Industrial bakeries have the opportunity to collect plenty of data about their finished products and the manufacturing process. PLCs on every piece of equipment as well as sensors throughout the production line pick up and store thousands of data points. However, without analysis that data doesn’t serve much of a purpose. Without a human employee looking at trends or evaluating the meaning of that data, it doesn’t really help a bakery that much. But even humans have their limitations.
“As humans we are somewhat limited in our ability to digest so much information,” said Liran Akavia, co-founder and chief operating officer of Seebo. “We have other things to do, and we only have 10 hours of work per day.”
This is where artificial intelligence can come into play. An algorithm, Mr. Akavia pointed out can process hundreds of data facts against a bakery’s goal continuously, and it also does not rely on human interpretation.
“What we see in many plants is that different people looking at the same data will come to different conclusions,” he said.
Seebo developed its AI as an algorithm that can conduct root cause analysis continuously to help bakeries improve the efficiencies of their production lines. Often this looks like reducing waste, as was the case with Barilla, Parma, Italy.
In a member research webinar presented by the American Bakers Association, Mr. Akavia and Giovanni Ballerini, vice president of global manufacturing strategy and global capital planning, Barilla, explained how the food manufacturer used Seebo’s AI to reduce product loss on its bakery line in Rubbiano, Italy, by 37% in a matter of months.
The Rubbiano bakery line produces Barilla’s Fette Biscottate rusk product. The plant produces hundreds of the popular product each minute. Mr. Ballerini said they chose this line as the test for the Seebo AI because the potential financial benefit of improving waste was significant.
“It also had scalability potential,” he said in the webinar. “It’s one of several rusk lines we have so if we could prove the value here, we could roll it out quickly to other lines.”
While some process parameters may be obvious to track and change, the algorithm’s ability to make connections and get to the root cause showed Barilla inefficiencies it would have never caught had it relied only on employees.
“Baked goods manufacturing is uniquely complex,” Mr. Ballerini said. “There are lots of interrelated points and data tags on the line: the speed of the conveyor, the heat source, the rolling, the cooling. Losses aren’t caused by one or two of these factors but many of them, and no human can pick up on the combination.”
The root cause analysis showed Barilla why the losses were happening, and the operations team could prevent them. In that way, Barilla was able to reduce waste on the line by 37% in a matter of months.
The key, however, Mr. Akavia said is to have a clear business case.
“We don’t collect data for the sake of collecting data,” he said. “We collect data to solve a business case. When you know what your business problem is, we can focus on collecting relevant data from the PLCs on the equipment.”
While this can require bakers to invest in data collection infrastructure they may or may not have, it can be simpler than anticipated. For Barilla, that meant a visual inspection system that detected parameters like size, color and weight of finished product. For other bakeries that might mean simply pulling data from PLCs that are already employed by equipment on the production line.
AI isn’t out to replace line operators either. In Seebo’s case, Mr. Akavia said they recommend most often using the AI to enhance the bakery team, not replace it.
“At Seebo we are most interested in creating the understanding for the operating team,” he said.
It can even empower bakery experts to support multiple production lines around the world remotely as the data is stored in a cloud-based system.
“A baking company can have their production experts sitting in their office or even at home and they can see a production line, analyze it and provide recommendations,” he said. “So an expert, wherever they are in the world can support multiple bakeries. The analysis from the AI happens automatically and every process expert and operator will get the same result.”
Baking companies continue to see increased demand for their products, and they are looking for ways for their production lines to keep up without sacrificing quality. Add to that a labor shortage, and meeting today’s challenges looks difficult. But Mr. Ballerini, originally an AI skeptic, pointed out the data is where the future lies.
“Data is one of your most valuable assets if you want to remain competitive, so you don’t have a choice,” he said. “We have more data than we could ever process or understand, but if you want to solve a problem you need to take all the data into account.”