This blog originally posted on Jim Gavigan's LinkedIn page here.
I recently read an ISA article on the Industrial Internet of Things (IIoT) and this is the second installment to a 4 part series of commentary on the article. There is a lot of great information in the article, and I am giving a "view from the ground" on several highlights from it.
Regarding standardization, the article states:
"With most of the technological components already available, concerns over cybersecurity, technology standardization, and intellectual property ownership remain the most prominent potential obstacles."
"Most industrial organizations purchase machinery from a variety of suppliers. In an IIoT context, this could ultimately result in a wide variety of vendor-specific interfaces being installed, resulting in added cost to maintain and update multiple installations."
"Some formal and de facto industry standards are already available to consider as part of an IIoT strategy. These include industrial Ethernet protocols, IEEE wireless standards, and software interfaces like OPC-UA, but there will be further developments as the IIoT progresses. Whichever is right for your company, the important point is to begin marshalling a non-vendor-specific device interface strategy in anticipation of an increasing flood of connected devices."
"In parallel with assembling a business case for IIoT adoption, industrial organizations should review standardization of network and software interfaces for connected devices as well as security and access control practices. This approach can save numerous headaches in the long run by reducing the number of operational variables and reducing or eliminating the need to maintain and update disparate technologies."
Standardization may be the single biggest challenge end users face in adopting the IIoT. Throughout my career, I have had numerous customers who were still running interfaces from their data collection systems to ancient DCS systems on VAX based machines. These same customers also had other control systems that were the "latest and greatest" in the marketplace, and they were communicating to these same data collection systems. Literally, they had hundreds of interfaces to control systems with vintages ranging from 15-20 years ago to right now that they had to manage. This challenge won't get better with the explosion of new devices and control systems coming with the IIoT.
While comparing notes with my peers, these customers' experiences are similar to many other customers across a wide range of industries. IIoT implementation strategies must acknowledge the unrealistic task for most organizations to standardize and update all the disparate technologies at the device and control layer. Many organizations today have come to pass through a series of acquisitions, and many of these acquisitions had different standards for control systems, MES/MOM systems, and ERP systems. All need to have some level of standardization, but there is only so much time, resources, and money to make this practical. ERP systems almost universally get integrated when one company acquires another for accounting, financial, and HR purposes. Rarely do control systems and MES/MOM systems absolutely require integration, which greatly impacts facilities as there is usually a lot of tribal knowledge built into or inherent in these systems. As the inevitable transitions of people within and/or out of organizations occur, the original functionality of these systems can be compromised or even lost.
Having standard interface strategies to the plant floor is easier today since most devices connect via OPC, but there are still a ton of relatively closed control systems running in industry today with proprietary networks or protocols that will need a specialized interface to connect to them. There are also places where high speed data collection is needed, and OPC isn't the best choice for higher speed data collection. My advice would be to standardize interfaces where possible (with OPC for instance), but there will almost always be outliers - you can't always retrofit them, and sometimes you have to leverage newer, or just other, technologies where appropriate.
Having disparate brands of control systems or various vintages of control systems is also not unique to one or two industries. This applies to every industry in manufacturing, in power generation and distribution, in building automation systems at large campuses like hospitals and universities, and almost everywhere in between. Standardization at the control system layer isn't always prudent or practical, especially not just for IIoT's sake.
So, what is a company to do? Is all lost and hopeless with regards to standardization? Definitely not! So, here are some suggestions:
Find a layer where standardization DOES make sense and is easier/less costly to implement - typically, this is the manufacturing intelligence layer (one could also use the term operational intelligence layer as well, depending on the industry). This is a layer of information typically above the HMI and SCADA layer, but underneath the ERP layer and is usually separated from a specialized MES or MOM system. See the figure below for clarification (replace building automation systems in place of PLC's or DCS systems if you are managing a large campus):
Regardless of what your company has or chooses to use at this layer, you want a platform that can bring in the data from any vintage and/or brand of control system and be able to deliver it to other systems like ERP (for financial reporting, raw material consumption, maintenance information, etc.) or MES/MOM systems (product genealogy, WIP reporting, schedule adherence, etc.). The key is to have a true platform that does the following:
Integrates data from disparate systems (control systems and other databases)
Adapts easily as plant floor systems are changed or are updated
Allows for enterprise deployment/connectivity
Adds context to the data (asset/event context, metadata, etc.)
Is extensible and allows the user to build applications on top of it if required
Integrates easily to other systems like ERP and Maintenance Management Systems via standard technologies (Web Services, ODBC, OLEDB, and the like)
It is infinitely easier to keep a manufacturing intelligence/operational intelligence system up to date and standardized than a myriad of sensors and control systems. Plus, this type of system doesn't care as much about what is attached to it - it tends to be more agnostic as to what is above it, below it, or beside it. MES and some MOM systems tend to have more specialized functions, and ERP systems tend to not have good control system interfaces natively, so neither are great choices for a point of standardization as it relates to the IIoT.