There is an ever-increasing need to integrate business strategies, manufacturing strategies and process control strategies (Figure 1).
Figure 1. The control strategy links the business strategy and the manufacturing strategy.
A little more than a decade ago, a group of mid-level managers of The Lubrizol Corp. rallied around a more holistic approach to close the gap between data and analysis to identify opportunities for eliminating waste, increasing production and boosting profitability. Today, that original vision remains intact, and what became institutionalized as the Operations Management System, or OMS, is enhancing relationships with vendors, improving productivity and contributing in meaningful ways to operational excellence.
From the outset, the over-riding goal of the OMS working team was to improve the efficiency of data utilization — not simply to report data. Analysis was key — to look for relationships and correlations, identify where variability was coming from, and act upon those results to further improve business dynamics and economics, product quality, throughput and time cycle while cutting variability and waste (Figure 2). Moving an organization from a data collection, data reporting mindset to an analysis mindset improves business and production knowledge.
Figure 2. Changing the mindset from data collection to data analysis increases profits.
The challenge for the OMS team was to gather the data being generated by disparate systems — from a variety of distributed control systems — and integrate the data across those systems. Every process was fair game. Targeted areas included production operations, maintenance, engineering and instrumentation, analytical lab results, environmental systems, asset management instrumentation, process control, calibration, quality assurance data, demand forecast linked to production planning, equipment and processes, and business data.
An early observation was that different departments have different needs for the same data. For example, compare the needs of a business group versus those of an operations group. The business group must know the inventory, i.e., how much material of each component was used to manufacture a batch of product. In addition, they require data on the quality characteristics (purity, properties, etc.) for making the product. The operations group must know exactly how each component will be added during the manufacture of a batch (the recipe), the quality characteristics of the component and the inventory quantities. How each component is used is necessary for correlating process dynamics to batch yield, conversion costs, and resulting impact on product quality. Thus, the needs of each group must addressed differently (Figure 3).
Figure 3. The same data serve many distinct requirements at different levels.
During the planning phase of the initiative, the team identified the need to establish a process historian. A historian is a database of time-related data such as temperatures, pressures, levels, device characteristics, quality and comments, all stored with a timestamp. It allows the review and analysis of such data. Trends can be identified and values searched and exported to various data analysis and reporting tools. Historically, much of the data that existed was in written form or stored in local systems. Thus, it required painstaking manual work to access important information and properly format it. The process historian helps address some of these problems.
After developing product selection criteria for the historian, and reviewing the offerings of a number of vendors, the OMS team selected the OSIsoft PI system as the corporate standard. Prototype implementation began at manufacturing facilities in the United States and Europe. Sensors were installed at key points in each facility’s manufacturing equipment and were wired back to controllers, such as a distributed control system (DCS), programmable logic controller (PLC) or data logger.
Once in place, the PI scanner essentially watches “tags” for certain criteria, usually a value change of a certain magnitude, at a particular frequency. These tags are pieces of data on the control system. When the criteria are met, the data and time-stamp are sent to the PI server. Whether a value (data and time) should be compressed and stored was determined by additional criteria configured in the server.
Non-critical information is discarded, allowing the basic fingerprint of what happened to be captured without requiring all of the data to be stored, thus efficiently using the disc. Users of the system can then access the PI server, or servers, that contain the data they wish to review. In the event of a quality concern, the data repository helps users determine how a particular product batch was made.
Once sorted and stored, data become available as input. Lubrizol could then move on to the next step: implementing data analysis tools to uncover process relationships, characterize process variability and correlate process dynamics. These tools are used to develop economic and product quality parameters.
Standardization and collaboration
Analysis tools are only useful if manufacturing processes are automated. After developing a strategy for success, the next step was to transition older, non-instrumented processes into controlled processes. A milestone was the decision to standardize on one DCS provider for the Lubrizol additives business. Another was the conversion of various legacy systems to the new global standard. Emerson’s DeltaV digital automation system was selected.