Detecting can edge defects right at the source: 100% inspection on the production line with the KEYENCE VS Series

The Cimanggis plant is no stranger to automated inspection. Before the VS Series, the plant used camera sensors on the swaging line to inspect battery can edges for dents and deformations. The issue was not the sensor’s ability to detect, but the limited information it could provide.

Camera sensors only provide binary decisions. They can only tell the line whether a part is OK or NG, and nothing more. For diverse defect types such as dented can edges, it became difficult for the team to develop inspections as inspection standards became stricter. Their previous vision equipment came from another manufacturer overseas; the results were acceptable, but when problems occurred, parts and support had to be sourced from overseas, and resolution took time.

The biggest cost impact, however, appeared at the end of the process. Dented cans that slipped through were not found on the production line at all — they were only discovered a week later, after aging tests, when the cans had already become complete batteries with jackets attached.

“Previously, we had to wait for aging for a week, so it was only detected when it had become a finished product. So the battery was complete, with a jacket already on. Now, with this, we only check the can itself, we can take the can, and then repack it. If it has already become a battery, then it becomes scrap. And for those batteries, we also have to pay to dispose of the scrap.”

When asked what his considerations were when choosing a vision system supplier, Abdul Basit, Electrical & Facility Engineer of the plant, placed support as the first priority, price second, and features third. What ultimately decided it, though, was not the spec sheet but whether his own team could operate the system without needing to contact anyone.

“Some cameras might be fully featured, but sometimes my team struggles with the settings. With this, we try to learn once or twice, and after that, we can do it. So for this KEYENCE camera, we already know 80% of its features, we know the settings. So we don’t have to ask the vendor for every little thing.”

The team’s ability to perform calibrations independently also helps reduce costs when issues arise. When the team can adjust inspections themselves, product changes or process deviations do not always require assistance from a technician team. They still consult with KEYENCE for things that are truly beyond their experience, but routine work is handled internally.

The second deciding factor was how purchases were evaluated. Instead of choosing from a spec sheet, the plant first discussed the application with KEYENCE: samples were tested, and units were installed on the actual production line before an order was made. A purchase order was only issued after the production team approved it.

“So before we ordered the product, we were probably 90% confident that we could do this. The production team was okay, approved it, and only then did we open a PO. So that was quite helpful for us.”

Because the application was already proven on the customer’s own samples, installation generally went smoothly: guidance was already available, the unit just had to be installed, and production was usually back up and running in one to two days. Post-sales support was also rated as meeting the team’s expectations: they could contact the sales team and, when needed, technical experts directly, with remote assistance when visits could not be made immediately.

The most visible change was the number of products that could be inspected. The line runs at 600 pieces per minute — 36,000 units an hour. With the previous method, an operator only took a few samples every hour.

“Ideally, we perform sampling inspection of 7 pcs per 1,000 products. But our production is 600 pcs per minute, so in 1 hour that’s 36,000 pcs, and if we only take 3 pcs from that, it’s not enough to guarantee the production lot.”

With the VS Series inspecting every unit, the quality of each product is checked and the results are recorded, rather than relying on a few samples. The system can also be set to automatically stop the machine when it detects a predetermined number of consecutive NG products, instead of relying on the operator to notice and intervene.

This has a significant impact on the inspection process. Manual visual inspection on this line meant one person was stationed at the conveyor, rotated every four hours across three shifts. Even with training, judgments remained variable: those that were consistent in the first hour had drifted by the next. Different people also applied different thresholds under different pressures. Automated inspection eliminates those variables, and the three positions previously assigned to this check across three shifts can be redeployed elsewhere in the plant.

Defective products that slip through the inspection process can have a direct impact on customers. A battery that is supposed to last two years might only last a few weeks in the customer’s hands if there is a problem with the can, and with the Energizer name on store shelves worldwide, the plant recognized that this would not only incur warranty costs but could also damage its brand reputation. Before cameras were implemented across the entire swaging line, a batch that had been stored for a long time led to leakage complaints from overseas customers. Since then, according to the team, leaks have become rare.

There were also benefits no one had anticipated. The camera was installed to look at the edge of the can; it turned out the camera could see more than that.

“We only focused on the can lip, but we also detected misplaced or folded components. And that saves us almost one pc per thousand. So if it’s one million, that’s a thousand batteries saved. And that wasn’t our parameter before. So in the future, that might become our parameter.”

The system also captures conditions such as off-center electrodes — not the plant’s primary concern, but now visible when they weren’t before. As Basit said, the inspection that was originally set for only one parameter turned out to be much more detailed.

Although stricter inspection increases the number of NG products detected during assembly, this is still considered positive because problems can be found earlier. Consequently, the number of NG products at the end of the process that must be discarded is also reduced.

The project focus at this plant has now shifted from the edges of the cans to the printing on them. Date and batch codes are currently applied using ink, which has several drawbacks, such as operators forgetting to refill the ink, faint printing, or code errors that are not immediately detected.

“We use ink, and we often face issues with that ink. Operators forget to refill the ink, and then it gets faint. Early detection is not fast, because several times it happened that we only found out after several pallets. Then there are also wrong codes. If we use a laser, for example, and then we can record it, that might be better maintained for us.”

The team is currently in discussions with KEYENCE regarding laser marking and code reading, as well as expanding the use of the vision system to detect smear-type defects that are currently inspected with measurement sensors. Camera trials for that application are currently underway.

When asked if he would still choose KEYENCE at other locations, Basit pointed to another location within his group: the company’s second plant in Indonesia already uses KEYENCE, and when affiliated plants overseas ask for recommendations, his team directs them to the equipment they use themselves.