Should Your Plant Switch to Automated Parts Inspection?
Jul 19, 2019 Comments (0)
Quality assurance is a very necessary part of any production line.
Without it, the wheels would come off of our bicycles in the middle of the street, babies would go hungry when their bottles clogged and grabbing a piece of fruit off the stand in the store could lead to instant stomach illness.
Seeing out the important task of quality assurance has always been a responsibility of humans. They dedicate hours, days, years of their lives even to ensuring that a given product consistently meets the quality standards consumers expect. It's difficult. Consumers are fickle, standards become more stringent all the time and humans are not always correct in their evaluations. They make mistakes. So, how to remedy this inherently human problem?
Removing the Human Factor
Computers seem like the obvious answer. In some places, they're already carrying out the job of quality assurance. It's called Automated Parts Inspection (API) and as you may have guessed it works great to verify that the sprocket on your bicycle does indeed have 36 teeth, that its radius measures exactly 2.868 inches and that it weighs out to the same number of grams of all it's fellow sprockets. These things are important. If a cyclist who patronizes your brand buys a new sprocket only to discover that it's missing a tooth or doesn't fit correctly on their crank, that's a problem. You could lose business.
So to deliver the highest level of consistency and eliminate the costs associated with having humans count every tooth on every sprocket, a computerized model of the part is made. Robots on the production line efficiently scan each new part to take its dimensions. The part is then compared to CAD template of a perfect sprocket. If a defect is discovered, logic in the production line automatically removes the defective part. It's undoubtedly more accurate than paying humans, who would require far longer, might get tired and stop paying attention, or even occasionally allow a part to go by without checking by accident. But this solution can't work for every plant.
Important Factors for Switching to Automated Parts Inspection
In the example above, we used a bicycle part. Perhaps that part would be manufactured by a component maker that only fabricates sprockets. Or perhaps it's being fabricated by a company that assembles complete bicycles. The two applications are extremely different from the standpoint of API.
Converting a plant that makes only sprockets to API would be relatively straightforward. There would be a high initial cost for equipment, however mapping each size of sprocket would be fairly straightforward. The different size sprocket templates could be synchronized with production equipment and a small number of robots would be able to accommodate all of the plant's QA needs. The initial investment would quickly be recouped from the savings accrued over the first few years of not hiring humans to perform QA tasks.
However, in a plant that makes every single part on the bicycle, things become more complicated. Because the various bike components are still fabricated according to design templates, API is possible in this scenario but it's more costly. You may need a hybrid approach. A completed bike cannot be test-ridden by a computer. The equipment needed to load-test a front fork to 450lbs. is not the same as the equipment for evaluating the sprocket.
Taking the expression a step further, API would be difficult to implement for a paint-mixing company that receives custom color requests intended to satisfy individual pallet selections for new homes. The sensor technology might exist, but you'd have nothing to compare the finished product to. You would have to create a test swatch, upload it and then compare the mixed paint to the initial one all as a matter of process for a single customer. Let's look at some real-world examples to further unpack this topic.
Inspecting Cologne Bottles for Fullness
Have you ever considered how fragrance manufacturers quality control their bottles to ensure that no one gets a half-empty bottle of cologne? You might think there's a physical sensor involved, but the process is much more high-tech.
The inspection occurs using light. Specifically, light that's not visible to the human eye. Only about 1% of all light that reaches earth's surface falls into the visible spectrum. At the edge of the visible light spectrum, infrared light uses a wavelength of 700nm and up, which can penetrate glass and reveal what's inside a container better than other more visible wavelengths. In industry, an 880nm infrared backlight and mirror are used to verify the fullness of cologne bottles on the production light.
Drones Used on Airliner Production Lines
What if your application requires you to inspect something massive? Like a jet airliner for example? French company Donacle has created a drone with access to a database of 3D images of the exteriors of modern airliners. When a new plane rolls out of production, the drones can check the plane's exterior for rivet rash, uneven paint and other signs of errors in the build process. But these drones aren't just intended for inspecting new planes.
Twice a year, in-service airliners must be checked for lightning damage. The stringent safety standards that air travel requires mean the plane's entire exterior must be checked for signs of a lightning strike. A job that would take a human crew six hours to perform is reportedly accomplished ten times quicker with help from the Donacle drones.
That's a real difference in the time required to get a plane back into the fleet and making money. And what's more, the drones are more consistent than their human counterparts. They're already outfitted to handle the popular Airbus A320 and being put to use by European companies like Easyjet and Air France.
Evaluate Your Options
In each of these cases, a highly replicable item is validated against a set standard. A computer can evaluate sprockets, airplane components and measure the fullness of bottles off a production line. Since workers can define a set standard, these instances make good use cases for automated inspection. Now it’s your turn to evaluate your own production plant. Do you have a product that is highly replicable? Can you easily measure for quality control? If so, you may benefit from automated inspection. If you have a case study of your own, share it in the comments section below!
(No biography information for Megan Nichols)