize and share your PI System data.

  • Sean Ely
    Sean Ely

    Sean Ely previously worked as Head of Product at TDengine.