Study the differences between parent metal and heat affected zone of weldments.

In my opinion, it is not only the grain size or shape, but you should check for chemistry, too (e.g. EDS). If bondings are weaker in one material at the grain boundaries, e.g. through precipitations or a high amount of high-angle GB, this will crack earlier.

Grain size, crystallographic texture, misorientation of near neighboring grains (microtexture), loading conditions (stress or strain controlled), elastic anysotropy  they all impact and contribute to differentiate fatigue crack initiation, though they are not all always important. Depends on the material.

Dear Dr. Mondi,

Interesting question!

As correctly stated by the researchers above, purity, texture and many other factors can play a role here. However, as per your question, I am assuming that all factors other than grain size distribution somehow remain constant.

In general, heterogeneous microstructure will contain points where hardness gradients, and hence strain gradients exist. These gradients could become crack initiation sites. Homogeneous microstructure will not have this sort of localisation (from grain size point of view alone). This is a simplistic understanding, so let us look at literature from different materials:

Ref [1] (on a Ni-based superalloy) shows that a heterogeneous microstructure starts creeping (secondary & tertiary stages) faster, but also elongates more. This could be because a heterogeneous microstructure will contain some fine grains, hence more grain boundariy density at these locations and more prevalence of local boundary sliding/migration. The greater ductility could be because fine grains are conventionally associated with more ductility due to shorter dislocation pileups etc.

On the other hand, the life under creep-fatigue is significantly shorter for heterogeneous structure. This is attributed to higher oxygen pickup at boundaries wherever fine grains are present. This suggests that with any oxygen-related damage mechanism, a homogeneous microstructure would be preferred.

Coming to hot deformation, the differences in grain size can also result in localised recrystallisation at some pockets, which then become softer than the surrounding matrix. These then result in localised flow [2], which could go on to manifest as shear bands and ultimately cracks [3]. This behaviour has been reported in steels [2] and Ti-alloys[3], so it is quite independent of crystallography. Of course, in such conditions you should remember that the grains will not all remain equiaxed; coarser grains tend to flatten faster. This can again result in preferred recrystallization, causing a sort of feedback effect.

Nevertheless, there is active research going on in using these very strain gradients to strengthen materials and retain ductility [4]. The philosophy is basically to have an orderly, alternating arrangement of fine and coarse grains, preferably of very different sizes. The gradients resulting from this arrangement cause strengthening. As I understand, this is something like reinforced composites...so the arrangement and directionality of the fibre (here the coarse grains) is important.

Refs:

[1] http://www.tms.org/superalloys/10.7449/2012/Superalloys_2012_15_24.pdf

[2](skip to discussions:)

https://www.researchgate.net/publication/318107910_Flow_Softening_Index_for_Assessment_of_Dynamic_Recrystallization_in_an_Austenitic_Stainless_Steel

[3]https://www.researchgate.net/publication/236945602_Microstructural_effects_on_adiabatic_shear_band_formation

[4] http://www.tandfonline.com/doi/full/10.1080/21663831.2017.1343208

Article Flow Softening Index for Assessment of Dynamic Recrystalliza...

Article Microstructural effects on adiabatic shear band formation

The history of Steel A is important. What is the reason of non-uniform grains for example? Is there any obstacle phases which limits or routes the grain growth?