Ferrite Test Accuracy: Fields vs. Points.

Date 02-23-2011 13:18

We have a supplier who keeps violating our test spec for the ASTM E562 image analysis for Ferrite %.

Our spec calls for a minimum of 15 fields with 36 points per field at 500X. They keep testing at 400X with 30 fields and only 16 points per field.

They are going to try to make the argument that more fields is better. But it is not an apples to apples comparison. There is no easy way to compare the two.

If I take a total point count, it is 540 our way vs. 480 their way. I don't see how they can say they have the same relative accuracy or confidence interval.

Aside from rejecting their test reports what else can I do? There is a lot of math involved and the ASTM spec is not straightforward in explaining differences in total point count.

By js55

Date 02-23-2011 13:55

My opinion would be that from a practical standpoint its a who cares. Its like arguing over how many angels can dance on the head of a pin.
Its a question of what is to be accomplished by measuring the ferrite. The crack/sigma fulcrum.
However, from a customer specification standpoint I'd tell the supplier to get his shyt together or he won't be a supplier long becasue I don't give a rats scrotum if he thinks his non conformance is better or not.

By OBEWAN

Date 02-23-2011 14:06

I rejected the report since it was not to spec. That means they have to write a supplier deviation and get it approved for the data to be accepted. I just don't want to be unreasonable. From where I sit though I am most often forced to take our specs as black and white - quite literally. But if there is a valid reason for their method being better then maybe the spec needs to be changed. This is a famous European test lab that relies on automatic image analysis systems. So far in my digging I have learned that there is ample opportunity for errors with these systems too. They can misread white and gray etc.... scary stuff for less than a Phd like me.

By js55

Date 02-23-2011 16:45

I would be surprised if after averaging it all there was any difference outside of normal scatter bands.

By 803056

Date 02-25-2011 19:37

I admit my limited knowledge and experience with ferrite determinations, but isn't this the reason they have moved to Ferrite Numbers?

Best regards - Al

By OBEWAN

Date 02-28-2011 09:32

Isn't a ferrite number a spec limit applied to a filler wire?

I am concerned about ferrite % which is a range of 35-55 or 30-60%. The fact that it is a range lets a lot of tests off the hook if they fall in the middle I suppose. If they are at the edge though accuracy is important.

On our filler wires for stainless we ask for 2-10FN. That is quite a wide range. But if I understand things correctly, when the %ferrite range tests fail, they are often corrected by adjusting the FN of the filler wire.

By js55

Date 02-28-2011 21:55

Al is right. FN is the chosen technology now. But even with FN there is considerable variation in results. The IIW ran a pile of round robin tests on labs for FN and found a very broad range of results.
A broad range of results in testing is OK(2-14, 2-16 could be fine depending upon the alloy, the wall thickness, the welding process). All you are tyring to verify is if the FN is high enough to minimize microfissuring and low enough to minimize sigma.
Some alloys are more sigma prone than others, for example 316 as opposed to 308 due to Mo.
And keep in mind that fully austenitic SS welds are made every day on alloys such as 310.

FN can be applied to filler metal and welds. Often you will see filler metal evaluation by diagram and weld evaluation by FN testing.
The problem with diagrams is that they don't account for cooling rate. The problem with FN testing is the inherent error of magnetic permeability testing.

By MMyers

Date 03-01-2011 17:43

"The problem with FN testing is the inherent error of magnetic permeability testing. "

Can you point me in the direction of some literature on this? I know it's used alot, but I've never gotten to play with it to learn the deficiencies.

By js55

Date 03-01-2011 21:11

The most info on this I know of are the tech articles through IIW which are based upon round robin reports of FN testing at multiple labs.
There may be some info on the web.