Machine Vision in Container Manufacturing: Glass Containers

Believe it or not, one application now considered a machine vision application actually predates my high school years. Back in the 'early days' of this country we were not the disposable society we are now. In fact, some of us old enough to remember undoubtedly benefited financially by collecting and returning empty bottles, collecting the two-cent deposit for each bottle (long before a nickel or a dime). Well, a major concern with re-using those bottles was that they be empty of foreign matter and be whole bottles before filling them again.

It is understood that as early as 1948 RCA was given a development contract to build an electro-optical empty bottle inspection system. The system used a photomultiplier to collect the light passing through the bottom of the bottle. Any light shift from a normal range for the particular bottle color, whether higher or lower, indicated the presence of an unwanted condition - debris in the bottom of the bottle, crack, etc. By 1953, RCA had installed over 3000 systems on beverage filling lines largely in North America. It is understood that Barry Wehmiller became the licensee of that technology and today, as Inex Vision Systems (a division of IWKA), continues to offer empty bottle inspection systems for filling lines, although those systems are now based on conventional machine vision techniques.

Back in the early 70's, companies like Powers Machinery (now part of Emhart Glass) and Ball were offering electro-optical check detector systems looking for fine cracks in the finish (mouth/seal area) of the glass container. These early systems were augmented to include inspection for checks in other areas of the bottle: neck, shoulder, sidewall, heel and bottom. Others engaged in offering these early systems included Emhart, Owens Illinois (though only for internal consumption), St. Gobain and Kirin Technologies. Back in 1957 Emhart published a brochure on 'Glass Defects Causes and Corrections' targeted at manufacturers of glass containers. It included a list of 60 specific glass defect types. Most had to do with appearance, bottle integrity (cracks, birdswing, etc.), stress related (bubbles, blisters, inclusions, etc.) and geometry and were related to specific zones on the container.

Today, these early electro-optical systems have pretty much been replaced with machine vision-based systems, with the exception of finish inspection. The market for glass container inspection systems is segmented first by applications in the glassware manufacturer or bottle filler and then within these segments by application. In the case of the glassware manufacturer there are applications at the hot end and cold end and at the fillers for pre-filling and post-filling applications.

A number of companies are offering products targeted at glass container inspection; however, they generally focus on applications either at glassware manufacturers or filling operations. Not all companies offer products for both market segments. Significantly, although inspecting re-fillable bottles is not a big application in the States since most bottles today are one-way bottles, there is still a substantial market in Europe and the Pacific Rim for empty glass bottle inspection systems for beverage filling lines.

At the glassware supplier, the hot end applications include mold identification based on reading a 'slug code' or actual alphanumerics found either on the heel or the bottom of the bottle, freak detection or detection of misshapen bottles based on height, diameter and lean, detection of defects such as stuck and down bottles, stones, cracks, blisters, mold dope, inclusions, heavy mold seams, distribution, bottom thickness and bird swings. Removing these defects at the hot end prevents the costly production and processing of defective product. By identifying defective product by mold number in real-time, immediate corrective action by an operator can eliminate the production of additional bad bottles.

At the cold end, as suggested, the litany of defects is based on appearance, zone and geometric concerns. Conditions like cracks are a must to be detected as well as over press and out-of-round, offset, chipped and lineover in the finish area and throughout the rest of the bottle conditions such as blisters, inclusions, bubbles, stones, ribbon tears, wrinkles, mold dope, bird swing, fused glass and uneven glass distribution (washboards, load marks, drag marks, laps and tears). Many of these systems are adopting cameras with nominal 1000 X 1000 resolution, increasing sensitivity to smaller size defects, especially stress inducing defects like blisters, bubbles, inclusions, stones, etc. Some are based on line scan cameras yielding resolution up to 2000 pixels. Often these systems offer zone-based inspection with adjustable sensitivities for defect identification based on the specific zone.

Certified System Integrator Program

Set Yourself at the Forefront of the Global Vision Market

Vision system integrators certified by A3 are acknowledged globally throughout the industry as an elite group of accomplished, highly skilled and trusted professionals. You’ll be able to leverage your certification to enhance your competitiveness and expand your opportunities.

GET CERTIFIED

Systems also can inspect the dimensions of the bottle. While full 3-D measurements are not done on 100% of the bottles produced, systems exist that can sample inspect on line and provide a full 3-D scan of the bottle for dimensional integrity (height, lean, tilt, thread, body dimension, out of round). In addition to dimensional and color variables these machine vision systems may also have to contend with appearance variables such as decorated or embossed containers or containers that are not round. Machine vision systems targeted at applications of the glassware manufacturer typically can handle up to 800 bottles per minute, though because of material handling constraints throughput is generally under 600 per minute.

At the filler, machine vision systems can actually inspect the bottles in the crates to sort crates with foreign bottles. Other inspections possible on the crates of re-fillable bottles to avoid interruptions in production include: over-sized crates, bottles lying on top of crates, upside down bottles, wrong size bottle, wrong color bottle, bottles with wrong neck finish diameter, foreign objects in the bottle, foreign matter lying between or on the bottles and inverted cups on the bottles. Faulty crates can be rejected before causing a line disruption. Machine vision systems can also be used to actually inspect the crate itself to make sure it is the right color and incorporates the right logo as well as check for other lettering, length, height, handle. Empty crates can also be checked for foreign objects, bent or missing dividers, protruding handles and misshapen crates before they reach the packer and cause downtime.

If not in the crate before getting to the filling line, re-fillable bottles are often inspected to make sure they are the correct bottle for the specific beverage (based on shape, size/height or color or specific hot transfer label) and to make sure that the bottle is not badly scuffed and that there is no cap on the bottle. Before filling machine vision systems inspect: base/heel for chips, finish/sealing surface, thread, neck, if the neck is plugged, shoulder and sidewall areas for cracks, etc., scuffing and for conditions such as mineral ring and residual liquid. Detection of residual liquid may be based on infrared, RF or X-ray-based analysis.

The market for post fill inspection systems includes both glass and plastic containers. In the case of post filled applications some systems check for conditions such as: missing cap, misapplied or cocked caps, foreign cap, missing or broken tamper bands, over and under applied caps, product fill level, lying or floating foreign matter in the beverage itself, loose caps, missing cap liners. Others can also check for label concerns: straightness, damage, corner folds, off center labels, label height, label tears, correct label, etc. as well as date and lot code presence and aesthetics/legibility and conditions such as overfill or cocked pour spouts.

Still other systems are used as in-line foam analyzers determining the integrity of the contents by optically analyzing foam collar parameters in a sealed container to identify bottles suspected of containing foreign matter. Machine vision systems targeted at applications on glass bottle filling lines typically perform their operations at a rate of up to1200 bottles per minute. Throughput is a function of the size of the bottle. Typically the machine vision systems used on filling lines require NEMA 4 or wash down housings.

After filling and the bottles are placed in crates, the crates can be inspected with a machine vision system to make sure they are full, however, this is a check most often done by weighing the crates. Cartons with covers can be inspected using X-ray-based techniques to determine underfill and detect broken or leaking bottles as well as missing bottle and flap related concerns.

Attached are several tables that attempts to classify the applications addressed by the respective machine vision suppliers. These tables are based on information gleaned from their literature and should be verified.

Special thanks for input to this article goes to Rich Hebel, Insight Controls, and Don Cochran of Pressco.