Figuring out Grain Size

[BluntCut MetalWorks]

I am happy with my current ht formula for low Cr steels. To determine grain size without a SEM is challenging. Late last night, I came up with a few image processing steps to trace strong grain boundaries - in the spirit of FFT, I amplified what look like good signal (strong grain boundaries), removed/dropped out noises.

* gotta play around with what I've first before spend a good chunk of $ for SEM & BSED imaging... *

Here is - what I think - my SQ 52100 grain size look like. I tried to be conservative, so quite a bit of GB probably got removed.

Is my processing look reasonable or totally off? Thanks for feedbacks/advices.

Edit: Moderator - please move this thread to Shop-Talk. Thanks

 

[BluntCut MetalWorks]

Most of the time, Retained Austenite is not easily seen, especially with an optical microscope. Well (lucky me), here is an image with 2 very large RA. I found about 6 of these huge RA in a sample size about half inch square.

btw, this cfv sample (detached from an attach coupon on a test blade) hardened with ultra low(barely above Curie Temp) austenite temperature.

There is a big IF on the pro/con on converting RA to martensite in this type of severe dislocation. *as it* is actually a state of equilibrium. If this defect end up on the apex, RA would roll and or fracture on subsequent damaging impacts. If RA got converted to Mart + tempered beforehand, fracture at boundary will taken place when encounter damaging impact anyway. Note: RA convert to Mart will increase dimensional wise about 4%, which put quite a bit of stress in surrounding area.