From Local Controls to Cloud Analytics: Tags to Bridge OT & IT

It’s sometimes hard to remember in these days of touchscreen human-machine interfaces, but the humble Programmable Logic Controller (PLC) was a significant advance when it started permeating industries back in the ‘70s and ‘80s. PLCs entered a world of bespoke control devices. These were finicky jumbles of diodes, relays, sockets, timing circuits, power supplies, and wiring, or – where electronics were too fragile – their hardier pneumatic logic cousins. These bespoke systems themselves replaced manual control of complex physical systems in the early-to-mid-20th century with the first automated, condition-based response capabilities. Still, PLCs were a major upgrade for standardized industrial operations.

PLCs could be programmed with "coils" (outputs) and "contacts" (inputs). In the more advanced PLCs that became standard in heavy industries, a structured programming approach incorporated coils, contacts, and any other variables, objects, or function blocks as "Tags" in a Tag Database. PLC programmers would declare a Tag and its data type prior to using any Tag. As with computer programming languages, PLC Tags support data types such as floats, strings, ints, and booleans, as well as user-defined data types, much like classes, structs, and objects.

WHAT’S THE IMPLICATION FOR IIOT?

As PLCs have been augmented or supplanted by modern control systems, the terminology persists. Today, every instrumented field device is assigned a unique identifier called a Tag number. For industrial analytics purposes, a "Tag" is synonymous with a sensor value.

The shift from bespoke control devices to PLCs carried over comfortable terminology for operators affected by change (coils, contacts, tags, etc.). Implicitly, this also brought along the associated thinking about data structures. We believe pushing operating technology data into cloud analytics is best accomplished through building on data structures implicit in currently installed control systems.

In cloud data architecture, a "data contract" describes how a client and server exchange data. Many cloud analytics applications are built for a variety of potential use cases, and therefore have a set of default data contracts. These require more complex data types to define data contracts. This very openness and flexibility, however, creates translation barriers. These barriers manifest when ingesting industrial data into a new analytical application. They also create challenges in writing output back into field industrial systems.

TAG STRUCTURES AS THE DATA CONTRACT FOR INDUSTRIAL ANALYTICS

Using Tag structures as the underlying data contract specifically for heavy industrial analytics applications enables not just easier data ingest to cloud-based analytical models. It also allows all further data manipulation – including advanced machine learning model outputs – to be written as Tags and sent back into field systems. This enables seamless communication to and from industrial user interfaces.

Particularly in the context of edge computing applications and other use cases requiring ongoing ingest of time series equipment data, Tag-based data contracts radically simplify the time and effort required to move from field data to analytically-driven insights that can inform operator actions. In analytics applications revolving around equipment monitoring – from assessing health and performance to detecting potential failures – this underlying Tag-based approach significantly reduces the time from initial data collection to tangible business value.