by Roeslein Roeslein

From the September issue of The Canmaker Magazine – Being able to predict when equipment on canmaking lines is likely to fail with better precision is expected to help a new beverage can plant be more competitive. Rick Pendrous reports

A beverage can plant being constructed in Florida will be the world’s first in the industry to use a predictive-analytics system when commercial operation starts later this year.

The $120 million project at Winter Haven managed by Roeslein & Associates is for Florida Can Manufacturing, which owns Caribbean Can in Puerto Rico, will eventually be capable of making 3.2 billion cans a year when a second production line is installed.

Operators at Florida Can Manufacturing will be using processes developed by Roeslein that are part of the St Louis, Missouri-based systems integrator’s data management system (DMS). Using specially-written algorithms, they will be able to identify what canmaking equipment components are most likely to fail and when they are likely to occur along with the confidence levels associated with that prediction – down to the hour of it happening.

The plant is part of a three-phase development project on 800,000sqft of land near Winter Haven’s CSX Intermodal Terminal, whose facilities will also be used for bringing in raw materials and exporting cans. The canmaking plant is being built in the first phase of the project on 300,000sqft of land and is expected to have an initial capacity of 1.6 billion cans a year.

The predictive analytics package that has been added to Roeslein’s DMS had originally been intended to be installed at Florida Can’s plant last year, but the project was delayed by the coronavirus pandemic. In addition to daily reports on equipment for supervisors and managers of canmaking plants, highlighting key performance indicators (KPIs) and identifying the top five causes of downtime for machines, predictive analytics will also identify when problems with particularly susceptible items of equipment or components – such as low oil pressure or problems with pressure switches on bodymakers, for example – are likely to occur.

“We are excited about the analytics and we have tested it and verified it against existing data,” says JC Harrison, director for systems engineering at Roeslein. “What this predictive analytics will do for Florida Can and other people in the future, in addition to standard KPIs in those reports, is use historical data to predict which downtimes will occur next in a particular day. And we have got the accuracy down to within the hour. In addition, we will give a confidence level for the downtime on that machine.”

This will enable Operators / Engineers to attend to potential problems on the most critical items of equipment where the confidence of impending failure is highest, adds Harrison. Where confidence levels are low, they can instead concentrate on other operational priorities, while having the risk of potential failure on their radar.

“We do have algorithms inside the program that are making these predictions but in order to make a good prediction you need more data,” he says. “The more data you have, the better your prediction and your analytics are going to be. We are giving the top five things that may occur in that shift. If you have a very efficient-running can plant the last two may be something that occur once a week, so it is hard to predict when they are going to occur.

“Now, if you have a can line that is running in the 60s or 70s [overall equipment effectiveness (OEE)], the equipment is all the time faulting and this will absolutely give them information to help improve their systems. But if you’ve got a 93 to 95 per cent running plant already, probably only the top one or two of the downtimes are something they are going to pay attention to because if a can plant is running that well they are not going to have a lot of downtime anyway.”

Harrison believes this is the first time predictive analytics has been used for canmaking operations and is convinced that it will help to improve line efficiencies. It is the latest of several recent advances that Roeslein has added to its DMS over the past few years. In response to a request from one particular canmaking client last year, Roeslein also introduced quality assurance (QA) and statistical process control (SPC) modules to DMS and OEE systems at two of that company’s plants in Central and North America. These systems are specifically tailored for use in canmaking plants and are primarily used for front-end gauge checks on bodymakers but they also undertake some subsequent checks on cans after the spray machines as well.

“We didn’t provide test equipment, but we interfaced with the test equipment and we have been able to put that statistical data right in with the same trends and the same reports as the production and downtime reports,” says Harrison. “They already had an outstanding SPC system, but they wanted to be able to bring all that data into common screens and reports and something that is easier to support. And since they already had our DMS doing all the OEE and downtime, it made sense to go ahead and try to move their existing system into our system and it worked out really well.” Because of the success of these two projects, the customer is considering installing the same system at new beverage can plant in the US this year, Harrison reports.

Roeslein has remained very busy throughout the coronavirus pandemic as demand for aluminium beverage cans has soared, installing 15 new can lines in 2020 alone. The company is currently working on projects with all the major canmakers and is also in the process of installing four DMSs.

More information from Roeslein & Associates, 9200 Watson Road, St Louis, Missouri 63126, USA.
Tel: 1 314 729 0055.

Website: www.roeslein.com