brittle material design

[Tmoose]

Got some danged castings made of hi chrome cast iron that keep provoking questions. One is justification of acceptance criteria for linear indications/heat checks/cracks.
I've made some decent models and used two different general purpose FEA programs to solve for max principal (tensile) stress as appropriate to determine "hot" locations and perilous orientations for what is essentially static loading.
I'll be making a few trial FEA runs with synthetic cracks, but expect realistically sharp crack geometry will just trigger a bunch of singularity warnings, as they should.

Fracture mechanics has been suggested as the necessary tool to determine acceptable indication size limits when oriented "badly."
So, now I'm starting at the very beginning independently learning about fracture mechanics.
One dilemma I have is that the "length" of actual indications is all we are going to have to work with. I think I know about how deep they are likely to be, so could make some reasonable ASSumptions there. Useful/necessary Material properties are pretty much nonexistent.

Does anyone have some hints to help ease this process ?

 

[3DDave]

I think that cracks of any kind will be bad unless there is some proof they are into high compression areas, which is how tempered glass survives (until it doesn't.)

On some high-strength castings it was a final work-around to settle on a casting process where the cracks would be segregated in the raw casting and then machine that volume out of the casting to eliminate them. The collaboration was to make sure the cracks ended up in a low stress area where machining did little harm. The worst was when the supplier took it upon themselves to 'improve' their process, which just moved the cracks back out of the machined target area. Wasted X-rays to find they had moved and then wasted stress effort and engineering effort to move the machining for just a few castings.

These castings were for high, potentially reversing loads, so your needs may differ.

 

[EdStainless]

Other than don't load brittle materials in tension?
We used to cast a screw like flow inducer that went on the intake of pumps out of 25% Cr white iron.
In the end we set up a flow stand to run parts, if they survived we used them, if they didn't we cleaned up and tested the next one.
We tried detailed probabilistic fracture mechanics approach, and gave up.
This stuff is amazingly strong, and some serious indications survived very well.
But at the same time some parts that looked good would self destruct.
Sorry that I don't have anything constructive.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, Plymouth Tube

 

[rb1957]

My suggestion is "don't" (permit cracks ... if cracked then you must scrap).

But possibly that's unrealistic, so
1) consequences of failure in-service ?
2) figure out your critical crack length, for either limit (extreme in-service) or ultimate load (factored limit or extremely improbable in-service load).
3) do you want to detect cracking in-service or wait till it breaks (ie inspect for broken part) ?
4) if you want to detect cracked parts, what crack length (which method) can you detect ?
5) estimate the life between these limits (or component fatigue test and observe the beach marks).

another day in paradise, or is paradise one day closer ?