Key highlights:
- The industrial data fabric concept helps unify disparate OT and IT data sources, allowing process engineers to access contextualized insights from across the plant and enterprise systems.
- By integrating with analytics tools like Aspen Mtell, a data fabric enables smarter maintenance strategies.
One way that developers and users make sense of formerly separate networks coming together in common areas is by discerning patterns in them, which leads to describing them as a mesh or fabric.
“An industrial data fabric takes raw data from a range of sources, including OT device like sensors and historians, and enterprise-level business systems like ERP. It then transforms this data into valuable, previously unavailable insights, making them universally accessible,” says Dwaine Plauche, senior manager of product marketing for industrial data fabric at Emerson’s Aspen Technology business. “This data can be used to create value with help from cloud computing and artificial intelligence (AI) either onsite or elsewhere.”
Most traditional data management strategies fall short of providing a holistic solution, mainly due to the difficulties of connecting many disparate systems. Plauche emphasizes that data, particularly OT data, is rapidly escalating in complexity, outpacing the capabilities of current organizing solutions. The data fabric concept collects information—and not just time-series data—from historians, DCSs and other sources, and aggregates and contextualizes it, so individuals and analytical programs can make use of it.
“This is a bigger problem than many people realize,” explains Plauche. “Existing data often can’t be trusted because it’s not clean, it’s not structured, it isn’t contextualized, and maybe it’s not even collected or available yet.”
To effectively address these challenges, a robust data fabric must possess several key attributes:
- Easy scalability,
- Vendor-agnostic and can connect to all available data sources,
- Data type- and version-agnostic,
- Distributable to all seven layers of the Purdue model for industrial control systems (ICS),
- Easily upgradable, and
- Reliable and secure.
“If a data fabric has these attributes, it will provide an infrastructure that can grow with increasing data generation, and help enterprises future-proof their applications,” adds Plauche. “Information production has been outstripping our ability to leverage it. A data fabric can close that gap, so information can be used for remote monitoring, prescriptive maintenance, making more complex models and other tasks.”
Get your subscription to Control's tri-weekly newsletter.
For instance, the AspenTech’s Inmation industrial data fabric software can be used in conjunction with its Aspen Mtell prescriptive and predictive maintenance software to generate more and better intelligence. Aspen Mtell is an example of a data-hungry application, so integrating with Inmation can build on its initial results.
“We have a midstream, oil-and-gas customer in Canada that does a lot of pipeline measurements, but the limited data it captured was going into a big data swamp that the company couldn’t maintain or scale to meet their requirements,” says Plauche. “Inmation’s architecture was able to aggregate and contextualize more measurement data—46 million rows of information generated daily—and presented it in dashboards that can be accessed quickly and easily.”
Plauche adds it’s crucial for a data fabric to be agnostic when it comes to suppliers and types of information because this is what allows it to access content in legacy equipment and brownfield facilities without disturbing existing operations. “Users can leave their old, cobbled-together historians in place because Inmation’s agnostic architecture makes their information accessible,” adds Plauche. “Inmation can access all kinds of data sources because it has hundreds of connectors, including web-based APIs, which let users prioritize the sources they want, and grow to include others. These types of connections could be made before, but it’s easier now because Inmation gives companies an IT compliant infrastructure that’s secure, instead of a traditioanl OT-based infrastructure that was usually walled off previously.
“In the past, establishing an industrial network required setting up historians, time-series data, batch devices and software, events and alarms, and other components and their information. Today, a data fabric can accommodate all of these, plus video and other broad and deep sources of data, turning them into value that couldn’t be captured before, and even relaying information and instructions back to the field.”
