Inline inspection in Roll Form mill 3

[Markerinhand]

Hey guys,
I work as an engineer where we use roll form lines to produce a beam. I was trying to investigate different methods for automated inspection at the unload table. An operator currently visually inspects the beam, and packs it.
I would like to start by looking for cracks in the beam between holes, these form as it is stressed by a bending procedure.
I have found so far Eddy Current, Vision, and Laser inspection methods.
I have very little experience with implementing these, and was wondering a few things. Besides what I have listed, is there anything worth me looking into. Second, what would be the most reliable, repeatable, and easiest to maintain / adjust out of these.

 

[BillPSU]

The primary means of reducing cracking caused by forming is to improve formability of the material. You haven't stated the material but if it is carbon steel going down to 1011 steel away from 1018 will improve the formability. In higher strength steels you would go to a high form type material or its equivalent.

Inspecting cracks out is not the solution, elimination of the cracking is the best solution. There are other solutions such as reducing the bending by redesign of the rolled cross section or adding additional roll stands reducing the amount of bending between each stand.

Bill

 

[Markerinhand]

Hi Bill,
Thanks for the answer. It is a harder steel (martinsite structure. Working on methods of eliminating the cracking within the means, cant change the steel since its specified, it doesn't crack in the rolls, it cracks in the sweep. Its thought that the mill slows down during the sweeping process, causing it at the weld to slow down. So we are working on this.
In the mean time I need to provide a method to prevent a cracked one of getting to the operators inspection / unload. This would be where the system I was asking about came in.

 

[CoryPad]

Ultrasonic testing, magnetic particle inspection, and liquid penetrant inspection are options, though usually not fast.

 

[Markerinhand]

I was looking at those and I do not think they can keep up unfortunately. Any experience with the eddy current?

 

[CoryPad]

Yes, that can be used successfully for crack detection.

 

[MintJulep]

http://www.automationwr.de/en/detect-hairline-cracks-autoflux

 

[MikeHalloran]

The part is fracturing because the sweep operation on top of the rolling operation is exceeding the material's ultimate strain. If you are getting any good parts at all, that would suggest that you are on the ragged edge of formability, and different lots of raw material will give different results.

If so, you may be able to sort the incoming coils by hardness, keep the soft ones, and throw away the hard ones instead of bothering to process them. It's expensive, but cheaper than making scrap and detecting it after forming.

Or, you may be able to reduce the strain in the finished article enough to prevent cracking with some adjustments to the tooling. To explore that, you'd want to measure the _exact_ geometry produced, and calculate the resulting strains, then compare the produced geometry with what's on the part print to see where you have room to adjust the tooling, e.g. with bigger radii in selected locations.

You might want to examine the mill certs for the incoming material, and have the material itself tested; forgery is not unknown in the steel business.


Mike Halloran
Pembroke Pines, FL, USA

 

[Ripz]

Fixing root cause and performing Quality Control checks are two separate items and both are necessary. Can't understand why everyone is harping on the first point.

Markerinhand, what is your output rate? There are manual methods of NDE that can accomplish multiple large parts within 10 Minutes eg Water Washable Fluorescent Dye Penetrant Testing. I'm assuming your cracks are open to surface?

Remember you do not need to perform NDE on all your beams. Having your operators perform 100% Visual testing while performing Dye Penetrant Testing on a sample percentage (maybe 10%) of beams, while also trying to fix your fabrication process to minimize of cracking, should be a satisfactory attempt at ensuring a quality product.

 

[Markerinhand]

Thank you Ripz. That was what I was trying to get at. I am working on the Root Cause, but I NEED to figure out a quality control.
Out is roughly 24 feet a minute, about 4 parts a minute, 2000ish a shift (7.4 hours).
The problem I am having is not in mass quantity (where die penetration in a sample would help), but in the flyers of 2 or 3 parts getting to my next operation.
They do 100% visual, required to touch key aspects of the beam, but its found to maybe be 80% effective depending on who is doing it.
Started speaking with some on Eddy Current inspection. Also trying to push to trial a vision system (failed in the past apparently).
Any other opinions would be helpful towards the detection side.

 

[Ripz]

From the sound of your output, manual NDE would't be applicable. And if you're only looking at teh bolt holes you'll likely need an Eddy Current Method, specifically made for your application. Your best bet would be getting a specialist NDE company to tailor make a probe for you. Give these guys a call and work from there.

You may need some operator training though.

Link

 

[Markerinhand]

Thanks Ripz. I have two different ones coming in to look at the application.

 

[Ripz]

Keep us informed with their proposals. Id be intrrested to know whst they recommend.

 

[Markerinhand]

I will. I am thinking I will get it going early in the new year so we'll see.