You can only have one aspiration. Suiker Unie in Groningen wants to be the nicest and smartest sugar factory in Europe by 2025. How will they do that? By working on the Smart Factory. But processing a natural product in an automated way is quite a challenge. Peter van Dam, Suiker Unie's technical services manager, explains a recent project: optimising the cutting mills.
What Suiker Unie does? The company processes sugar beet. That means its production is what is known as a campaign business. Depending on when the sugar beet is harvested. When the beet is ripe, there is the time of harvesting. Beets are then harvested, stored and processed. Beet processing is done in bulk, regardless of the delivery, variety, soil type from which the beet originates, etc. Moreover, during a certain period of the year, grass and frozen beets are processed mixed. That means a relatively unpredictable process. What can intelligence add to that?
First of all, they migrated the DCS system with Schneider Electric. In addition, they worked with Emerson on asset monitoring. People wanted more than just primary control of the plant. Van Dam: "We then started with the so-called low-hanging fruit. Optimising the cutting mills. In doing so, we wanted to take an in-depth look at the so-called Sugar OPE, the Overall Process Efficiency. Until now, it was actually the case that the knowledge and experience of the operator was the main input for process operations. We wanted to improve those operational decisions by deploying intelligence. With that, you make the decision independent of the person and therefore the quality of your process and your product becomes more constant. But in doing so, of course, you have to get the operators on board, because the quickest way to get a techie's heart racing is to not involve him in an optimisation process."
Cohesion
"Cutting mills work in groups of five and they jointly deliver to one brewing trough. The interrelationship between the cutting mills listens very closely. To see how best to deal with this, we first organised a workshop with the people from the factory. In a brainstorming session, we determined which parameters and data sources were relevant for the analysis of the cutting mills. The paramaters that were examined were the number of rotations per minute, the torque, the power required and the amount of operations."
They then chose to put one sugar mill under the microscope. "This mill we started analysing and modifying extensively to see what the effects are. This involved paying attention to everything from vibrations to the number of stones coming out of the mill. Only then did we start scaling up with the other mill. We wanted to know if we could predict what problems might occur in which batch. We then started analysing a number of factors, the spacing of the blades, the different shifts - morning, afternoon and night, the shift schedule, the status and action logs and Wonderware's Historian programme. That data was then processed into information and relational data and from that came some surprising data for us."
Predictable
Van Dam continues: "For instance, we were able to start predicting the remaining cycle time of a cutting mill. But other things, which we had expected, suddenly turned out to be much less predictable. A cutting mill has a stone hatch, through which stones are fed out of the mill. The expectation would be that a quick change would then occur after running a stone, but there appeared to be no direct relationship with this. The mix between grass beet and freezer beet turned out to have a much greater impact on blade wear than we thought. And with thicker cutting material, the torque was found to have much less variation. That had a bigger impact on the consistency of the cutting than we had suspected."
"In short, we found many new insights into the influence of various parameters and also how cutting mills influence each other. A mill with recalibrated blades has an impact on performance of the other cutting mills in the same group. And the role of the operator of the cutting mills also turned out to be greater than we expected."
"The methodology chosen for analysis based on Proof of Concept has reduced the risks and at the same time strengthened shared learning. We have now developed a solution through which we expect faster and better decisions in the upcoming campaign, called the Operator Cutting Mill Assistant. With automation, everyone has access to the same data at the same time."
By: Janet Kooren
