When Alten suggested a chicken slaughter line manufacturer implement predictive maintenance, the software service provider initially met with skepticism. "Predictive maintenance sounded nice," the customer said, "but why replace expensive parts earlier than necessary? Why fix something that isn't really broken?"
Because of his long-standing relationship with this machine builder, Alten still got the confidence and green light to get started. Meanwhile, predictive maintenance is helping the machine builder cut costs by minimizing maintenance costs and ensuring uptime at end customers within tighter limits.
The customer's initial reluctance to replace parts too early is quite understandable. There is a good chance that some costly items will continue to run for months without intervention. Even if they wear out, use just a little too much energy or are already well past their useful life. In general, the credo is: rather not spend money if it is not really necessary.
After all, keeping a chicken slaughterhouse running is careful balancing. This business is all about margins of pennies per product. Every minute counts. Not for nothing are service level agreements (SLAs) very common in this world: slaughter line suppliers guarantee uptime. If the lines are down, they share in the damage suffered. To ensure continuity, many parts are kept in stock at abattoirs. Examples are the knives for cutting chicken fillets. If those are blunt or they are at the end of their life, the slaughterhouses choose an appropriate time to replace them. Bolts and nuts, v-belts and other small items are always in stock.
The trade-off is more complicated when the price of spare parts runs into the thousands of dollars. A catastrophic problem - such as the failure of a crucial engine - can shut down a slaughter line entirely, but stocking tons of expensive spare parts at each customer site is also expensive. "These are complex trade-offs," says Richard Bernards, business manager at Alten. "You can stock everything, but there's no guarantee that critical parts will work when they come out of the box after one or two years."
This love for technology, which Marina felt at a very young age when she began to take an interest in how machines work, surprised some colleagues when she became the first female expert to participate in the project! "The perception of society is changing," she stresses. "I quickly felt comfortable in my role because I quickly proved that I had the skills the team and the project needed."
Until recently, machine suppliers were looking for a balance. Inexpensive parts were kept in stock; more expensive parts such as motors were only "flown in" when the need arose. Restarting a slaughter line in such a case may cost hours, but that was taken for granted. Alten suggested to his customer to start predicting catastrophes with more expensive parts. If a complex system part then threatened to break down, it could be replaced at a time when the slaughter line was down anyway. That story proved harder to sell than imagined. Says Bernards, "The main argument was: we're not going to spend money if machines aren't broken."
In the customer's perception, Alten's proposal created a new problem. He was going to incur costs at an earlier stage. Bernards: "They had a different idea of predictive maintenance. They imagined, for example, that they would get a signal the moment an engine had been running for three years.
If it was still running well at that time, it would be a shame. However, we saw opportunities to predict the time of replacement fairly accurately."
Alten proposed to start discovering patterns in sensor information from the slaughter line and use that to make judgments about whether something is about to break down. The software house had built up a good relationship with the machine builder in recent years. Despite his reservations, the latter gave Alten the green light to dive into the matter and investigate whether there might be opportunities here.
One of the major advantages proved to be the overwhelming availability of data generated by existing sensors in the relevant chicken slaughter lines. For example, hundreds of sensors monitor the quality of drumsticks, wings and fillets through temperature measurements. If the value is too high, the meat is immediately ejected from the line.
Bernards: "There is plenty of data. For example, a slaughter line also makes choices and executes actions based on sensor data. There is also data available on power consumption and engine speeds, light locks see passing chickens, and image sensors provide much-needed information for filleting."
Numerous sensors turned out to be able to provide more information than was needed until recently. For example, each microprocessor is equipped with a sensor that protects the electronics from overheating. If the temperature exceeds a specific value, the microprocessor shuts down.
This information was used to enrich the already available data and thus increase the quality of forecasts. The abundant availability of very diverse data proved to be a huge advantage. In fact, the accuracy of the prediction lies mainly in combining, matching and comparing the data. If a motor is spinning faster, it may no longer feel resistance due to the lack of a transmission. It could also be that an internal shaft is broken. "A speed sensor only tells you that the revolutions are increasing," Bernards said. "To know what's really going on you need other data as well."
To properly assess future failure, however, smart matching during data collection is not enough. "You can't say: if incident 1 and incident 2 occur at the same time, you have to stop the machine," Bernards says. "You also have to look at the historical data, recognize patterns in it of situations where catastrophes occur and use that to estimate what kind of action toward the future you need to take." The big advantage was that this customer was already capturing large amounts of information in its latest generations of slaughter lines, either centrally on a PC or in the cloud. This eventually led to a system that combines data, learns itself and keeps getting better. From that mash of data, it is possible to estimate the probability of failure. "We train the model to make a recommendation to replace something quickly, says Bernards: "We can't say that something will fail this afternoon at five past four, but we can say, for example, that there is a 90 percent chance that a specific component will fail within now and two weeks."
Predictive maintenance helps avoid SLA penalties, ensure uptime and keep maintenance costs down. But there is still plenty of room to set priorities. For example, the machine builder can suggest to the end customer that, for a specific amount per month, he guarantee that the slaughter line runs 99 percent of the time. After all, if the guarantee is 100 percent, he has to go to extremes to incur costs while risking replacing parts too early. Bernards: "Predictive maintenance allows you to choose the right time better. But you can still make the trade-off between intervening quickly or going to the edge." Bernards expects the technology to bring benefits to both the machine supplier and its customers. " For example, customers get a discount on their SLA if they take software version with predictive maintenance, or higher uptime."
Brainport Digital Factory works every day with 23 active members (business and technical education) to realize the Digital Factory. We have a modern open systems manufacturing platform, on which we apply new Digital Factory technology such as AR, Low code machine models, (private) 5G, precise localization and AGVs.
By combining in-depth domain knowledge from our members, arrive at highly innovative business and engineering solutions that concretize the digital factory for both SME and large-scale manufacturing companies.
By combining in-depth domain knowledge from our members, arrive at highly innovative business and engineering solutions that concretize the digital factory for both SME and large-scale manufacturing companies.
No posts found!
Dutch