Home Forums General Discussion Media, Shot, Beads, etc. AMS 2431/4C - Shape

Sections 3.7.2 3 & 4 all refer to Table 2 - Shape Requirements. I would like some clarification on what is required. AWS17 has a nominal size of 0.43mm +/-0.025 (0.405/0.455mm) the table indicates an area of field of 1.588mm2 is to be used allowing a maximum of 2 of unacceptable particles.

Am I reading this correctly as this would mean that if the wire was at the top end of the spec you could only fit 9 pieces into that area?

Also does the 'Number of Fields Viewed' mean that I take 11 samples of this size to conduct the tests of which on 2 can be unacceptable?

Thanks for your help

The area of field for AWS 17 is 0.0625 inch or 40mm. Although it's not real clear this actually means 40x40mm. At the high end of the sizes you could theoretically fit 9x9 particles or 81 pieces in each field.
Since 11 fields are required you would prepare 11 sample cards with cut-out of 40x40mm and count the number of marginal particles and unacceptable particles. You could have as many as 2 unacceptable particles (total) from all 11 fields.

The nominal size of AWS 17 is 0.43mm which means you could get 93x93 pieces into a 40mm x 40mm area

You would normally expect more than 93 x 93 pieces in a 40mm x 40mm square. That is because you assume "square-packing" - which is the exception rather than the rule. "Close-packing" gives more pieces than "square-packing" for a given area.

If using a recess apposed to tape, dose the sampling area have to be square or can is be say circular? Obviously the area contained with in the shape would be the same.

I must disagree with Jack's interpretation of Table 2 of SAE AMS 2431/4C. The area of field is clearly stated to be in units of a square inch - with the equivalent of 1 square inch being given, correctly, as 645 square millimeters. For AWS 17 the required field of view is 0.0625 of a square inch - which corresponds to 40 square millimeters. 40 square millimeters is the area of a square that has sides of 6.32 mm. Using "square packing", spheres with a diameter of 0.43 mm could be arranged 15 by 15 giving a total of 225 spheres in each of the 11 field areas to be viewed.

For consistency, the field area tested should be square - a spherical field area of the same area as a square field area would accommodate a slightly different number of spheres.

Socrates is correct.

645mm sq = 25.4 X 25.4 mm
151mm sq = 12.7 X 12.7 mm
40 mm sq = 6.32 X 6.32 mm

I was president of math club in high school. Fortunately my math teacher is deceased.

Not all of the spheres will be the same size or shape, by using a square and the 'square packing' method surly that biased and unscientific.
By square pack when forming a single layer in a recess, only conforming media would fit to achieve the neat row formation.
Your basing the square packing idea on that all the media is already of the same diameter and shape.
Is not taking a percentage of fails in an area where alignment of particles is not forced fair, by expecting that a predetermined amount will fit neatly in the sample area you are not testing accurately but predetermining that all media in the sample are already conforming to the required size and shape.
I thought that the media was to be poured into the recess and excess removed, how can they be poor into rows again this is biased and not accurate or fair in my opinion.

Referring to table 2 of AMS2431/4c
It states that the area of a quarter of an inch squared is 161mm and that one sixteenth of an inch has an area in millimeters of 40mm. Both are wrong and contradicts table 2B of AMS2430 rev.S and the analysis procedure.

Inch mm Area mm
1 25.4 645.16
0.5 12.7 161.29
0.25 6.35 40.3225
0.0625 1.588 2.521744

I do not have /4C in front of me but here is the information. Hopefully this will help. AMS specifications are in inches, feet ..etc. So any conversion to metric are approxment. Does this corraspond with /4C? If so, then it is the same as 2430 and all the 2431/ specifications for steel media.

1" X 1" = 1 sq in or 25.4mm X 25.4mm = 645 sq mm
1/2" X 1/2" = .25 sq in or 12.7mm X 12.7mm = 151 sq mm
1/4" X 1/4" = .065 sq in. or 6.32mm X 6.32mm = 40 sq mm

Whether this method is bias or not, the question that needs to be asked is; does it control the process either manufacturing or peening to produce a desired result? I believe this test method has been in place for decades and has been proven. But, that does not mean that a proposal to the committee for a better method should not be considered. Please feel free to make that proposal and if not already become a member of the committee.

I hope this helps.

sorry

Inch mm Area mm
1 25.4 645.16
0.5 12.7 161.29
0.25 6.35 40.3225
0.0625 1.588 2.521744

I presented a model that allows an estimate of particle numbers to be made using simple arithmetic. For the model spheres were assumed, not irregular particles. For identical spheres 225 is correct for square packing. Hence we have a ball-park figure of 2475 particles for 11 fields. For actual AWS 17 particles the number will obviously be different - as you have pointed out. It will not, however, be way out. With a maximum of 2 unacceptable particles being allowed that will correspond to 'less than one in a thousand'.

We should find some interns with spare time to do actual count of particles in a sample field.
Several years ago David Francis of Metal Improvement Company made model estimates that seemed to support the number of particles for each media size maintaining a (fairly) constant percentage.
I'll search for that document (in my spare time)

For the three largest sizes the estimated number of particles to be inspected lies between 217 and 312. For all smaller sizes the number varies (with size) from 1852 to 2945. These numbers are based on my own Excel-based model.
The biggest problem with AMS 2431 shape analysis is that it is based on subjective visual inspection - different people will give different answers for the same samples. It would be very useful if a set of reference images could be produced. These only require a digital photograph to be produced for each of the field being viewed. These images could be displayed on any computer screen to be 'visually inspected'. Assessments by different individuals could then be compared. This could lead to a future insistence on computer-based image analysis!

You are right visual inspection is very open to different people opinion of correct shape but this is random uncertainty which is common in all testing and inspection.
The main problem I can see is the range of media as you identified can potential be massive, my boss found similar results to you Socrates.
I believe that the range is far to big to have a numerical figure depicting the total number of unacceptable particles. It should be a percentage that is based upon the screened sample currently being inspected.

We have to be practical! A percentage would require that more than two thousand individual shot particles would have to be counted (except for the rarely-used huge particles). That would be very tedious and time-consuming. With the present specification we only have to identify marginal and unacceptable particles. This could easily be converted to a percentage to the nearest significant figure i.e <0.1% but would not add anything useful.

Why do you need to know how many marginal and unacceptable particles there are in the sample, If you don’t know how many particles make up that sample, the results don’t mean anything?

As you said there is a huge range as to the potential number of particles in the sample, that is true!
Unless you have number as how many particles are in the sample it doesn't mean anything.
All you know is that the sample you had has 2 failed particles. Does that mean the stock is no good because it had 2 fails in an indeterminable number?

I wouldn't suggest counting them it’s not practical but a percentage of failures based on weight could be useful.
We don’t know how many particles are in each container but we know the weight fairly accurately.

We may even know the hopper weight for instance but as the amount in the hopper it’s impossible!!!
If there is an acceptable amount of fails based on weight the experiment has more meaning.

The results do mean something - quantitatively! Consider, for example, that we decide to set a limit of 0.1% of unacceptable particles in a defined batch. 2 unacceptable particles would then be the most that could be tolerated in a batch of 2000 (2 being 0.1% of 2000). We now have to ensure that the sample size examined contains at least 2000 particles. As shown previously there will be at least 2000 particles in the total number of fields specified for each normal particle size.

So you would have to go to some extent to count them?

Because 2 fails is based on 0.1% of 2000 particles you need to first ensure there are at least 2000 particle to screen.

But if it was calculated against weight you could have any number of particles you would only have to do some math to calculate a percentage of fails by weight a sample of 2000 particles of CWS62 has a mass of approximately 43.6g. It would be approximation of-course and not as accurate as counting but a lot easier and the results could be used a lot more effectively.

Percentage to weight is already used to calculate size failures.

We disagree!

you disagree but you used a percentage of a predetermined sample to explain yourself in your last post I,e 0.1% in a sample of 2000

The point i am trying to make is that there is no predetermined amount to be inspected and there is a numerical figure dictating how many fails should not be exceed. you proved my point by using it,

There is no meaning to the test with out parameters.

what if you find two fails in a sample of 2400

is that a fail? no

because the figure is based on 2 fails in 2000 but the spec does not state that!

I have thank you i have very much enjoyed are debate on the issue

Thank you too. Just one final comment before I must withdraw from this topic. I agree that two unacceptable particles in 2400 would not constititute a fail. On the other hand, three in 2400 would constitute a fail. Three in 2400 is more than an assumed 0.1% -the authors of the specification not having declared that assumption.