Emerson’s Jonas Berge shares his thoughts on a recent ARC Advisory Group article, Unifying Device Integration Across the Automation Landscape. The article highlights the author’s recent conversation with the FieldComm Group’s leadership team. It:

…served as a reminder of the essential work of creating a unified, interoperable, and consistently defined data flow from the device level.

Here are Jonas’ thoughts that he posted in LinkedIn after reading the article.

Interesting post from ARC Advisory Group Inc. – my additional thoughts:

#AI is often likened to a brain, and in this context, we must also remember the sensory neurons, the muscles, and the nervous system carrying the signals between them. AI without sensors and actuators is like a brain without senses and muscles. Before you “can connect the dots”, you must first “collect the dots”, and then you must “correct the dots” – take action to make a difference. The ‘nervous system’ connecting these #sensors and #actuators is industrial digital #networks.

For AI to be successful, the recommendation is for plants to invest more in sensors to take the place of mechanical gauges and portable testers, and in actuators on control valves and isolation valves to enable valve remote control.

Think about it, phones today have more sensors than the first-generation smart phones, cars have more sensors today than your parent’s car, planes have more sensors too. Cars motorize sliding passenger doors, boot/trunk, rearview mirrors, power windows, seat adjustment, sunroof, fuel door, and external door handles slide out. The driver can operate the car without getting out of the seat: operators should be able to run the plant without going to the field.

Most AI companies will tell you “you already have all the data – you just need to copy all of it to a single platform, a ‘single source of truth’, and then train your AI model”. But you don’t have all the data, and not in real-time, – and therefore rely on manual data collection. This data collection must be automated by sensors. You have a lot of PROCESS data because you have many process sensors. But you don’t have enough EQUIPMENT data because equipment data is collected manually, infrequently, or not at all. These are missing measurements. You can’t predict EQUIPMENT problems using PROCESS data. So plants need many more sensors as the real-time data foundation for AI to stand on.

PA-DIM making device data available through OPC-UA in a standard format will open new possibilities. I’d like to see PA-DIM even without using device files, e.g. directly from HART-IP to PA-DIM, at least for ‘parameters’.

I see PA-DIM as a unifying information model mainly for Process Automation protocols like HART, FF, and PROFI-PA. The “PA” in “PA-DIM” means Process Automation and the information model is very much built to accommodate the NAMUR NE131 ‘core parameters’ defined for common PA device types. It could extend to devices like variable speed drives found in both process and discrete factory automation, but may not be a good fit for pure discrete factory automation devices like robot arms etc.

Data from this intelligent field is a key part of Emerson’s Boundless Automation℠ vision for an industrial automation platform that integrates operations across the intelligent field, edge, and cloud to empower enterprise-wide visibility, optimization, and autonomous operations.