5

I am researching about ways to continuously determine black spots (size 50+ micron) in our product stream. This stream consists of imperfect spherical polymer beads (size 3 mm) of different materials (ABS, SEBS, etc.) and colors (opaque and transparent). I want to be able to count them as well as characterize them (size).

I have been looking into a couple of options:

  1. Visual observation using computer vision
  2. NIR spectroscopy:

enter image description here

What other options do I have and which one is most effective (preferably also least expensive)?

Update: In our production plant we produce at 500 kg/hr and the pellets are transported along a conveyor as a monolayer of around 50 cm width at around 5 cm/s. However i can imagine that is simply too complex to begin with and i am not opposed to having a secondary stream with much lower massflow and then upscale in a later stage or just have labscale setup to check samples taken of the productstream. I am ok with an approximate count as long as it is reproducible. I would also be ok with initially only detecting 100+ micron black spots.

nluigi
  • 167
  • 2
  • 13
  • 1
    What other techniques have you already considered so we don’t waste our time looking at repeats? – Solar Mike Dec 09 '17 at 18:57
  • @SolarMike mostly option 1, using blob detection in opencv does the job but I am unsure if I can get the right hardware to pull it off with such small spots. There are some commercial machines from OCS but they cost upwards of 60k euros. Option 2 I found online for continuous processing of grain where they are able to detect spots of 80+ micron. I am familiar with spectroscopy but I wonder how it is possible to distinguish such small spots in spectra. – nluigi Dec 09 '17 at 23:33
  • 1
    @nluigi while 60k euros sounds a lot, consider what the cost of your work is, what the cost of failure is, and what the cost of equipment you need to buy is. Getting your system to work at the dpeed you need is also quote tricky. Also consider that a vendor also does support for their stuff, which in your case is you. Ultimately, you shouldnt do this decission unless your the business owner or manager. – joojaa Dec 12 '17 at 06:35
  • @joojaa, sure that is true... however i think my question is more general in that i would like to know what different methods there are to detect such small defects. If in conclusion, the only way is to invest 60k to buy a commercial machine then i will of course report that to management and the ball is then in their park. – nluigi Dec 12 '17 at 09:40
  • @nluigi I feel it is very unlikely that anybody would do this kind of work for you in manner that would satisfy you. Even if i would have this sort of list i wouldnt give it out for you. – joojaa Dec 12 '17 at 10:16
  • @joojaa, it is kind of the point of a Q&A site... If this question doesn't receive any answers then so be it. In the mean time i will continue my own research and if i find anything usefull i will update here. – nluigi Dec 12 '17 at 10:44
  • @nluigi yes kindof, but in this caseits more like a question to do your work for you. Since in fact enumerating all the ways to do something is part of the research process. Get a list fo all measureable things and start ticking of the ones that are not suitable for your usecase. – joojaa Dec 12 '17 at 10:45
  • @joojaa, right and i will continue my own research. In the meantime i hope that any of the multidisciplinary engineers who frequent this site are willing to provide some additional information I would not have thought of myself. Two heads are better than one, imagine a thousand. – nluigi Dec 12 '17 at 11:42
  • @joojaa, this becomes something like we do for him because he is honest enough to tell that it is. Everyone do ask question. And there are people kind enough to guide him to the right track even though it is for his gain. What do we gain from this? NONE. just waste of time. But knowing that we had helped someone, some people find it noble. :) Cheers. – Jem Eripol Dec 13 '17 at 00:57
  • 1
    How wide and how fast is the product stream? Knowing this would help determine the type of system you could use. Is knowing the approximate amount of black spots good enough or must you know precisely? Judging by the NIR analysis you wouldn't mind an approximation..? – CraigC Dec 13 '17 at 05:57
  • @CraigC - check my update – nluigi Dec 13 '17 at 12:47
  • So, that is about 139 grammes in a space of 50cm by 5 cm if we take 1 second. Then are they clumped together in the middle of the conveyor or evenly spread out so that it is easy to see them? – Solar Mike Dec 13 '17 at 13:31
  • @SolarMike - easily and uniformly spread out – nluigi Dec 13 '17 at 16:18
  • So how do you turn them over to see the other side? – Solar Mike Dec 13 '17 at 16:21
  • @SolarMike good question, I don't know exactly. The pellets fall into a box after being conveyed, perhaps two sensors could be placed on in front and behind side to analyze. Another option is to add some vibration to have them change orientation during conveying. Honestly, I am just interested in any options so don't be limited by my specific production setup. – nluigi Dec 13 '17 at 18:40

1 Answers1

4

The simplest method will be a multi-step approach. The first step will be a defective product detection, that gathers general data about the product stream and quarantines defective products. On a parallel path, a small batch of samples will be also sequestered from the main line to be analyzed continuously for detectable defects. A laboratory analysis could then confirm size and count using digital tools.

Recognizing that these are, for the most part, polarized plastics, they should have a fairly consistent low to modest relative permittivity, double to triple the value of air. Regardless of opacity and/or material, a product line with high amount of defects will be unpolymerized or coked plastics, consisting primarily of carbon, which has a detectable higher relative permittivity, nearly five times the value of even the most polar plastics. A static (ultra low frequency) dielectric detector can detect these defects, and quarantine the offending batch.

enter image description here

As the relative voltage across the plates changes, the increase in stored voltage can be detected as a defect for sequestering.

On a parallel path, a small portion can run through a dielectric spectrometer. These have had some uses before, (See here for detection of different polymers based upon response spectra, and here where it was used to detect areas of low polymerization in LDPE.) It works the same way as the dielectric detector, but uses an alternating current. This can get a full spectrum analysis for comparison between polymers:

enter image description here

Since this detection method is slower, it will need to be able to analyze samples one at a time. However, this will ensure the main detector is not fluctuating substantially due to contaminants in the air or other problems. It will also review a history of defective compounds. As the library of spectrum is built, this could replace entirely the need to utilize laboratory detection, except for the cases of control chart defects (see below).

Some advantages of this is that it can detect defects inside the beads that would not be visible via normal spectroscopy alone.

Final laboratory counts to measure black spots could then be utilized to verify the automated machines are working within parameters. The 7 tools of quality, particularly the control chart, will be useful as a countermeasure to ensure these detectors are working properly.

enter image description here

Mark
  • 5,333
  • 13
  • 47
  • Mark, thank you. This is exactly why I posted this question, I would never had thought of using this particular property of plastics. – nluigi Dec 14 '17 at 06:26
  • No problem. When we thermoformed the extruded sheets for fabricating liners for composite pressure vessels, we used a spark tester to detect pinholes and cracks. It also picked up contaminants from the mold that embedded themselves during gel. While it never tells the complete story, electrical conductance of insulating plastics is a great indicator of defects and lets you know where to focus your attention. – Mark Dec 14 '17 at 07:38
  • Is this sensitive to trace amounts of metal (for example iron oxide) which may be present due to the use of natural fillers (chalk, talcum etc)? It seems iron oxide has a dielectric constant of 14.2 compared to 1.5-2.5 for PP/PE and 1 for air. Does this mean that any detection of black specks will be overshadowed by the presence of trace amounts of iron oxide? – nluigi Dec 14 '17 at 08:32
  • 1
    If the trace amounts were in the plastic already when it was tested (and found to have a value of 2.5), this shouldn’t affect anything, as you’d be looking for a delta in production line conductivity from the control line. This is why a control chart would be needed. The trace amounts would affect conductivity, but only up to a certain amount. A good clean mold would certainly still leave atoms of rust in our thermoformed parts, that’s just the nature of the device. But only when it was to a level where we could detect it with a magnifying glass or so did our crude setup detect contaminants – Mark Dec 14 '17 at 17:54
  • 1
    @nluigi I unearthed a swiss patent that utilized these methods (http://www.google.ch/patents/US6897661). The patent shows how 10L of water was circulated in the flow stream in this liquid based detector. When some blood was injected into the holding tank at 500 ppm concentration, an initial spike can be detected, and as diffusion allowed the blood to spread throughout the tank, the capacitance remained elevated. In a similar sense, the initial trace iron oxide contaminants will elevate the capacitance, but the detector will need to find the spike of new contaminants. – Mark Dec 14 '17 at 18:19