The early results are in: Investing in data capture and analytics can lead to big payoffs for food and beverage manufacturers who take action on the information they collect. Sample results tell the story, from a global processor that achieved sustained OEE ratings above 85 percent to a baking company that leveraged data to increase production by 5 percent and reduce cycle time by more than 20 percent.
Processors learning their way to better running efficiency, product quality and productivity call it a journey that often requires overcoming both behavioral and physical hurdles.
Yogurt maker Chobani began the process five years ago, when Hugh Roddy, VP of global engineering and project management, joined the company. As a native of Ireland with extensive experience in the European approach to manufacturing engineering, the idea of using data to continuously improve processes came naturally.
“People want instantaneous data at their fingertips. We have projects and changes that are complete in six months that take other companies a year or more to do,” says Roddy. “Collecting data is useless unless you act on it; you have to collect it, display it and then be able to analyze it. Our engineering, operations and continuous improvement teams work in tandem, using the data we collect to learn, change and get better.”
According to Roddy, not every change is perfect, “but we stay focused on finding out where we still need to improve, making investments based on the potential income to be gained. We started at the packaging/filling level, so we’ve probably touched on only 5 percent of what we can do.”
Among the most important metrics for Chobani are KPI improvements, OEE monitoring and targeting goals, as well as throughput measures and input materials to output yields. “We employed all the tools from Rockwell Automation that we have embedded, but without using the Ignition platform from Inductive Automation, which allows unlimited clients without paying additional licensing fees, collecting all this data would be prohibitively expensive. We’ve also utilized mobile clients extensively to drive the data to the plant floor,” Roddy adds.
“We’re most interested in driving change from the plant floor and empowering our people,” he says. “Our systems allow operators to record what’s happened on the line and take real-time action to correct problems. At the same time, someone has to be accountable for taking action, so you need a team focused on change.”
Roddy can look at data from all of the company’s plants from his smartphone. “We noticed, for example, that the second and third shifts at our Australian plant were routinely producing less than the first shift. It turns out that operators on those shifts sometimes weren’t at their stations when actions were required. It’s one of the reasons we’ve put live dashboards in front of operators at all our plants. It makes people accountable and motivates friendly competition to achieve goals. As a result, we’ve seen huge changes in operator performance, productivity and efficiency.”
Walk before you run
It is common practice for companies to begin extensive data collection and analysis with one process or at one plant, and then extend their efforts to other areas as they gain confidence and see the benefits.
For Campbell’s Soup Company, the process began at its Pepperidge Farm facilities.
“While data gathering helped us with the initial formation of our strategies for what data to collect and how to collect it, at the beginning the effort was very disparate and paper-based, as well as very tactically driven,” explains Mike Hughes, senior director of global supply chain process excellence. “We often collected the same data multiple times for different projects.”
The company has been measuring line performance, OEE, downtime, yield and other factors for the past 10 years, but Jim Prunesti, VP of global engineering, says digitizing the information has made it much more usable for decision-making at all levels. “We act on the data by using it for daily direction-setting and shift overlap meetings, as well as rolling it up to the broader leadership. Summarizing our findings across plants has been very helpful, since many of them deal with common issues. We’ve already seen at least a 10 percentage point improvement in line performance.”
Prunesti cautions that, despite these benefits, “the reality is that implementing this kind of program is a significant challenge. There’s no solution out of the box, especially when you’re dealing with both new and legacy equipment. You have to find creative tools to achieve your goals and it takes years to implement,
especially when you’re rolling it out over 20-plus plants worldwide.”
Hughes says data collection needs to start at the lowest level in the plant. “If you don’t have data detail, you can’t make changes. We start with micro data sets and then roll it up.” But more data often creates more problems to solve. “We’ve had to retool our database architecture on the reporting side to support data analysis and make our findings useful,” he states.
“We’re trying to understand our process from the back-end to the front-end and identify where the opportunities lie,” states Hughes, “but we’re already seeing this as an investment with real financial returns for the company. More workers are also engaged and embracing data because it’s a lot easier for them to see it digitally. Decision-making has gotten easier as well, because now it’s based on facts rather than emotions.”
A platform to integrate
One of the advantages of digital data collection over manual, paper-based efforts is that it can help identify and calculate minor process disruptions that are normally overlooked.
“Micro-stoppages, which typically last five minutes or less, are a major source of downtime,” says Reid Paquin, industry analyst for food and beverage manufacturing and consumer product goods for GE Digital. “It’s too difficult to record the data on this manually, but these hidden losses contribute to a significant amount of lost production.”
Paquin says that only a few companies, typically larger ones, are using advanced analysis of their collected data to drive change. “Approximately 80-90 percent are still focused on real-time data collection. Many companies are still collecting data manually with spreadsheets, often in silos and for a single purpose.
“You need a platform to integrate and automate data collection and analysis,” he explains, while admitting that it can be a massive task. “One of our customers has over 5,000 manufacturing applications in its production systems.”
Particularly troubling is the lack of data to support risk management. “You must have data to assess risks to the company and their potential impact. A lot of companies are lagging here because they lack the essential data, especially in connecting areas of risk across an enterprise.”
Starting fresh with data
Smaller manufacturers, especially those with newer plants, often have an easier time organizing a robust data collection and analytics program. That’s the case for Freshpet, a maker of fresh foods for pets, with 140 employees working over three shifts.
When the company bought an old factory in Bethlehem, PA in 2012, it was able to gut it and install an Ethernet-connected infrastructure. Consequently, Engineering Manager Sam Torres routinely monitors more than 600 data points for operational and maintenance purposes.
“Our IDF (intermediate distribution frame) enclosure is the principal component in a well-designed plant data detection system,” Torres explains. “We have a total of 180 nodes on our network, collecting as few as four data points to as many as 20 from each process device. We use dashboards to track what’s happening now as well as what happened before.”
The data helps Freshpet understand the issues and the next steps it need to take, which is reviewed with engineering and operations staff every morning. “Case counts, for example, are useful so operations can gauge productivity; while cycle counts help the maintenance team schedule work to prevent equipment failure,” Torres says.
“The same concepts apply whether you’re making candy bars or putting milk in a bottle,” states Torres. “It’s essential to be able to see data over time, as we can do with FactoryTalk Historian, because that’s where the real learning takes place. By identifying trends, you’re able to understand the capabilities of your production systems and discover the root causes for why you’re not meeting goals.”
Torres cites one example of how data was used to justify the purchase of a new film splicer for a chub packaging machine. “Before the automatic splicer, the operator had to shut the machine down to change the film roll. Each changeover took 6-7 minutes, which resulted in 1.5 hours of lost production time every 24 hours,” he says.
“Without data you can’t prove anything, and you can’t manage it if you can’t see it,” declares Torres. The company is now considering purchase of an MES system to enable more detailed reporting. “We’re still doing too much manual, paper-based data tracking. We’re collecting a lot of data but there’s not always a good, friendly way to display it.” (Read machinery case study on Freshpet’s Bethlehem facility.)
According to Greg Hood, global product manager for Rockwell Automation’s RAPID Line Integration Solutions, implementing a standard data access model is essential for food and beverage manufacturers, because their production lines, more than in other industries, must connect disparate machines from many different vendors.
“Program goals must be specific, measureable, on a timetable and achievable,” he says. Once the program goals are established, ensure that you have the data required to monitor goal achievement.
Hood suggests starting at the machine level. “Data required to monitor those goals must be available and collecting it must be repeatable throughout the plant or across the enterprise,” he says. “Then focus on line level context and display. The biggest challenge is in getting real-time data that’s accessible so operators can monitor events and root causes in order to make changes as opportunities arise during the production run.”
Hood says getting everyone from plant workers to enterprise users to have confidence in the accuracy of the data is crucial. “If you’re going to monitor operator actions, the people on the line must agree on what the goals are and support them. They‘ll be more willing to buy into a system if they see it as an enabler for their performance. Once they’ve bought in, you must communicate and celebrate the wins.”
Achievable goals, data standardization and accuracy, context, integration and buy-in are the foundation for success, according to Hood. “Recognize that you can’t achieve every goal at the same time. See it as a journey where you walk first and learn as you go along,” he states.
The what and why of data
Making data actionable starts with a framework for data collection that’s based on company priorities. For the Kellogg Company, that framework includes OEE criteria and sustainability targets.
“Data points are selected based on what can influence OEE, such as minor equipment stops linked to inputs, so we can do a root cause analysis,” explains Larry Turpin, VP of shared services, global engineering. “We’ve changed some of our data points along the way because we questioned how actionable some of the data we were getting really was.”
Turpin says everyone realizes data is important, but the cost to have it available in all desired instances is difficult to afford and not always financially viable. “Key data priorities for the company are uptime, minor stops, throughput, product quality, including scrap and overweights, and raw and pack material usage,” he says.
“The best way to make data actionable is to review it with the right personnel, usually a combination of operators and plant managers, and to build it into each plant’s daily direction-setting meetings,” adds Turpin.
The data collection efforts can sometimes run into barriers at the plant level, says Kellogg’s Tamesha Murphy, director of power, controls and instrumentation systems. “IT infrastructure is not always kept current in facilities, especially smaller factories, but we’ve found that data collection is foundational to efficient plant operations.”
Tighter process control
A thorough understanding of your existing process is essential before you can start improving it, says Maryanne Steidinger, director of industry solutions for Schneider Electric. “Start by reviewing your process, learn where the critical points and potential bottlenecks are, and how your equipment performs, whether it’s the temperature of an oven or the speed of a conveyor.”
Steidinger says the goal is to control the process as close to 100 percent as possible. “Most companies run at 65-70 percent of their capabilities, which leaves a huge amount of room for improvement. Significant gains are achievable without a large capital cost just by fine-tuning your process controls,” she says.
“Training then becomes imperative, so your workforce can understand and use the data they’re seeing. You have to put the numbers into language employees can relate to, such as 12 pallets of products thrown out rather than a 12 percent failure rate.”
Acting on data
System integrators can be a great resource for manufacturers looking to make operational data more available and actionable.
“There’s no shortage of raw data from production systems, but you can’t make it actionable unless you give it a context,” says Divya Prakash, director of the manufacturing systems and solutions division at system integrator Matrix Technologies, a member of the Control System Integrators Association (CSIA). “That’s the only way you can get to the root cause of a problem and actually fix it, rather than just applying a band-aid,” he says.
“You may have data that tracks the temperature profile over time, for example, but you also need to know what was being made at the time, who was making it, which shift and what were the environmental conditions, so you can compare that to the same situation on another day,” he states.
The more manual a process, the more variability there will be in the results due to the actions of human operators, Prakash notes. “The proper action then would be to either provide operators with more training or automate more of the process.”
Process data can be peeled like an onion, allowing you to understand more about your processes as you dig deeper into the details.
“The more details you have, the greater control you can exert over operational results,” says John Tertin, director of manufacturing information services for system integrator ESE, Inc., a CSIA member and Rockwell Solutions Partner.
“Priorities also change as you get deeper into the details. You need to look for the root cause of a production issue, such as whether heating element temperatures are operating above or below set points. This could be causing a significant percentage of your product rejects.”
As customers demand more customized products, data can also help you determine whether you can take advantage of new business opportunities. “A customer that makes dipping sauces, blending ingredients together so they crystalize over time, received a large order to coincide with the Super Bowl,” explains Tertin.
“By analyzing their historical production data, they realized that stabilization chemistry limited their ability to make sufficient additional product in the amount of time they’d been given.” It was better to turn down the order than risk a dissatisfied customer.
Old habits often die hard, especially when it comes to looking at the big picture. “People get very set in their ways, which means projects often focus just on getting the data,” says Chris McLaughlin, SCADA and MES specialist for system integrator Vertech. “To make data actionable, however, you have to make it available at the right time to the right people.”
McLaughlin says most plants have a feeling that something is wrong in a process or machine, but they don’t always know exactly what’s causing it. “That’s why playing Sherlock is one of our most important functions. If you can show them the reasons for downtime, the numbers behind OEE become much more powerful.”
Training people how to implement continuous improvement practices makes a huge difference in their ability to make change happen if they understand not just how to use data, but why to use it, according to McLaughlin. “When you can get different people with different roles, priorities and experiences to collaborate, using the same software and sharing ideas, rather than functioning as islands looking at their own spreadsheets, it becomes much more powerful, motivating and enabling of change.”