The Current Situation
Now that the power of AI has taken a big step forward for consumers to take advantage of it in their daily lives, the need and desire to use the same tools at work has risen exponentially. The question now is how to do you bridge the various worlds of “digital signals”, MES process data, and ERP business data to bring about a wholistic data management layer for business. (Please note that "Digital Signals" includes traditional Automation/OT systems and also the newer IoT approach, as for our purposes here they are the same).
A lot of businesses over the years understand the need to have integration between their ERP and MES layers. There is a clear understanding in the value of having an integrated MES and the operational historian layer. Where I think the limitation has been in the past, is having all three of these systems integrated together and keeping them aligned over time. The value it has to the business to take on such a honorable project is clear, but ultimately can be costly to manage these three systems individually and also the integration of them individually. The scale of coordination across all locations to manage teams of people with very different objectives and goals in mind (IT ERP, IT MES, and OT Historian) to keep up with the fast-changing pace of manufacturing and shifting needs of a business has proven to be contentious and fragile. Resulting in these teams working on integration as the afterthought to ensure their core system remains stable and aligned to its immediate stakeholders. This brings about a lot of room for errors and inconsistencies and again has proven to be easier to merge it later in a more manual way rather than in near real-time.
Evaluating the Current Approach
So, the next logical question to ask is, what are the forces at play that are pushing these three systems apart. I think a lot of the focus can be on the heterogenous nature of manufacturing and the fact that one single vendor cannot be responsible for Level 0 up through Level 4/5 in a business (See Purdue Reference Model). This variety leads to the systems getting fractured more and more over time as manufacturing must keep running and adapting over time. Identifying where the most diversity in the landscape is yields a clear answer in that this exists at Level 0 through Level 2 systems, and up to Level 3 but many try to standardize this already. Most companies will agree that they have an enterprise standard for Level 4/5, even here there are exceptions, but these are smaller by comparison to the plant level and the reasons are smaller and smaller every day.
Looking closer into the plant level the physical machines and assets represent a very real, and very interesting, challenge of how to solve efficiency problems in production. Each asset is picked, or built in-house, for very specific and important reason. The speed it operates at and the function that it performs represents a key benefit to operations and tied to the value proposition of equipment vendors. Therefore at Level 0, and the outbound digital signal layer of Level 1, it is not a simple or cost effective approach to standardize everything from a single hardware vendor so it will remain heterogenous to a degree. There is an industry push to standardize this layer Modbus, OPC-UA, MTConnect, and groups like Open Industry 4.0 Alliance but this only really applies to newer equipment and some equipment still uses proprietary protocols behind a paywall.
Moving one level up from here are the Level 2 systems (i.e. Batch Control, SCADA, DCS, HMIs, etc) which provide a very specific role of controlling the various engineering system(s) and are tightly coupled to Level 1. This software layer does have maturity around adoption industrial standards (i.e. OPC, MQTT, RESTful, etc.) to interface to on their top end by the Level 3 supervisory systems. This is where the first real integration challenge exists around how to manage the mapping heterogenous engineering/controls systems to a diverse set of manufacturing processes that are, hopefully, defined by enterprise business processes and keep them aligned over time. The catch here is that modifying the control system is not something that you “just do” as there are many consequences to this behavior of which the operations technology team will have no tolerance for.
At Level 3 this is where the manufacturing processes are defined and the priority and often the first-time end-users’ interface to the physical systems and automation layer around them. Weather it’s a central control room for a process area/cell or a production line/cell where the operator is performing tasks the MES is there to coordinate, guide, enforce, track, and report. It’s a lot of tasks for the local system to manage and can become unwieldy for the end-user if information is not driven from the technical system(s) in place. The coordination of production also has an impact on supporting functions at the plant around material supply, quality, maintenance, safety, environmental impact, and the list goes on. This need to blend the various digital signals together in a way is where the simple/traditional ‘linear’ namespaces struggle; but more to come on the UNS approach later.
Finally at the Level 4/5 layer this is where the central ERP coordination of the various business processes come together as a singular master and transaction structure. The business functional topics of Finance, HR, CRM, SRM, and Supply Chain all reside here and provide coordination across the various groups to align processes and ensure work is done in a coordinated manner. This is all well and good but often does not include the granularity required to run manufacturing operations and manufacturing operations often lack the central standards and linkage back to ERP, as things change over time.
These systems and the teams that manages changes and the alignment of change top-down, bottom-up, and middle-outward all have the best interest of the business in mind but without proper tools and processes in place create unintended churn across the organization.
Rethinking Data Management
Now that we have a baseline for how operations are behaving today...the challenge is, and will continue to be, how can we do things better? Responses here will range from tactical OpEx improvements to strategic CapEx projects. The bottom line is that there needs to be a consistent way to guide the business so that priority of OpEx and CapEx can be assigned. This has traditionally been done via centralized analytics and standardized KPIs and resulting in a view of the business that is useful but lacks the ‘near real-time’ impact that operations bring. These improvements are useful, but they are all still ‘after the fact’ and can help prevent issues but are not going to catch them in-flight.
The key part here is to find a new way to tackle the old problem of relating digital signals and the applications processes in a way that influences both manufacturing systems and people to improve efficiency. This combination of MES and Historian has been a mainstay for many industries for a long time and the combination works. However, there is a new way to go about combining them that uses newer technology which gives scale, reduces overhead, and has a bigger reach across the systems. This approach takes the best of a proven approach and combines it with modern technology, application design, and industry standards. This modern approach also happens to align with what is being delivered ‘today’ with SAP Digital Manufacturing in combination with SAP Datasphere to deliver what we are referring to as the SAP Inverse Historian Reimagined.
