Hardness testing methods and procedures - what is right and what is ludicrous

[Dan Watson]

Hello. My organization has been going around in circles trying to get a hardness test procedure in place for years. Some of the past discussion points have been:

- just test one piece pass/fail. No sampling plan needed. If asked for result, send end customer sub-contractor certificate
- do not record values since "we know" the variation in a batch of parts will give values above or below specification range
- must have one seating hit and three hits after to average test result
- Automated tester is more accurate than a semi-manual tester. (QC uses semi manual tester due to proximity to receiving dock. Automated unit on other side of facility, used constantly in-house heat treatment)
- must use end of the tenon of part rather than head of the part (see attached pic)
- must remove 0.040" of part for surface to test
- must use acid to remove plating before testing
- can test case hardened parts with Rockwell tester
- Do not need SOP (attached) for operators, just training

Has anybody had the same type of naysayers for proper incoming inspection testing and recording values? Iam at my ropes end and just give in to their "known" methods. I really could use some guidance.

 

[Ron Rompen]

Dan: I don't know if you require an SOP for hardness testing, although some of the statements you made lead me to question that;

'Just have one piece pass/fail' - I would strongly urge you to use a higher number of samples - my past practice has been to take 3 random samples. Providing your supplier certificate as evidence of hardness is fine.

'One seating hit, and 3 hits afterwards' - No 'seating hit' should be required. Record the 3 individual readings - that is the REAL data.

"we know" variation in a batch......below specification range - So you are aware that (and ok with) providing nonconforming parts to your customer, apparantly on an ongoing basis? How can you justify this?

'Automated tester is more accurate...' - What is this statement based on? Unless you are doing microhardness, the results from a manual and automated (or semi-automated) machine should be similar UNLESS one of the two is not correctly calibrated/operated

'must use end of tenon.....' - Why? If the requirement is for overall hardness, then you are required to MEET that hardness anywhere on the part. If you feel that you should measure only in one area, this should be agreed to by the customer.

'must remove 0.040"...' - Again, why?

'Must remove plating with acid....' - Depends on the plating, however most electrocoating does not have an impact on the surface hardness.

'Can test case-hardened...' - Case hardening only affects the surface layer of the material - the 'soft' material underneath allows deformation of the surface without the appropriate 'spring back' and can give incorrect readings on a 'regular' tester.

'Do not need SOP.....' - Although this may be a correct statement, with the questions noted above, I would suggest that you DO need an SOP, just to ensure that operators know how to perform the testing correctly.

 

[Dan Watson]

Dan: I don't know if you require an SOP for hardness testing, although some of the statements you made lead me to question that;

'Just have one piece pass/fail' - I would strongly urge you to use a higher number of samples - my past practice has been to take 3 random samples. Providing your supplier certificate as evidence of hardness is fine.

'One seating hit, and 3 hits afterwards' - No 'seating hit' should be required. Record the 3 individual readings - that is the REAL data.

"we know" variation in a batch......below specification range - So you are aware that (and ok with) providing nonconforming parts to your customer, apparantly on an ongoing basis? How can you justify this?

'Automated tester is more accurate...' - What is this statement based on? Unless you are doing microhardness, the results from a manual and automated (or semi-automated) machine should be similar UNLESS one of the two is not correctly calibrated/operated

'must use end of tenon.....' - Why? If the requirement is for overall hardness, then you are required to MEET that hardness anywhere on the part. If you feel that you should measure only in one area, this should be agreed to by the customer.

'must remove 0.040"...' - Again, why?

'Must remove plating with acid....' - Depends on the plating, however most electrocoating does not have an impact on the surface hardness.

'Can test case-hardened...' - Case hardening only affects the surface layer of the material - the 'soft' material underneath allows deformation of the surface without the appropriate 'spring back' and can give incorrect readings on a 'regular' tester.

'Do not need SOP.....' - Although this may be a correct statement, with the questions noted above, I would suggest that you DO need an SOP, just to ensure that operators know how to perform the testing correctly.

Thanks Ron. I needed a sanity check. Science, physics, and good methodology are bad words at this company. They can make up their own facts and call it evidence.

 

[Miner]

The seating hit only needs to be done if you have removed/replaced the penetrator. I agree with Ron's comments.

 

[mattador78]

Fyi if you are stripping off plating with a HCL be aware of de-embrittlment. As Ron says a few microns of zinc for example wont cause too much of an issue

 

[John Predmore]

naysayers for proper incoming inspection testing and recording values?

I think everyone has an opinion. Even self-proclaimed "experts", with self-professed experience, have no guarantee their experience fits your circumstances. Listen to the parts instead.

I think a data-driven study is your best approach. Measure one part 10 times. I am estimating it takes less than a minute to do one measurement. Plot the readings in sequential order. Maybe 20 or 50 re-measures is necessary, until you see variation settle to a fixed range, and you satisfy yourself you understand the nature and baseline of measurement variation. If you can't literally hardness test the same spot, then move the part a few millimeters each time (follow the equipment manufacturer's instructions). You don't know what variation you might see unless (until) you measure more times than you think you ought to. Then re-fixture the same part 2 or 3 or 10 times, and take another couple readings in the new position(s), to learn how isotropic is the part (what you show is a symmetrical part, but I don't know how it is made, so I am not willing to assume without data that hardness is isotropic. Material properties don't care what the drawing shows; its the way individual parts are made that ultimately matters in variation studies).

Look at your graph{s} with a critical eye: using inductive reasoning, how many readings (and in what position) do you need to get the best picture of what you consider a most representative value of hardness?

Then do the same procedure with random parts from different batches. If there are other factors (you listed a few) which can vary, introduce those into your study one-at-a-time so you can assess how much (or whether) the other factors contribute significantly to variation. As you learn more about the physics and the material properties, you will gain understanding which variables you must control in your procedure and which don't matter much. Once you understand what to expect from the measurement variation distribution, you no longer need dozens of readings on one part to reach what you consider the most representative value.

Write up your findings in a measurement study. State your conclusion and your recommended procedure. If a naysayer criticizes your study, invite them to conduct their own study of similar scale, and present you with a report and a full set of data. Then you and he can have a scientific discussion about what the evidence tells you. Chances are, the naysayers don't want to do the work to give evidence-based input.