Author, Brett Weintz. Updated 23 March 2020.
Metal fatigue fracture is an often encountered failure mode of mechanical components. The process involves a repeated/ fluctuating load having a range that exceeds a threshold and a cumulative number of cycles exceeding an (endurance) limit. Both conditions are necessary for a fatigue failure to occur. For example, if a component has suffered a torsional fatigue fracture, then the component had to be subject to a torsional shear stress/ strain of sufficient magnitude with accumulated cycles exceeding a limit. Unfortunately, fatigue typically does not provide external indications of progression (prior to the final few cycles culminating in failure) for conventional condition monitoring systems (save for acoustic emission) to provide an alert. Avoidance (of failure) is generally reliant on traditional periodic visual inspection.
This page provides a brief outline of metal fatigue fracture within the context of rotating machinery failure. A case study is included.
You can find the following Brabon Engineering Services site pages and case studies regarding machinery failure/ damage analysis:
- Machinery failure/ damage investigation discussion page.
- Diesel engine failure discussion page.
- Metal fatigue fracture failure discussion (this page).
- A machinery failure investigation case study.
- A metal fatigue fracture failure case study.
- A sterntube bearing failure case study.
Metal fatigue fracture characteristics
Metal fatigue fracture is characterised by curved ‘beach’/ arrest marks (macro scale) centred on the initiation point(s) and microscopic (micron scale) striations (in ductile metals) across the surface as well as a general brittle appearance. Fatigue fracture of ductile metals do not exhibit yield (necking). The fracture surface is characteristically different to that of an overload failure at macro and microscopic scale. Fractures of castings (typically brittle materials) need to be carefully examined to correctly classify the failure. Where a through fracture occurs there will be a final rupture/ overload area with (perhaps) shear lip(s). ‘Beach’ marks are generated by changes in the loading pattern during the fracture propagation.
The shape/ orientation of fracture face indicates the type of stress having caused propagation. That is, a fracture face perpendicular to the axis of a shaft would suggest rotating bending fatigue, whereas a spiral fracture face would be associated with torsional fatigue. It should be noted that the surface shape may change with variations in the dominant stresses as the crack propagates through the component section.
Understanding the cause(s) of a fatigue fracture can hinge on examinations of the initiation point(s) to identify features that may be associated with initiation. For example, metallurgical imperfections like non-metallic inclusions/ voids or surface damage marks (having occurred prior to failure). However, smearing of the surface due to contact/ abrasion between the two faces of the fracture can obliterate details of features. Examinations of the component section(s) adjacent to the fracture for general characteristics, e.g. an evaluation of (location) distribution of non-metallic inclusions, may assist in indicating the initiation mechanism.
It is worth noting that all structures contain defects. Manufacturing and quality control examination processes aim to prevent defects smaller than the maximum significant size occur in regions of elevated mean and cyclic stress. It should also be noted that adverse environments typically have a significant effect on the fatigue strength of any component, e.g. corrosive conditions or microscopic surface damage/ cracking, say due to fretting.
An assessment of the cycles and estimated stress levels can provide correlation of the initiation mechanism, e.g. if the estimated load cycles on a steel component (total hours run since commissioning x operating speed x harmonic order) significantly exceeds the typical endurance limit of 10 x 10^6 cycles (for polished steel specimens), then initiation may not have involved a manufacturing imperfection, but perhaps an adverse event during the service life.
Metal fatigue fracture failure case study
Diesel engine and incident
The case involved a large four-stroke, vee form, medium speed, trunk piston engine prime mover for a diesel generator. While running on load at about 90% of mcr, the oil mist detector was triggered causing an automatic shutdown of the engine. Fortunately, the operators decided to open each of the crankcase doors and conduct visual inspections. One of the crank pins was found to be fractured which had consequentially caused failure of both (A and B bank) connecting rod large end bearings.
Inspection of the crank pin after dismantling the running gear found two cracks that crossed the drive end crank web fillet radius at either side of the 6 o’clock position (with the crank pin at top dead centre). The cracks joined on the crank pin journal and spiralled through 90° in a counter to drive direction along the crank pin to the lube oil outlet cross-drilling. The crack extended down the oil outlet with multiple bifurcations radiating from the outlet. Given the rotational direction driving torque from the cylinders towards the non-drive end would tend to open the cracking.
The machinery monitoring trend history showed a satisfactory level of torsional vibrations at the engine flywheel prior to the failure. That is, the torsional vibration damper was likely to have been functioning effectively. The engine output flexible coupling was also dismantled for inspection and found to be in satisfactory condition.
The engine had been running for about four hours at the time of the failure. The engine had accumulated about 9,900 running hours and 400 starts prior to the failure. The large end bearings had not been recently dismantled and all had been fitted during original assembly of the engine. The first maintenance overhaul was not scheduled for another 2,000 running hours. In addition, the crank web fillet radius was recessed below the crank pin surface and into the web face such that the radius surface was below the point where the face and pin surfaces intersect. Thus, if a large end housing was to move axially to contact the crank web, then the bearing could not normally contact the fillet.
Fatigue life estimate
Considering the accumulated running hours, then one cylinder would have produced 148 x 10^6 cycles. Six cylinders (from the non-drive end) would result in 888 x 10^6 cycles at the fracture site. The estimates exceed the mean endurance limit of 10 x 10^6 cycles for polished carbon steel specimens. However, there is typically a significant distribution in individual endurance life results and given the crankshaft was an alloy steel, then the time of the failure was likely consistent with an original manufacturing defect.
Each crank web fillet to each crank pin and main journal was subject to non-destructive testing using the phased array ultrasonic technique. Indications (of internal defects) were found in fillet that had fractured as well as two other fillets.
A section of crankshaft including the failed crank pin was submitted to a laboratory for metallurgical examinations. While the fracture faces were found to be mechanically abraded, there were beach marks and arrest marks characteristic of fatigue fracture. Sub-surface initiation points were found in each of the (two) cracks that transitioned the crank web fillet. Each crack in way of the initiation points was angled parallel to the main spiral, i.e. both initiations involved torsional stresses. There was no (external) surface damage to the fillet in way of either initiation site.
Metallographic sections in way of each crack initiation site were examined using an electron microscope. At crack Z1 (non-metallic) sulphide and oxide inclusions were found in the initiation area with one sulphide inclusion about 0.3mm long that intercepted the apparent initiation site 0.2mm below the fillet surface and which would be a sufficient size to cause initiation. At crack Z2 one small artefact was found in the area of the initiation site.
Steel cleanliness evaluations using DIN 50602 , method K, rating 4, were conducted on metallographic sections in way of crack Z1 and Z2 initiation sites as well as 75 mm below the crank pin surface. The overall results for all three samples were K4-oxygen 45 and K4-sulphur 52, giving a total K4 index of 97. The manufacturer’s inspection certificate for the replacement crankshaft specified a maximum allowable overall K4 index of 50.
(1. DIN 50602, Microscopic examination of special steels using standard diagrams to assess the content of non-metallic inclusions.)
An etched macro section through the crank web and crank pin found an adverse grain flow which intersected the crank web surface resulting in non-metallic inclusions at the crank pin fillet surface.
It was concluded that the crankshaft had suffered a torsional fatigue failure had been due to metallurgical imperfections associated with forging defects in the original manufacture.
Have you suffered a sudden inexplicable machinery failure?
Brabon Engineering Services will make a thorough investigation to discover the root cause of the failure and help prevent further incidents.
Call for a discussion: +353 87 383 5043
email for a proposal: email@example.com
Send a message via our contact page
Connect on LinkedIn
Metallurgical laboratory examinations are closely managed to avoid nugatory work and costs.
The following additional case studies are available on request:
- Engine Crankshaft (Fatigue) Failures, July 2019
- Intermediate Shaft Fracture, June 2019
- Propeller Shaft Fatigue Failure, August 2020
- Ship Rudder Stock Problems, July 2019
- Sterntube Aft Bearing Whitemetal Fatigue Damage, May 2019
The Wikipedia article regarding metal fatigue (linked) provides discussion of the fundamental metallurgical process and contributory factors as well as history and notable historical failures. Wikipedia also have a separate article concerning fracture mechanics.
There are continual advancements in understanding of the fundamental metallurgy of material fatigue and state of the art information can be found from the IMechE ( Proceedings, Part C: Journal Of Mechanical Engineering Science), ASME and other institutions. There is also the ESIS (European Structural Integrity Society). Numerous open source publishers operate on-line, including MPDI and Academia.