The volume of data being generated, captured and processed daily—estimated to be approximately 400 to 500 exabytes (0.4 to 0.5 zettabytes)—continues to grow by leaps and bounds. Machine-to-machine (M2M) connections are responsible for nearly half the data sources and between 25–35% of the data proper. Roughly 10–15%, or nearly half, of the M2M data is from industrial (process and manufacturing) sector, which makes sense when you think about the number of data sources in a typical operation and their cyclic update rates.

However, generating data, versus making use of that data are two different issues, and it is being able to use that data that provides value.

One thing that has consistently been a challenge in converting data to knowledge and value is being able to interpret and use it, preferably with minimal human intervention. AI tools will help with the human intervention problem, but they must have the data presented to them in usable format. 

This is simply the latest permutation of the dreaded third-party interface issue that’s been around for as long as I have been in industry. Fortunately, with the migration to commercial off-the-shelf technology (COTS) and the adoption of standard industry protocols, the data integration within the control environment is easier. Historically, industry was reliant on system integrators (SIs) to address these problems for them, then main automation contractors (MACs) came along to take responsibility for the full spectrum. The introduction of COTS has created other challenges such as obsolescence management cybersecurity so the MACs and SIs still have a lot of work, including data management into the IT realm.

Industry protocols are like languages, so if all the systems use the same protocol, the data should be available as part of the device configuration. As we know, languages have dialects (or accents in spoken communications) that make it difficult to understand each other at times. When a manufacturer adds extensions to the protocol, which is often allowed, they are in effect making their implementation of the protocol a dialect. Because machines are unable to understand dialects potentially more difficult to understand.

Get your subscription to Control's tri-weekly newsletter.

The process industry is fortunate to have several widely accepted protocols, from the traditional Modbus, to the now more common OPC and MQTT approach. Likewise, the power industry is supported by DNP and IEC 61850. Other industries are not as fortunate, with a hodge podge of protocols, many still proprietary. 

While at the APTA TRANSform & Expo in early October, I got to see this firsthand. A large bus manufacturer stated they often have 15 different systems to connect on every bus they manufacture. Likewise, I saw that bus people, train people and power people all have different interpretations of SCADA. One reason for this may be because other industries have so many different players vs process and power with several large global companies to push the protocol concept to reduce their integration pains.

Protocols are typically only within the domain for which they were developed, but the real value resides in the supplemental data beyond process variables that need to be transferred to the IT domain for asset management and maintenance. The transition from the real time domains is still a work in progress; however, the need for AI systems to have consistent data is likely to be the driver for a single language across systems. 

If pending regulations such as Digital Product Passport in the EU, and GHG reporting requirements do not drive the common language, or ontology at least, the economic value and potential sustainability objectives to be realized from increased productivity and throughput possible by optimizing not only the life of your equipment but also facility capacity along with potentially one or a consortia of the big AI companies will be impetus enough. Good news is that when they are ready to make this happen it will be those of us familiar with how to make ontologies and protocols that will be able to provide valuable insights possible to make it happen.