Original Juice Co found itself with a challenge when it came to checking bottle cap placement onto a product after the filling stage. It wanted every bottle cap that gets twisted onto all of their different bottles to be applied fully, to be applied straight and not skewed, and to ensure that the tamper-band is not broken on products that require one. The company looked to install a vision system that could inspect a variety of imperfections not tolerated on high-speed lines and filling equipment. The shoulder-strength of the various bottles is very important, and must withstand significant head pressure and torque in capping and filling stages of operation. Any defect that could compromise the integrity of this area is an inspection attribute of the vision system. In addition to performing reliable and repetitive applications, the vision system would need to inspect at a rate of up to 300 bottles per minute to accommodate the production requirements of Original Juice Co.
To address this, Original Juice Co began working with SciTech, who initially performed an on-site trial. Its success proved that this project would help every bottle leaving the plant to be entirely free of imperfections. The trial focused on one bottle product only. On conclusion of the trial period, work began on implementing a full system, and also be scalable in order to accommodate any future bottle and cap types.
SciTech chose to use the Cognex In-Sight range of industrial machine vision cameras (pictured above). The In-Sight 5100 incorporates a die-cast aluminum housing and sealed industrial M12 connectors and achieves an IP67 rating for dust and wash-down protection on the factory floor. These environmental attributes would prove to be crucial in withstanding the wet, citric-acid environment of the inspection site as a result of the juices produced, as well as the cleaning chemicals used in the area. This also required that the fixtures and fittings were built using only stainless steel material. The overall system consists of a touch-screen industrial PC, incorporated into a stainless steel enclosure. The enclosure also houses the Ethernet hub, the digital power supply of the lights, a PLC and various power distribution components. After bottles have been filled and capped they travel down the conveyor line, where two cameras sequentially inspect the bottles. The first camera looks directly at one side of the bottle and inspects the bottle cap from this side only. A red LED backlight provides the camera with a silhouette image of the bottle. Backlighting provides maximum contrast between the product outline and its background, and is ideal for measuring external part edges. This results in images that work extremely well for the vision sensor’s measurement and inspection tools. When the bottle comes within the camera’s field of view, a sensor is triggered and an image is taken. The Cognex In-Sight software tools then analyse the image for defects and determines whether a bottle is defective or not. In the event of a failure being detected, a fail signal is sent via one of the camera’s outputs to the PLC. The PLC then triggers a reject mechanism, which removes the bottle from the line. After passing the first camera, the bottle will travel a little further before the second camera acquires another image.
The second camera, mounted similarly to the first camera except at the opposite side of the conveyor, focuses on the other side of the bottle cap. The same inspection criteria apply equally to this camera. All defective parts are knocked off the conveyor line into a reject bin. A red beacon also becomes illuminated for five seconds when a bottle defect is noticed, notifying the operator. Good parts are simply allowed to continue, unhindered, on the production line. During this inspection a pattern of lines is projected on to the bottle cap and bottle, and the vision system is used to detect any deformity in cap height on the bottle, cap presence or absence, tamper-band presence and quality, and cap skew. To complete these tasks the vision system uses edge detection and histogram software to measure the cap, analyse the angle of the cap and determine cap presence.
Because of the number and variation of bottles involved, easy changeover was essential. To this end, the cameras, backlights and sensors were mounted onto one fixed bracket that could be moved vertically, via a turn-wheel, by the operator to accommodate the different height requirements of the bottles involved. The changeover procedure requires a height-determining tool to be temporarily attached to the fixture. Using a turn-wheel, the operator screws the tool down to a position whereby the tool rests gently on top of the bottle lip. When the tool rests on top of the bottle lip, the correct camera/light/sensor position for that product has been set. On completion, the operator removes the height-determining tool from its slot.
The final step in the changeover procedure is when the operator makes a selection, via a custom application, using touch commands on the computer’s screen. This screen uses a tab-style interface to segregate the different bottle types appropriately. Each tab has a descriptive name indicating the different bottle categories. On selecting the appropriate tab to the bottle type being run, a corresponding button appears that when pressed loads the correct vision file associated with the new product being run on the line. On successful completion of this step, the main software interface screen will contain the descriptive name of the bottle type selected. The changeover procedure is now complete and the line is ready to run.
When the line is running, a custom application on the VGA touch-screen monitor shows the last failed image acquired from each camera. The cameras also FTP their data back to the PC, which is stored in text files. This provides quality engineers with more information about their process.The vision system’s key asset is preventing defectively sealed bottles from being shipped to customers. Most importantly, the system ensures that every bottle leaving the plant is free of imperfections.
In the can
The EU legislation on disposable packaging is changing in May this year, when retailers that sell beverage cans or disposable bottles will be obliged to take back all empty disposable beverage containers. With experts predicting a comeback for the beverage can, the market looks good for high-speed image processing systems used in the manufacture of these cans to ensure quality.
The US group Ball Packaging Europe manufactures beverage cans, with one of its can-end factories based in Braunschweig, Germany. The quality of the aluminum can ends produced there is ensured by TCVision, a high-speed image processing system from Puchheim-based company Quiss, which uses image processing components from Stemmer Imaging.
Each machine in the production hall in Braunschweig churns out 2,400 aluminum can-ends a minute. The can-ends are produced in a process that involves several steps. First, the blanks are punched from an aluminum sheet. The can end is then moulded into shape in successive steps, and the tab, which is moulded simultaneously, is then attached to get the finished article.
Before they are packaged, every can end is inspected using the TCVision image processing system. ‘Random samples used to be taken to check the quality of the can-ends,’ says Helmut Gruber, the sales and marketing director of Quiss’s Packaging division. ‘However, because of the products’ field of application, that is no longer enough. The quality of metal packaging must be extremely high. The economic conditions and high inspection speeds required made it almost inevitable that high-speed image processing systems would have to be used.’
The TCVision system checks the can ends for problems such as malformation, scratches, impurities, forming errors, burring, edges or other flaws. In addition, the position of the tab, the overall diameter and the rivet diameter are checked. ‘There is a combination of checks involving contour and surface analysis and presence and position monitoring,’ continues Gruber. ‘The demands are considerable. The can-ends run past the inspection station, on three parallel lines, at a speed of around 800 can-ends a minute. For some criteria the detectable tolerances are a mere 15µm.’
To meet these technical requirements on the image processing side, a combination of suitable image processing components were required, in addition to specially-developed software.
The progressive-scan CCD cameras M10 and M4+CL from JAI were used, and PC-Vision and PC-CamLink frame grabbers from Dalsa Coreco were used for rapid image capture. The cameras work under a movable cover with application-specific lighting consisting of standard LEDs, with flash control developed by Quiss to meet the requirements of high-speed inspection.
The system can detect a number of different flaws on the can-ends, reject flawed parts and shut the press down immediately when certain defined types of flaw occur. The system is fully configurable. It is possible to control its behaviour to suit the type of flaw detected and, in the case of slight flaws, for example, to permit five flawed parts per thousand. Serious flaws, on the other hand, can be defined as criteria for an immediate shutdown. Defective parts are always rejected.
The TCVision runs cans through the inspection station at a speed of around 800 cans per minute.
The TCVision analysis software checks the test object on the basis of a contour model that can be compared with a CAD drawing. Deviations from the required contour and flaws in the surface can be detected repeatedly, even in the sub-pixel range. Consequently, not only can a wide variety of checks of dimensional and positional accuracy of contours and surfaces of all shapes be carried out; but it is also possible to check for malformation, dents, cracks, impurities and completeness. This new type of analysis permits the robust and repeatable detection of defects that image processing systems have previously failed to detect. This makes it possible, for example, to check other designs such as gold-coloured can-ends or other variants.
The software also allows the user to create new product types or alter the inspection parameters. The contours are detected by the system in advance, and the user simply has to select the contours to be inspected. The user can assign them go/no-go criteria, such as the maximum size of crack that can be tolerated or the maximum and minimum rivet diameter. ‘It is thus possible for the system to learn arbitrary free-form areas and curve geometries within a short space of time,’ stresses Gruber.
With a network connection it is possible to load parameter sets and images from the production level and analyse and optimise them offline. The updated software can then be transferred back to the production system without interrupting production. As the system is networked, it can be operated and monitored entirely from the office, or by means of a teleservice or remote control. Thus, even camera settings can be transferred to all lines in this way.
In addition, the production process can be optimised and documented on the basis of the data obtained. The system also offers various statistical functions for the visualisation and data analysis of the results obtained in the inspection process, as Gruber explains: ‘The results of the analysis over time can be displayed in a bar chart, for example. The trend view allows the proportion of faults and fault classes to be displayed.’
The statistical functions are assigned filter criteria. It is thus possible, for example, with a mouse click, to extract and display line-related, station-related or fault type-related data. You can also set warning thresholds and display them in the statistics. Ball Packaging Europe is highly satisfied with the results of this inspection station. ‘In addition to providing 100 per cent, documented quality, the detailed analysis and visualisation functions of TCVision allow changes that take place in the process to be detected quickly and in good time,’ says Gruber. ‘You get not just quality control but also the opportunity to improve the whole process and dramatically reduce the reject rate.’
Finally, Sick IVP has also developed a smart camera system suitable for use in canning applications, where it is essential that the lid of the can is correctly attached and that there is no damage on the top surface. The IVC-3D camera can be used to check the surface and height of the cans for defects. The camera combines imaging and analysis into one camera housing, using the same image processing tools used for 2D cameras. The smart camera can be set up using a PC, and can then operate standalone or as part of a network. Inspection results can be sent directly to a PLC or handling equipment, and results monitored via
Krones has introduced the Linatronic 735, a compact empty-bottle inspector with high-precision IRIS technology and ‘BIRD’ intelligent image processing for avoiding incorrect rejections, plus hygienic machine design with inclined outer surfaces. The inline machine inspects glass and plastic containers with diameters from 50 to 110mm and heights from 60 to 400mm for damage and soiling, so that the containers in which the product is subsequently filled are flawless in terms of both appearance and hygiene.
The requisite inspection units are selected to suit the particular application category involved. They can also be retrofitted at need. The standard inspection units provided are base, sealing-surface and residual-liquid (HF/IR)-monitoring. As optional inspection units, the client can integrate check routines for side-wall, inside wall, thread, lateral neck finish or foreign bottles. The automatic type change-over feature means that no handling parts are required.
Container sorting with the Sekamat system has now been upgraded to include a colour camera for differentiation by means of the bottle’s colour, thus providing enhanced sorting convenience. In addition, the revamped linear rejection unit has been constructed to meet stringent standards of hygiene.
The upgraded Sekamat inspection system monitors the bottles by criteria like height, diameter, contour, colour or glass embossing. It also inspects them for the presence of closures, and detects any full containers. It differentiates between PET and glass bottles using a high-resolution CCD colour camera, whose signals are processed in realtime using DART technology.
If bottles do not conform to the production specifications, they have to be rejected. In the new Sekamat, this is preferably accomplished with a design-enhanced version of the linear rejection unit. For linear rejection, the containers are pushed gently sideways to the end position onto a parallel conveyor. The Sekamat can be used at a machine speed of up to 72,000 containers an hour.
