The failure looks like torsion failure as indicated by the angular surface
I agree with the torsional failure idea. From looking at textooks, if the shaft is under torsional stress then the plane of maximum
shear tensile stress is at 45 degrees (can’t really tell from pictures - does it look like 45 degrees op?).
Slip leading to ductile failure occurs in the plane of maximum shear stress. I think most shafts are designed not to be brittle... Brittle failure occurs in the plane of maximum tensile stress. My reference [attached] shows this type of failure pattern would correspond to a brittle failure under torsion. Does it make sense that the shaft would be brittle? (what is material and operating temperature)?
Since the shaft diameter decreases significantly at the step, it would be a logical place to have torsional failure at the smaller portion of the shaft. I imagine that the torsional stress in the shaft is highest at the top of the impeller and lowest at the bottom of the impeller (using vertical pump terminology). So we might expect the failure to occur near top of the narrow portion / top of the impeller. Does the other side of that slanted crack (hidden from view in the photo) pass near the top of the impeller? Is the shaft step radiused and the impeller face radiused slighly larger?
Another question that comes to mind, did keyway fail first or shaft fail first:
[ol 1]
[li]Let’s say the shaft failed first, then the impeller would drop and grab onto a stationary part, the slanted shaft might continue to drive (by virtue of the slant) resulting in increased stress on the key / keyway. I can imagine that.[/li]
[li]Let’s flip the scenario and assume the key / keyway failed first. Then impeller might spin on the shaft. It wouldn’t go anywhere (to create increased stress on shaft) unless the impeller locknut also came out. And impeller seat on shaft doesn't seem to show any sign of spinning anyway. Seems a less likely scenario.[/li]
[/ol]
Was the impeller locknut still threaded onto the shaft in the as-found condition?
Got any pictures of the key or of the two failure surfaces of the shaft? How about the impeller and adjacent stationary surfaces (for indication what contact may have occurred… or did something get stuck between the impeller and stationary parts etc).
Edit - changed first paragraph. Added attachment from ASM Handbook Vol 11:
"Fig. 7 Free-body diagrams showing orientation of normal stresses and shear stresses in a shaft and the single-overload fracture behavior of ductile and brittle materials. (a) Under simple tension. (b) Under torsion. (c) Under compression loading"
=====================================
(2B)+(2B)' ?