Improvements in sensors, servo technology and programmable controllers enable manufacturers and industry to monitor processes to prevent expensive failures, quickly react to problems that do occur, and customize production.
“Within GE, as we put sensors on machines and products, we’re able to predict, adapt and react more quickly. For example, we can see premature vibrations, signals that we’re losing a pump or losing a motor. That helps us manage unplanned downtime,” Bart Weihl, executive, global operations, advanced manufacturing initiatives group at General Electric, said. “More people today are moving in that direction. I think companies are still in the process of understanding the full benefits of what a connected machine will get you.”
Connected machines provide real-time visibility for industrial applications, including energy production and manufacturing, said Matt Newton, director of technical marketing for Opto 22. “Enabling sensor data to be processed and monitored in real time through IP connectivity built into devices like programmable automation controllers (PACs) allows enterprises to obtain real-time situational awareness of almost all aspects of their production and maintenance.”
Analyzing the now-available data facilitates analytics, diagnostics, high-speed control and automatic configuration over a network. Instead of performing an autopsy on a failed piece of equipment, engineers and technicians can act to save the machine before it breaks, shuts down a line and ruins the part being manufactured.
“Motion control is really key on smart machines,” said Bob Hirschinger, principal engineer at Rockwell Automation. “With better motion control, the machine is more flexible and performs better. Sensors are really important in motion control. They can send back information on temperature, vibration and horsepower to facilitate direct control and/or diagnostics information. There is a lot of status and diagnostic information we can make available in the PLC and directly up to the cloud.”
Motion control technology is being adopted increasingly in the packaging industry, Newton said. Applications include product synchronization, product spacing, product labeling, wrapping and knife control and form sealing.
At a plant making wrappers for Snickers candy bars, the improved controllers and sensors could alert someone immediately if, for example, one wrapper printed “SNIC” twice on the same wrapper, Hirschinger said.
“We’ve seen reduced cycle times and improved product throughputs,” he said. “The control systems are so much faster. We are sampling and analyzing 50 measurements in milliseconds. With the sensors, we can take that data, analyze it and make changes.”
Renewable energy markets are adopting motion control, Newton noted.
“Wind turbines are using motion control to autonomously move the turbine into the most efficient wind direction. Concentrated solar power involves moving receptacles that follow the sun to collect heat to warm a medium that drives a turbine.”
To make this possible, PACs and programmable logic controllers (PLCs) have advanced in many significant ways since the turn of the century—including managing more drives and collecting more data now available from improved sensors.
“In the past, one PLC could control a limited number of drives,” Hirschinger said. “Now upwards of 200 drives can be managed directly through one PLC.”
At the same time, sensors have improved. Sensors are now more robust and able to provide more critical data, Weihl said. Deviation in these areas can signal that a part is about to fail.
“That’s the first step with GE’s Brilliant Factory—putting sensors on the machines, tying in to the controllers and talking to the network,” he said.
Big gains to be had
GE’s Brilliant Factory Manufacturing software analytics track multiple data points and manages machine function and maintenance. GE’s case studies have shown that Brilliant Manufacturing software helped reduced unplanned downtime 10–20%, resulting in 20% decreased inventory, and delivered 20% capacity recovery by deferring a capital expenditure for one year.
Consider a cutter making expensive parts used in aerospace.
The cutter becomes dull and compensates by running at a higher-than-optimal horsepower—resulting in three possible scenarios, Weihl said.
The cutter could fail and would have to be replaced. At several hundred dollars a cutter, that is costly but not as bad as other potential outcomes.
The cutter could jar the machine and knock it out of alignment. That realignment could take up to four hours, resulting in lost production time.
In the worst case, higher friction temperatures from the dulled cutter could weld the cutter to the part it was making. “Then you’ve lost the part, which is very expensive,” he said. “It could cause nonconformance, have to be repaired and submitted to the customer for approval. It could be weeks or months before the customer could respond back.”
Now, sensors monitor the horsepower of such cutters and via the PLC can alert someone immediately when horsepower exceeds normal parameters, Weihl said.
When a machine breaks, a factory of 10 or 20 years ago might show a red light on the machine, said Jeff Morgan, president of Hermitage Automation and Controls in Richmond, VA. Or a manager might notice workers heading outside for a smoke break because their machine no longer functioned.
“Now, instead of just seeing a red light, a signal goes to someone’s office in maintenance and an email is sent to management saying ‘line two is down’,” he said.
Machines more friendly today
Although today’s machines are increasingly smart, they often no longer require an expert to set up and monitor.
Consider a robot with three joints, which are actually three servomotors.
“One master controller knows the position of each servomotor,” Morgan said. “Ten to fifteen years ago, there was a requirement for a specialist on your team that could deal with that. Now, you don’t have to be a specialist. Someone with general knowledge could program it and make it work. You don’t have to be a PhD anymore. They’ve made it a lot easier for the end user of the product.”
Installing, configuring and tuning machines also is easier.
“In the past, these products were difficult to configure, commission and deploy,” Hirschinger said. “In many cases, when you employ a servo on a machine, the servo has to be tuned and optimized for the mechanics the machine is running on. That used to be analog, then digital. You needed an expert to set those up. More recently, advances in servo and servo loops technology allows these drives to self-tune. These drives can recognize the mechanical configuration and measure data that is coming out of the machine.
“We can automatically tune the drive based on the mechanics it’s connected to. Today you can take a servomotor with a well-designed mechanical machine and easily integrate it. You don’t have to enter catalog numbers and fine-tune it. It tunes itself. It allows you to commission a machine much faster.”
The available data also helps as manufacturers transition from selling only products to selling products-as-services. Both the system manufacturer and end user can receive data critical to monitor and rate machine effectiveness. Case in point: “I scheduled this machine to run six hours: did it run six hours and did it meet quality standards?” Weihl said.
The company 1200 miles away that made the machine also might get a notification when something goes wrong, Morgan said.
At one company using machine-monitoring sensors, a milling machine had high-speed spindles that could be knocked out of service by unexpected vibrations, lack of lubrication to bearings or other catalysts—resulting in the need for a $50,000 repair and a three-day delay as a replacement was ordered, Weihl said.
The company was using the data to do an autopsy after a machine failed. Weihl wanted to be proactive.
“You’re monitoring the spindle to do a post-mortem,” he said. “I wanted to use it as a preventive tool. I wanted information from the vibration sensors and temperature ported out so I could watch for changes in the bearing temperature.
“We wrote a program to watch and chart it. In the first month, I was notified that bearing temperatures were beyond our parameters. I called the machinist. He said he was monitoring the situation and that the rear bearing was 10° hotter than it should be. Our machinist was monitoring the situation and said that the rear bearing was 10° hotter than it should be. Upon further exploration, we found the lubrication pump had failed. It was a $300–$500 item but, left unchecked, the repair would have cost tenfold to fix. Instead, we were monitoring those temperature points. We actually caught and prevented the failure with sensors on the machine.”
These improvements mean manufacturers that have already evolved from reactive maintenance to preventive maintenance can now save money by moving to predictive maintenance.
“With sensors and the data you harvest from machines, you can move yourself into more condition-based maintenance,” Weihl said. “It saves the company money and prevents downed assets.”
“Predictive maintenance saves time and money in almost all sequences of the maintenance cycle,” Newton said.
Opto 22 recently worked with a customer that was automating a chain of convenience stores and looking to make its maintenance department more efficient, he said.
“Traditionally when something at a store broke down—for example, a refrigeration system component—a maintenance technician had to drive hundreds of miles to the site,” Newton said. “Then the technician had to debug the refrigeration system, identify the failed component and wait for a replacement component to be sent to their location for installation and calibration. Combining today’s connected and intelligent sensors with cloud-based predictive analytics software offers 24×7×365 autonomous monitoring of all system components.”
“After moving to a predictive maintenance model, now the customer knows which system components are beginning to wear out before they fail,” he said. “They’re monitoring things like compressor temperature, oil temperatures and levels, all remotely and in real time. This gives them the ability to know which system components need to be replaced before a failure occurs. Even better, field technicians know exactly which system component to bring to the store before they leave their shops. This decreases downtime during maintenance cycles and makes each store more profitable.”
Savings from speed
The savings can add up fast.
Consider a cigarette plant. For one customer, machines with improved sensors and controllers cost $175,000, Morgan said. The customer bought 50 machines and payback time was measured in months.
“The advanced machines made the process go that much faster,” he said.
Improved sensors and controllers can help companies monitor remote assets—assets that in the past might get checked only once a year, said Kyle Horsman, product specialist in Turck Inc.’s sensors division.
“We sell a linear device in the oil and gas market that goes on a machine that pumps crude oil out of the ground,” he said. “The sensor tracks to make sure the pump is running its full stroke of 20′ (6.1 m). The information is blasted from the control system over a satellite. If it gets below a certain stroke or is running at full stroke but taking twice as long, the company can send a technician out. When you’re talking about volume being pumped out, it can make a big difference.”
The well might be drying up; the cavity might be crumbling; or something else might be wrong, Horsman said.
“Previously, they may check on these things only once a year,” he said. “With the old legacy systems, there was nothing there giving them any type of feedback. By putting sensors and controllers, it’s greatly improved the chance of catching something. With continuous feedback, it’s live data. The biggest thing is: somebody gets notified. They or we can take care of it immediately. That can correlate to thousands of dollars to the customer.”
Sensors and control systems now provide valuable information on how to optimize production. A food and beverage company might discover that instead of a long cleaning cycle once a day, running shorter cleaning cycles every five hours helps the system run more smoothly, Horsman said.
In some cases, smarter machines can compensate for problems. “If a bearing starts to wear out, the drive automatically compensates,” he said. Even if the machine doesn’t compensate, it sends data about its performance to people who can quickly adjust, repair or shut down the machine.
These improvements make it much easier to customize production.
“One of the trends in manufacturing is a broader variety of different packaging types and customized products,” Hirschinger said. “One of the key things to support that trend is more flexible machines.
“Manufacturers want short runs of different configurations of products. In the past, you’d have to make a lot of manual conversions, mechanical changes and manual reconfigurations to reconfigure the machine for a new product. In some cases, it could take half a day,” he added. “Today’s machines are easily reconfigured to run different products. The machine automatically runs that new product without having to do any manual or mechanical changeout. The machine is automatically and quickly reconfigured based on the information downloaded.”
Consider a plant that makes diapers.
In the past, it might take hours to reconfigure the machine to switch to a smaller size, a different brand or different thickness, he said.
“On the same machine, you might produce a batch of generic diapers and a batch of Huggies. Some of the diaper lines have up to 200 different attributes of motion control. It could take a half hour per drive to reconfigure the machines. Now, machines are easily reconfigured to run different products. You just put in the new recipe. The machine automatically runs that new product without having to do any manual or mechanical changeout.”