There has been a fascinating power struggle on
HCSC makes Maui brand Hawaiian Raw Sugar using 100 percent cane sugar grown on its own, 37,000-acre central
Cane grinding, juice extraction and boil-down are performed at the company’s mill situated near Puunene,
John Rivera, power management analyst for HCSC, came to
The company itself is not slow to modernize, but when Rivera came on board, the electrical generation side was essentially stuck in the mid-1900s. Data connectivity was not part of the picture, and information from conventional metering and relay-based distribution was not able to give timely warning of power glitches, sags and failures. “Our manufacturing intelligence application began in the power plants,” Rivera says. “We learned how devices with common protocols like Modbus could talk to each other—and to us. Within three years, we had upgraded to a fully digital Wonderware InTouch HMI [human/machine interface] system.”
The power group was then able to monitor electricity from both its own plants and the
As a result, Rivera has overseen the installation of Web-based, company-wide intranet systems that provide real-time visibility to supervisors, managers and line employees in a variety of areas. “I have a strong belief that when you open information to the masses, you allow everyone to become part of the picture,” Rivera says. “It takes all sorts of functionality and all sorts of approaches to piece together the whole picture. The more people who are available to help, available to analyze and support, the better the bottom line.”
The context for manufacturing intelligence is easy. Set a goal. Create a road map for reaching that goal. Set checkpoints along the way, then use manufacturing intelligence to poll those points as you drive by. Manufacturing intelligence can also help when you reach your destination by looking in the back of the truck to see that all your cargo is still there. And, it can do a walk around just to see if the wheels are still on.
There is no rigorous, single definition of manufacturing intelligence. The term itself derives from military intelligence, which gathers information from all possible sources, legal and illegal, to make guesses about what the enemy is doing now and what it will do in the future. This considerably less bellicose variant also looks wherever it can, but our information sources are (hopefully) less prone to secrecy. A manufacturing intelligence tool can take the form of a dashboard, spreadsheet, chart, report, even a light that illuminates—any medium that will convey the right snapshot of operations. The only complex part is defining some of the details, such as the goal, the road map, the checkpoints and your destination.
Manufacturing intelligence in the context of performance management is a tool that tells you how operations are performing. The heart of performance management itself, of course, is setting and reaching goals or key performance indicators (KPIs). The exact nature of these KPIs is almost always specific to a given situation. Manufacturing intelligence can then be thought of as an overlay placed on top of manufacturing data, revealing the meaning of these data as seen through the lenses of the KPIs. In other words, forget about seeing whether valve 146-27 is open or closed, but whether this valve and all of its upstream and downstream points collectively tell you that x amount of good product is heading out the door.
“Manufacturing intelligence is not just raw data about your manufacturing information flow,” says Claus Abildgren, marketing program manager, Performance Product Management, Wonderware, a Lake Forest, Calif.-based manufacturing software vendor. “It’s higher level aggregation that allows you to use the data to improve performance, to make better product tomorrow.”
Manufacturing intelligence is closely tied to production management. From Abildgren’s point of view, production management’s function is to execute production orders, which includes equipment set-up. It also includes the orchestration of that equipment to start, stop and perform activities to make product to fulfill orders, as well as managing operational master specifications and procedures. Finally, production management handles equipment and product formulas and operational procedure process limits. Manufacturing intelligence’s function, on the other hand, is to draw from production management data to visualize production status and history, including the current status of machines and orders. “It also captures production events, and detects deviations,” he says. “Manufacturing intelligence does what is necessary for the delivery of accurate, real-time information for improved decision making, including real-time data collection, notifications and alerts, plus process analysis and traceability vis-à-vis operational KPIs.”
Deriving manufacturing intelligence from all of the information flows from all of your equipment in all of your facilities is a process that involves both abstraction and translation. Abstraction (or aggregation) merges details to arrive at the smallest and sharpest picture possible. Translation then maps this picture to business processes. “You definitely need to focus on business processes,” says Jeff Tropsa, senior business consultant, Simatic IT Business Unit, for Siemens Energy & Automation Inc., the Alpharettea, Ga.-based automation vendor. “What do users need in their roles to do their jobs? Where is this key information located? What sources offer the best information? Answer these, then put it all into views that are easily navigated, and the gains are tremendous.”
Christina McKeon, director of product marketing for Performance Management, for enterprise software vendor Infor, in
In gathering and aggregating data, there are certain themes that are common to all aspects of performance management, not for just manufacturing intelligence. Chief among these is the need for consistent data. “In multi-line facilities, and especially in multi-plant contexts, there has to be consistency in what is being measured,” Abildgren points out. “You have to compare apples to apples. It’s best if everyone can hash out a common data model and capture information in well-defined templates.”
Behind any specific implementation of manufacturing intelligence is a need to:
1. Ensure the availability of data
2. Pinpoint who sees what
3. Define the media for delivery.
There is, of course, quite a bit of interaction between the first two. Only by understanding the range of data available to you can you get a sense of what views, abstractions and metrics may be drawn from the data. Conversely, once you figure out what people want to see, you can quickly determine the best sources for the desired information. The explosive growth of manufacturing data connectivity infrastructures, from standards, networks and buses, to intelligent components and peripherals, means that virtually any motion (or stoppage) can be monitored pretty much in real-time.
The manufacturing data infrastructure is not yet 100 percent plug-and-play, but it is a magnitude easier to tie equipment together today than it was even five years ago. What remains is the need to move from raw data flow to intelligence—that is, the need to winnow through millions of bits of data to come up with the one or two, or a dozen, or two dozen things that are important to you. This winnowing is at the heart of manufacturing intelligence systems.
For most systems, there is always a fallback to use systems integrators to make this move from raw data to intelligence. Good system integrators put trained implementers to work, mapping your needs to software and hardware. When it all works well, these individuals both know the capabilities of the tools as well as the skills needed to ask the right questions. But integrators are not always necessary, as many systems have built-in algorithms, tools and set-up routines that make the mapping easier. These can range from pre-defined templates that can be selected as users build up their data models, to individual objects or components that can be linked easily to overall needs.
In the long run, however, installing manufacturing intelligence is less a technical issue than a procedural one, because the technology, when all is said and done, is merely a medium for the information. As we have already indicated, the process begins with the chief executive. Practically from that point on, everyone’s solution is going to be different, but there are certain commonalities, and those are what we can focus on here.
Tropsa of Siemens says, “At the beginning, there can be concern if people worry about big brother watching, but it’ not about hitting people over the head, it’s about all of us looking at the entire manufacturing process and seeing where we fit. You have a natural tendency to pay attention to your own roles and the things you’re already familiar with, but the real question is how can we—everyone in the enterprise—optimize throughput, not just this or that specific line, train or unit.” He emphasizes the need to rethink and understand the true metrics that drive the business, a commodity that is “not typically known, because automation and advanced control can exist at many different levels. Without a global view of the enterprise, you’re likely simply to optimize your own sub-system.”
The touchstone in many facilities is seeing how much profit any given asset contributes to a business unit. “If you can truly analyze profitability, and keep a high-level view of the control environment, you can see gaps and lost opportunities,” Tropsa says. “And you can begin to see what has the greatest impact right now. For instance, if a motor is tending toward failure, should you fix it right now? What if it’s doing important work? If you can risk waiting a few hours and still escape catastrophic failure, you might want to wait on the repair.”
Wonderware’s Abildgren sees the application of manufacturing intelligence as part of a given company’s evolving automation systems. “You can represent your physical manufacturing processes in what we call a ‘plant model’ in an IT setting,” he says. “This provides supervisory functions of communication, alarming, visualization, security and so on to monitor and control physical processes and activities.” You can then evolve the existing plant model later, now to include additional functional capabilities for, say, formula management, or the capture of production events, or the identification and qualification of equipment downtime. “It’s not about building another, separate application/functional silo, and then having to integrate this to the supervisory system—it’s a natural extension of the same logical model.”
Once fundamental operations are understood and under control, manufacturers typically evolve their views so they can understand true capacity and overall manufacturing performance across several operations or lines, “many times involving statistical process control and higher level compound metrics,” Abildgren explains. “If your systems are truly modular, you can add further functional components to the existing plant model. It won’t matter where you start or where you want to end.”
Infor’s McKeon points out that implementer profiles can be key. “You need people who know technology, processes and above all, people,” she says. “Details on the skill sets are specific to the needs of specific companies, but the overarching need is for a team that can focus on the business objectives and understand the processes well enough to spot both best practices and unwanted duplication. People skills are mandatory from the earliest stages, to allay fears about micromanagement and champion the positives of knowing what you’re doing.”
The benefits of installing manufacturing intelligence almost always repay the work involved. And the benefits extend beyond the manufacturing suite. One consumer packaged goods manufacturer worked with Infor to gain intelligence on the financial side of manufacturing. The company was taking weeks to finalize budgets in a chaotic scene where hundreds of people each worked with their own spreadsheets. By installing a system to keep tabs on projected and actual costs, the company was able to reduce its time to close yearly budgets by 50 percent—making forecasts far more timely, and near-term objectives far more grounded in reality.
For more information, search keywords “manufacturing intelligence” and “performance management” at www.automationworld.com.