Alison Smith: The time has come for the Operations Data Warehouse
The market’s appetite for manufacturing operations performance visibility continues to blossom, maturing slowly from an early fascination with key performance indicators (KPI) to a more sophisticated recognition that while the right KPIs have value, key performance drivers (KPD) are needed.
At the technology level, this is driving a convergence between the world of traditional business intelligence (BI) and enterprise manufacturing intelligence (EMI).
For those new to the EMI acronym, it refers to applications and architectures that turn the volumes of data generated by the manufacturing environment into business-oriented intelligence for real-time decision support. This converged capability set blends the real-time aggregation, contextualization, and analysis of operations data with sophisticated analytics, mining, modeling, and scenario analysis techniques that have traditionally been the hallmark of BI applications.
The first generation of EMI was about creating KPIs. Operations intelligence progresses this idea to KPD.
Simply put, a KPI tells you how you’re doing, but doesn’t necessarily tell you what to change to influence the KPI. KPDs, on the other hand, are the major levers that you can exercise to influence the KPIs. As the complexity of production increases—maybe in terms of product mix, shared assets, or floating bottlenecks—so does the need for KPDs.
The goal of identifying a KPD is to understand the correlative relationships between the various variables that you can control to influence your performance against the goal that’s being monitored—i.e., the KPI.
Of course, this type of analysis presumes that you’ve got something to analyze, which means collecting data and stashing it somewhere—although if you read my April column, we’ve clearly got some work to do in the area of basic data acquisition.
That said, depending on your industry, you’ve likely got randomly distributed databases associated with desktop applications. Getting to the big picture of what’s happening—and why—requires the addition of a new architectural element: an operations data warehouse.
The goal of the operations data warehouse is to provide a repository for operations data not just from one or two shop-floor systems—e.g., MES and CMMS—but from any of the systems whose performance influences product outcome. Discovering these interactions takes a broader information set than analytics on a siloed application data store can reveal.
The process industries are ahead of the curve on assembling strategic manufacturing architectures that include operations data warehouses—whether it’s in the form of an enterprise historian or a traditional relational store—for the purposes of operations intelligence.
Manufacturers with discrete styles of production should take a look at what their counterparts in process are doing. The benefits in terms of improved performance on first pass yield, first time in control, overall equipment efficiency and availability are large, not to mention additional performance benefits gained by understanding the impact of changeovers, schedule variance, mix, machine and tool maintenance on overall product quality and process performance. It’s time for manufacturing operations to put on their IT hats and incorporate this simple yet powerful element into their operating environments.
|Alison Smith is a director within AMR Research’s Market Services group, where her current focus is on manufacturing operations. Smith offers insight on applying manufacturing execution systems, enterprise manufacturing intelligence, and asset performance management solutions across vertical industries. She can be reached at firstname.lastname@example.org .|