In the last entry, we saw several photos of parts that were broken due to loads applied to a long skinny part in a way that caused the part to bend, initially creating a crack perpendicular to the length of the part. The tensile stresses responsible for the cracks eventually give way, in some cases, to shear stresses. We saw two situations: one where the shear stresses caused the crack to make a “hard turn” and run perpendicular to the original direction; secondly, the shape or configuration of the component had been changed so much that the remaining crack ligament was now primarily feeling tension or compression rather than bending, and the shear stresses created a final crack at 45 degrees to the original crack.

So in both situations, the crack is initially macroscale brittle and changes to a macroscale ductile crack before final separation. Consequently, in both cases part of the crack is brittle and part ductile. Once again we are reminded that if the material is more or less homogeneous, the material itself must be capable of behaving in a ductile manner.

Most cracks that form in bending do initiate in a macroscale brittle manner. Many or most of these cracks change crack-propagation mechanism to an essentially ductile shear crack either in tension, compression or bending. Even if the load that caused the crack stops before the component separates completely and there is no visible evidence of the change to ductile crack propagation, we see now that it would be wrong to jump to the conclusion that the MATERIAL was brittle and that the heat treater made a mistake.

If we think back on why we started this “macro-fractography adventure,” which was to try to find a way to describe cracks that would allow heat treaters to be able to defend themselves from accusations of doing a bad job of heat treating, we can now see that there is a potential defense EVEN when the entire visible crack LOOKS BRITTLE.

Why have we spent all this time on bending? Well, many real cracks happen due to bending loads or bending loads combined with tension or compression. To make things even more complex, torsion or twisting can be added. But we’ll worry about torsion later. For now, the important thing to realize is that there are really not that many loads in metallic components that are in pure tension or pure compression.

If we have the simplest shape we can imagine (a cylinder with a uniform cross section) and we could glue it between two plates, we could initially load it uniformly. But very soon, as the part lengthens, the cross section starts to shrink. And now, if the ends are glued in place, they can’t shrink. Right away we set up a complex stress state that is mostly tension, but significant bending at the ends, where the shrinking part “fights” the part that can’t shrink. This essentially creates a change in diameter, and a bending stress is created at the diameter change. Where is the crack most likely to happen? At the area of the complex stress state. A tie rod in an extrusion or plastic molding machine is close to pure tension. At some point, however, there are screw threads or a reduced section so the part can be held in place. Where there is a diameter change, there are accompanying bending stresses, at least at the local level.

Another application, which may be pretty darn close to pure tension, is reinforcing bar in prestressed concrete. Yet even here the rebar has to have textured features, which destroy the perfect symmetry of the component and create local bending stresses. In this case, if the rebar starts to crack in nominal tension – probably at a “gripping feature” from localized bending – the concrete is likely to hold the bar in position, thus maintaining the nominal tension and minimizing bending.

I had to think really hard to come up with these examples of the closest things to pure axial loading. Bending may not be everywhere, but it is in a lot of situations. Just knowing that fact and that bending cracks most often start in a brittle manner, even when the material is ductile, should help a heat treater to protect themselves.

A final hint to see if an apparently totally brittle crack created by bending might be at least in part ductile works most easily in rectangular or square sections. Simply lay the part on a table with a cracked end protruding over the edge. Compare the width of the initiation edge to the width of the final edge to separate. Since bending loading compresses the final portion to separate as it stretched the initial portion, the initiation edge might well be narrower than the final fracture edge. If this is the situation, we again have some evidence of measurable shape change, impossible without some amount of ductility.

In a later installment, we’ll take a break from all this loading geometry stuff and look at some classical fracture surface features, so you will soon know how to tell which edge was the initiation edge of the crack and which the final edge.

If anyone reading this has some photos of parts that were determined to have cracked in a brittle manner that you are willing to share with the readers ofIndustrial Heating, please contact to pass them on. If appropriate, they will be shared with commentary.