Corrosion initiated rotating bending fatigue failure of a fertiliser conveyor belt head roller

The following describes the experimental process and findings of a forensic investigation of a failed fertiliser plant conveyor belt head roller shaft. 

A 25 year service life was expected but failure occurred following only 10 years. A gearbox which provided direct power to the roller had been replaced only several weeks earlier. The primary aim for the plant owner was to determine if the gearbox replacement had caused failure of the shaft, or, if the two incidents were unrelated. If incorrect instalment of the gearbox had caused failure, liability would fall on the gearbox suppliers in terms of mitigation.

However, it was revealed that the shaft had failed due to low stress rotating bending fatigue, over an extended period of time. A premeditated change of material selection at the manufacturing stage, substituting carbon steel for 304L stainless steel, resulted in reduced corrosion resistance. Fertiliser ingredients tend to pose limited problems in their dry form in terms of corrosion; the presence of moisture however can change the ingredients into aggressive corrosion species. Thus, corrosion allowed multiple fatigue cracks to initiate from corrosion pitting. The cracks then eventually joined together to form a single fatigue crack which propagated through the cross section. 

The design drawing for the head roller specified the shaft component should be forged 080A22 carbon steel (formerly EN43) with the end plate, drum wall and boss constructed using grade 304L stainless steel. All welding was specified as tungsten inert gas method (TIG). The jointing of stainless steel to the carbon steel was to be made using a 309S92 stainless steel based filler rod and 304L to 304L made with 308S92.

Macroscopic Examination

The first stage of an investigation following the gathering of evidence and operating information is the visual examination of all components. Key features presented by the fracture and ancillary components can provide vital preliminary facts relating to the type of failure, even before investigative techniques such as optical or high powered electron microscopy are performed. 

Failure of the head roller was due to full wall fracture of the shaft on the bearing journal side, i.e. the gearbox side. The fracture location was not associated with geometry changes or keyways, a common failure location, but was in the weld region of the 304L stainless steel boss.

Some of the findings included:

• General corrosion on the majority of the external surfaces.

• A shallow band of corrosive attack approximately 2mm in depth was observed around one half of the fracture. 

• Orientation perpendicular to the shaft axis indicating fatigue propagation. 

• Relatively flat surface across the majority of the fracture.

• Small, ~10% of the cross-sectional area (c.s.a), jagged area. 

Shaft failures are well documented and the visual features can be analysed and translated to determine the not only the quantity but also the type of stress which resulted in initiation and propagation. The features indicated a fatigue mechanism. Causes of fatigue may be the result of misalignment or other bearing related issues. 

Scanning Electron Microscopy (SEM) & Energy Dispersive X-ray Analysis (EDX)

EDX analyses confirmed the corrosion was related to the presence of dried fertiliser ingredients which had most likely been hydrated by air moisture or other sources.

SEM examination revealed multiple ratchet marks around the edge of the fracture approximately 180 degrees from the final failure point. Ratchet marks are features associated with fatigue propagation from multiple origins. Individual fatigue cracks which originate on slightly different planes, propagate and eventually join together, producing step like features termed ratchet marks. Shaft fractures exhibiting multiple initiation points are often the result of a rotating bending load.

In the central region of the fracture surface the introduction of a mixed mode mechanism was seen. Secondary tears and the onset of cleavage, due to increased crack speed were identified; the speed increase was due to a continually reducing c.s.a. The mixed mode mechanism is a result of variations in rotation speed and loading placed on the roller, as a crack continues to grow the applied stress remains constant but the localised stress on the remaining section increases resulting in the small secondary tears (localised overload) and isolated cleavage.

The final stage of fracture (~90% c.s.a.) was sudden overload when the remaining section was not able to support the stress. Ductile overload results in jagged morphologies with a deformed and twisted appearance, a result of the two surfaces pulling apart. Microscopically, this produces the cup and cone dimples known as microvoid coalescence. 

Optical microscopy

Fracture surface

The fatigue origins were shown to have initiated at the weld toe / heat affected zone (HAZ) of the shaft to boss weld. HAZ regions are typically considered stress concentration points due to a change in mechanical properties and microstructure resulting in a more brittle material. Corrosion pits are a further cause of fatigue initiation, as the pit acts as notch or stress raiser. Therefore, corrosion of a stress critical weld region further increases the susceptibility to fatigue initiation and propagation.

Microstructural analysis

The shaft material possessed a ferrite pearlite microstructure with elongated manganese sulphide inclusions running in the longitudinal direction of the shaft length, typical of forged EN43 bar stock. However, the microstructure of the boss material was not an austenitic structure typical of 304L stainless steel, but possessed a banded ferrite pearlite structure, i.e. low carbon steel.
The end plate did however possess a structure consistent with rolled austenitic 304L stainless steel plate.

Chemical analysis

The composition of the shaft material was within specification for 080A22 low carbon steel. The boss however possessed a composition similar to 080A15 (formerly EN3) low carbon steel and not the specified 304L stainless steel.

Analyses of the welds showed a carbon steel filler rod had correctly been used join the boss and shaft, and not 309S92 stainless steel, Table 2. This indicated that the material change had been a pre-conceived alteration as the welder was aware that carbon steel filler was required at the time of manufacture. Analysis of the weld metal joining the boss to the 304L end plate was consistent with 308S92 metal specified for welding carbon steel to stainless steel.

Conclusions

1) Failure occurred due to corrosion initiated fatigue initiating from multiple points within the notch sensitive heat affected zone of the shaft to boss weld. The fracture was consistent with low stress rotating bending.

2) Sudden failure due to a gearbox replacement was ruled out.

3) The substitution of stainless steel for carbon steel in the boss and weld reduced the corrosion resistance and mechanical properties, particularly at a stress critical zone.

4) Correct material usage would have reduced the risk of corrosive attack and improved fatigue resistance. However, 10 years of service may also have been extended if fertiliser build-up had been systematically removed and all surface cleaned off, particularly in the event of the product becoming hydrated.