It may sound a little far-fetched, but experts say that the causes for more than 90% of all plant failures can be detected with a careful physical examination using low power magnification and some basic physical testing. Inspection of the failure will show the forces involved, whether the load applied cyclically or was single overload, the direction of the critical load, and the influence of outside forces such as residual stresses or corrosion. Then, accurately knowing the physical roots of the failure, you can pursue both the human errors and the latent causes of these physical roots.

Before explaining how to diagnose a failure, we should review the effects of stress on a part. When a load is put on a part, it distorts. In a sound design the load isn’t excessive, the stress doesn’t exceed the "yield point", and the part deforms elastically, i.e., when the load is released the part returns to its original shape. This is shown in Figure 1, a "stress-strain" diagram that shows the relationship between loads and deformation.

In a good design, the part operates in the elastic range, the area between the origin and the yield strength, the part will be permanently deformed. Even greater increases in load will cause the part to actually break.

Figure 1 illustrates a very basic point of design, and applies when the load on a part is relatively constant, such as the load on the frame of a building or the stress in the legs of your desk. It is a very different case when fluctuating loads are applied, such as those in a hydraulic cylinder or in an automotive connecting rod. These fluctuating loads are called fatigue loads, and when the fatigue strength is exceeded, a crack can develop. This fatigue crack can slowly work its way across a part until a fracture occurs. (Corrosion can greatly affect the fatigue strength).

Figure 1

Machine components can fracture from either a single overload force or from fatigue forces. Looking at the failure face will tell which of these was involved. A single overload can result in either a ductile fracture or a brittle fracture.

DUCTILE OVERLOAD VS. BRITTLE OVERLOAD FAILURES

A "ductile failure" is one where there is a great deal of distortion of the failed part. Commonly, a ductile part fails when it distorts and can no longer carry the needed load, like an overloaded steel coat hanger. However, some ductile parts break into two pieces and can be identified because there is a great deal of distortion around the fracture face, similar to what would happen if you tried to put too much load on a low carbon steel bolt.

The term "brittle fracture" is used when a part is overloaded and breaks with no visible distortion. This can happen because the material is very brittle, such as gray cast iron or hardened steel, or when a load is applied extremely rapidly to a normally ductile part. A severe shock load on the most ductile piece can cause it to fracture like glass.

An important point about failures is that the way the load is applied, i.e., the direction and the type, can be diagnosed by looking at the failure face. A crack will always grow perpendicular to the plane of maximum stress. Below we show examples of the difference in appearance between ductile overload and brittle overload failures.

Figure 2

From the examples above in Figure 2, we know we can look at an overload failure and knowing the type of material, tell the direction of the forces that caused the failure. Common industrial materials that are ductile include most aluminum and copper alloys, steels and stainless steels that are not hardened, most non-ferrous metals, and many plastics. Brittle materials include cast irons, hardened steel parts, high strength alloyed non-ferrous metals, ceramics, and glass.

One note of caution is that the type of fracture, ductile or brittle, should be compared with the nature of the material. There are some instances where brittle fractures appear in normally ductile materials. This indicates that either the load was applied very rapidly or some change has occurred in the material, such as low temperature embrittlement, and the material is no longer ductile. An example of this was a low carbon steel clip used to hold a conduit in position in a refrigerated (-50 F) warehouse. The clip was made from a very ductile material, yet it failed in a brittle manner. The investigation showed it had been hit by a hammer, a blow that would have deformed it at normal temperatures.

In a brittle overload failure, separation of the two halves isn’t quite instantaneous, but proceeds at a tremendous rate, nearly at the speed of sound in the material. The crack begins at the point of maximum stress, then grows across by cleavage of the individual material grains. One of the results of this is that the direction of the fracture path is frequently indicated by chevron marks that point toward the origin of the failure. Example 1 is a photograph of the input shaft of a reducer where the chevron marks clearly point toward the failure origin, while Figure 3 is a sketch of the cross section of the wall of a ruptured 20ft. (6.1 m.) diameter vessel. In both cases, by tracing the chevron marks back to their origin, we knew exactly where to take samples to determine if there was a metallurgical problem.

Notice how the chevron marks (high-lighted) point toward the origin of the fracture.

Example 1

Figure 3

So far we’ve talked about the gross overloads that can result in immediate, almost instantaneous, catastrophic failures. A very important distinction is that fatigue cracks take time to grow across a part. In a fatigue failure, an incident of a problem can exceed the material’s fatigue strength and initiate a crack that will not result in a catastrophic failure for millions of cycles. We have seen fatigue failures in 1200 rpm motor shafts that took less than 12 hours from installation to final fracture, about 830,000 cycles. On the other hand, we have also monitored crack growth in slowly rotating process equipment shafts that has taken many months and more than 10,000,000 cycles to fail.

Figure 4 shows a simple fatigue crack with the different growth zones and the major physical features.

The fatigue zone is typically much smoother than the instantaneous zone, which is usually brittle and crystalline in appearance. Progression marks are an indication that the growth rate changed as the crack grew across the shaft and don’t appear on many failure faces.

Figure 4

There are some complex mechanisms involved in the initiation of a fatigue crack and once the crack starts, it is almost impossible to stop because of the stress concentration at the tip.

A stress concentration is a physical or metallurgical condition that increases the local stress in the part by some factor. A good example is the shaft shown in Figure 5. We see that the stress in the area of the radius varies depending on the size of the radius. A small radius can increase the stress dramatically.

Figure 5

Stress concentrations, indicated by the symbol Kt, can be caused by changes in metallurgy, internal defects, or changes in shape. There is extensive data that indicates that the resultant values depends on both the type of stress, i.e., bending, torsion, etc., and the general shape of the part.

Stress concentrations have a great effect on crack initiation because of their effect on increasing the local stress. The crack can start solely as the effect of the operating loads or it can be multiplied by the stress concentration factor.

The face of a fatigue failure tells us both the type (bending, tension, torsion or a combination) and the magnitude of the load. To understand the type of load, look at the direction of crack propagation. It is always going to be perpendicular to the plane of maximum stress. The four examples in Figure 6 reflects four common fracture paths.

Figure 6

Figure 6 brings up the question "what type of bending?" Was it one-way plane bending, like a leaf spring or a diving board, or was it rotating bending, such as a motor shaft with a heavy belt load? As seen in Figure 7, looking at the fracture face again tells us the type of load. Notice that "rotating load" on the right causes the crack to grow in a non-uniform manner. In general, when the divider of the instantaneous zone does not point to the origin, it shows there was a rotating bending involved in the failure cause.

Figure 7

Fatigue failures almost always start on the outside of a shaft at a stress concentration, because the local stress is increased. However, the instantaneous zone (IZ) carries the load in the instant before the part breaks. By looking at the size of the IZ, you can tell the magnitude of the load on the part. Figure 8 shows a comparison between a lightly and a heavily loaded shaft for both plain bending and rotational bending.

Figure 8

If a part is relatively lightly stressed, the cracking will start at only one point and the result will look like one of the examples above. However, if a shaft is more heavily loaded, then cracks can start in several places and work their way across the part. In Figure 9 we see a sketch of a rotating shaft that failed in only a few weeks. Inspecting it, you can see the instantaneous zone is very small, indicating it wasn’t highly stressed. Also, the crack is straight across the shaft, showing the cause was a bending load. But if the load was light, why did the shaft fail? The answer is stress concentrations.

Figure 9

Looking at the fracture face, you see a series of ratchet marks. These are the boundaries between adjacent fracture planes, i.e., between each pair of ratchet marks is a fracture origin, and as these individual cracks grow inward they eventually join together on a single plane. The small instantaneous zone indicates the stress at the time when the shaft finally broke was low, but the multiple origins and the ratchet marks show us there was enough stress to cause cracking at many points around the perimeter almost simultaneously.

From this you can conclude that there must have been a significant stress concentration. (The calculated stress concentration was in the range of 4.0, so the stress in the area of those origins was four times as much as it should have been.)

With this information on the type of load and the magnitude of the load, we can start looking at some failures and diagnosing where they came from. Following are some examples of failures and an explanation of their causes.

Figure 10. A torsional fatigue failure resulting from a loose hub fit. Note the severe fretting (from looseness) and the cracked shaft.	Figure 11. A rotating bending fatigue failure from a motor shaft. Notice the small instantaneous zone that shows the shaft was lightly loaded at the time of failure.
Figure 12. By tracing the progression marks backward, we can see the failure started at the corner of the keyway. But, the instantaneous zone is tiny. This indicates the shaft was very lightly loaded at the time of failure and further research is needed.	Figure13. Impressive brittle fracture of a large universal joint. The chevron marks point to where the failure started. The fact that the surface has uniform roughness tells us that this was an instantaneous failure.
Figure 14. A testimony to an inept repair. The weld repair of the shaft should never have been attempted. The four gross weld flaws initiated fatigue cracking of a very heavily loaded shaft.	Figure 15. Typical rotating bending failure. Moderate sized instantaneous zone. Rotating bending failure origins surround the shaft.
Figure 16. Ugly plain bending failure.

Neville Sachs, P.E., is President of Sachs, Salvaterra & Associates, Inc., which was founded in 1986. The consulting firm specializes in improved plant and equipment reliability and technical support services. Among the firm’s capabilities are vibration monitoring, mechanical failure anlaysis, corrosion and materials engineering, design reliability analysis and a wide variety of nondestructive examination methods. Previously, Neville was Supervisor, Reliability Engineering for Allied Signal Corporation where he was instrumental in developing one of the first large predictive maintenance inspection programs in the nation. Mr. Sachs received a Bachelor of Engineering Degrees in both Mechanical and Chemical Engineering from Stevens Institute of Technology. Visit his web site at http://www.sachssalvaterra.com.