Although sometimes fatigue isn't the mechanism, and a shaft will fail as a result of another mechanism, such as simple overload, this is usually the exception rather than the rule. Overload failures generally result from an unexpectedly high stress condition associated with failure of another component or some abnormal operating condition.

     The mechanism of fatigue requires the simultaneous presence of three things: 1) there must be cyclic stresses on the component, 2) those stresses must be tensile in nature, and 3) there must be plastic strain. The process of fatigue is considered to consist of three stages: 1) initial fatigue damage (involving plastic strain) leading to crack initiation, 2) crack propagation that continues until the remaining cross section of a shaft (or other component) becomes too weak to carry the imposed loads, and 3) final sudden fracture (stress overload) of the remaining cross section.

     Fatigue failures are insidious because the stresses responsible for crack initiation and propagation are generally much lower than the nominal yield strength of the material; fatigue failures often occur under normal operating conditions and are therefore a big surprise to equipment operators, maintenance personnel, and engineers.

     Although fatigue failures in rotating shafts often have similar crack propagation and final failure mechanisms, the root cause of each failure (i.e., the conditions responsible for fatigue crack initiation) can vary markedly from failure to failure.

     Fatigue failures in shafts nearly always initiate at the surface, generally at points of mechanical or metallurgical stress concentration that locally increase stresses or reduce the material's fatigue resistance. Mechanical stress risers include small fillets, sharp corners, grooves, keyways, and press/shrink fits. (If it were not for stress concentrations, either those designed into a part or those inadvertently caused during manufacturing, many metallurgical consultants would go hungry!) Manufacturing operations such as forging, machining, plating, cladding, and heat treating can introduce metallurgical defects that initiate failure; these metallurgical defects include hydrogen embrittlement, grinding damage, quench cracks, laps/seams, and weld defects. Another very important factor in fatigue initiation is service-related damage caused by corrosion and wear.

     The following case histories describe a few of the rotating shaft failures we have seen in the recent past.

Accumulator Roll Journal Failures

     Sudden failure of an accumulator roll journal on a coated-steel paint line caused immediate concern to our client because he had an additional 27 rolls identical to the one that failed still in service on the same machine. We started our on-site investigation with a nondestructive inspection of the remaining shafts and found cracks in three additional journals; this was a pretty good clue the first failure wasn't just a fluke event that could be handled in a “fix-it-and-forget-it” manner.

     Analyzing the failed and cracked shafts at our laboratory confirmed the failure mechanism was rotating-bending fatigue (as expected, given the operating environment). We found that a sharp machined radius at a diameter change in the journal had acted as a severe mechanical notch and was primarily responsible for the cracking (photo at right). Although the sharp radius was the primary factor in the failure, it wasn't the only factor. Misalignment of one row of accumulator rolls had imposed unexpected bending moments on the journals, increasing the cyclic stresses necessary for the fatigue crack initiation.

     This failure represents one of the classic causes of fatigue failure; a sharp machined radius inducing a mechanical stress concentration. The seriousness of a severe notch cannot be overemphasized; local stresses can easily be magnified 3 to 10 times, compared to the nominal applied stresses.

     We recommended that in the future, the replacement journals be redesigned to include a generous machined radius at the diameter change to eliminate the severe stress concentration. To avoid additional failures on the remaining journals still in service, we recommended they be modified to eliminate the sharp radius. Additionally, we recommended the roll alignment be periodically inspected and adjusted to reduce the bending loads imposed on the journals.

Drum Washer Stub Shaft Failure

     A pulp and paper industry client experienced several 9-inch diameter stub shaft failures on his drum washers. As a result of the failures, the client inspected several other drums and found similar cracking that had not yet led to failure. We were asked to examine the fractured shaft, establish the cause for the failure, and recommend a solution.

     The shaft, which had previously been repaired, was constructed of AISI 8620 carbon steel, with a nickel-based alloy weld overlay and an austenitic stainless steel sleeve on top of the overlay; the sleeve was welded at the ends.

     Our analysis showed the shaft had failed by rotating-bending fatigue; this time, the cracks initiated at the toe of the weld joining the sleeve to the shaft. The primary cause of the cracking was the presence of a significant geometrical stress concentration at the toe of the weld (see photos). Interestingly, fatigue cracks also initiated at the root of the weld, due to diaphragming of the sleeve on the shaft. This failure occurred in spite of the fact that the change in diameter of the shaft was properly executed by machining a generous radius at that location (just beyond the sleeve weld).

     The fillet weld joining the sleeve to the shaft created a very large mechanical stress concentration where fatigue cracks initiated under service conditions. Using the sleeve was a “quick fix” to avoid applying additional weld overlay to the shaft, which of course would have required additional machining to properly profile and dimension the overlay material; unfortunately, the quick-fix sleeve also provided for a quick failure.

Screw Shaft Flange Weld Failure

     Our final example involves the failure of a weld joining a drive shaft to a flange on a lime slaker classifier screw. This particular shaft failed after being in service for only three months. The drive shaft/flange connection had been recently redesigned; the failed connection was constructed according to the new design.

     Our analysis showed that the drive shaft, flange, and weld all complied with the material and manufacturing specifications. Although the weld joining the shaft to the flange failed by fatigue, the final overload portion of the fracture comprised more than half the fracture, indicating the failure location was subjected to relatively high loads.

     The failure was caused primarily by the following factors related to the design of the shaft-to-flange joint: 1) the design of the weld joint resulted in a sharp notch at the weld root regardless of the amount of weld penetration, 2) the weld had high nominal stresses due to the small weld cross section relative to the size of the shaft and journal, and 3) the shaft was welded to the flange on the side of the flange where the highest bending stresses in the shaft were located. These points are illustrated by the cross section of the drive shaft and flange shown in the photograph above.

     We recommended the components be redesigned to eliminate the multiple stress concentrations introduced by the current (new) design; this included eliminating the machined step in the drive shaft and not welding on the side of the flange where the highest bending stresses in the shaft are located. Additionally, the weld joint needed to be larger, because the size of the weld specified was insufficient for the applied loads and the size of the components.

Fighting Evil

     We hope our discussion and examples will help some of you avoid experiencing a surprise shaft failure. Just remember, there is no such thing as a good stress concentration.

Broadcast Towers

     You've all seen the large, red and white radio and television transmitter towers scattered along the West Hills in Portland, but have you ever stopped to consider just how tall they really are? Well, they vary in height, but the tallest ones are over 1,000 feet tall. “So what,” you say? Well, have you ever stopped to think about the poor fellow who has to climb to the top of those towers to change the light bulb? Figure about four hours - 2 1/2 hours up, 10 to 20 minutes worth of work once you're up there, then 1 1/2 hours or so to get back down!

     Recently, our electrical testing division, American Product Safety Co., had an interesting assignment involving two of the newer towers and one older tower. Doing what? Well, unlike the older towers, which only have ladders, the newer towers have elevators, and like all electrically powered devices, state and federal laws require they meet state and federal product safety standards. (Imagine that, right? In fact, these laws apply to nearly all electrical products, from toasters and alarm clocks in the home to nearly every type of electrical equipment in industry.)

     Because the elevators were not manufactured as “listed” products, the law requires a “field evaluation” be performed after final installation to assure they meet the requirements of the appropriate safety standards. Basically, a field evaluation is a thorough review of the entire system, including the electrical system's construction relative to the requirements of the standard. Any nonconformances (and yes, there always seems to be at least a few) are identified so they can be rectified by an electrician, after which we apply a field label certifying the product as installed meets the standard's requirements.

     Oh, and just in case you're imagining one of those fancy, mahogany-lined elevators you might find in a downtown law firm's office, trust us, the ones in the towers are not quite that cushy. They consist of a simple, one-person box, with the sides and floor made of expanded metal (i.e., see-through, wind-through, rain-in-your-face, grating-type material.) There's no piped in elevator music, but the view out is spectacular! And the trip to the top? It takes 17 minutes, or about 2 hours less than climbing!