Anatomy of a failure

A wire rope broke in service whilst lifting a load of reinforcing steel that weighed between 2.5t-3t. This load was well within the capacity of the crane. The precise sequence of events leading to the failure was not known, but the load did not reach the ground because the rope jammed in the gap between sheave and support bracket. The rope was two years old and was stated to be maintained by regular lubrication. Bounce of the load was stated to have been possible during lifting. The rope sheave diameter was 520mm, and it was known to have been in service longer than the rope.

The insurance loss adjustor (whose job is to advise the insurance company on the quantum of loss and whether the existing policies covered the causes and consequences of the failure) commissioned a failure investigation to ascertain whether the rope was overloaded at the time of failure, or whether poor maintenance had been responsible for premature breakage. The resolution of these issues would determine whether or not insurance cover existed for the various consequences of the accident, and would also indicate if an increase in premium was necessary to offset a possible higher level of risk.

Three pieces of rope were supplied to assist in this investigation – these comprised the two broken ends together with a section of rope taken well away from the failed ends. The purpose of this latter piece was to use tensile testing to check the load capacity of the rope at the time of failure.

Visual Observations

The rope was a general engineering 18 strand non-spin type designated as 12×7(6/1)/6×7(6/1) – see the illustration in Figure 1 below. This designation is a shorthand form which summarises the information contained in the image. Thus this rope had an inner layer of 6 strands of wires, with each strand comprising 7 wires wrapped as 6 outer wires around 1 inner wire, while the outer layer is formed by 12 strands of wires wrapped in the same way. The inner core of the wire is fibre.

As received, rope lubrication was deficient to dry, with slight corrosion evident on the outside of the rope. Close inspection of the rope near to the fracture plane showed that a number of wires were cracked in outer and inner strands on both sides of the fracture (Figure 2). Cracks on both inner and outer layers were associated with flattened regions on the wires. Cracking was also observed on wires well away from this region also associated with flattened regions (Figure 3).

All parties were agreed that the rope should not have failed under a load of 3 tonnes, so an immediate action was to cut a section out from the rope and send it for tensile testing at a specialist facility. Alongside this process, it was necessary to determine the original grade and size of the wire rope.  This would then establish what degradation of properties had taken place over the service life, and provide an indicator of the severity of service and quality of maintenance.  The only information that the operator could supply was that the rope was in the 1770 MPa grade category.  Table 1 gives manufacturer’s data for wire diameter, nominal rope diameter and tensile strength for their range of 1770 MPa ropes.

In order to establish the original failure load of the wire rope as supplied, it was necessary to measure the diameter of wires near to the break (average approximately 1.5mm) and the rope diameter (approximately 21.5mm) and to deduce the most likely tensile strength, allowing for the rope stretch (and consequent reduction in diameter) that would have occurred during the failure. Wire diameter should be less affected than rope diameter, except immediately adjacent to the fracture plane of the wire. An inspection of the data in Table 1 leads to the conclusion that the rope was most likely to have originally been 24 mm in diameter, with 1.5mm wire diameter (measurement error with Vernier callipers is at least 0.02mm) and a breaking load of 332 kN.

Tensile testing was performed on a 1.5m length sample using a wire rope testing machine, giving a measured failure load of 232 kN. It should be noted that this load did not represent complete failure of the rope, but rather fracture of 11 strands (66 wires) out of a total of 18 strands (108 wires).  Two inner and five outer strands remained unbroken. Examination of the 66 wires that broke indicated that only 10 of them had a flat fracture surface, which would be indicative of the presence of an initial fatigue crack, with the rest showing ductile failure modes. The reduction in tensile strength is around 30% and the evidence from the broken wires indicates that around 15% of them would have had their breaking load greatly reduced by the presence of fatigue cracks. Thus the presence of fatigue cracks and lack of lubrication provide a rationale for the observed rope breaking load. Nonetheless, this breaking load of around 23 tonnes is very significantly higher than the load being carried at the time of the failure.

Initial Inferences

At this point in the investigation enough information had been gathered to consider some possible contributory factors to the failure. These are given below and comments made as to their feasibility and contribution:

• High strand: this is a condition in which one or more strands stand proud of the body of the rope and hence are worn before adjacent strands. This is usually caused by improper socketing of the rope, or seizing, kinks or dog-legs. Examination of Figures 2 and 3 shows that the flattened regions are not wear damage and hence are not due to a high strand condition. Close inspection reveals cracks associated with the flat regions. Little evidence of any severe abrasion is evident.

• Small sheave: Premature fatigue cracking of wire ropes can be caused by repeated bending over a sheave that has too small a diameter for the rope. The manufacturer’s recommendations for non-spin ropes were that the sheave diameter (520mm) should be at least 18x the nominal rope diameter of 24mm. The ratio in the present case is >21:1, so this should not have been a factor in this failure. This observation is supported by appearance of the rope in Figure 3 which supports fatigue cracking over a sheave of normal diameter. Fatigue cracking over a small sheave leads to individual wires being bent out of place.

• Fatigue over normal sheave: The observed damage to the rope appears to correlate with fatigue damage occurring through bending under normal loads over a sheave of the correct size. The flat regions are surface deformation, possibly occurring as a result of groove wear during service. Plastic deformation to regions of the surface of some wires has occurred during service. Fatigue cracks have been induced at some of these hardened regions during bending over the sheave.

The inference to be drawn from this investigation thus far is that fatigue cracking was second order factor in this failure. During tensile testing of the rope some fatigue cracking of wires was present and the breaking load was reduced by some 30%. Nonetheless, the failure load of 232 kN (23 tonnes) is much higher than the load reportedly being lifted at the time of failure (2.5-3 tonnes). The fatigue cracking observed in the wires is unlikely to have been a direct contributor to the failure, unless one rope section was more defective than indicated by this single tensile test.

To provide confirmation that the section of rope where the failure occurred was no more susceptible to low load failure than the section that had been subjected to tensile testing, a number of the individual broken wires were subjected to a fractographic examination of their mode of failure.

Fractography

Fractography is the study of fracture surface features to determine causes and mechanisms of failure. The total number of wires in all rope strands was 108, and a total of 31 wires (29%) were selected for examination; 20 wires being taken from the outer strands and 11 from the inner strands. This higher weighting towards the outer strands reflects the likelihood that fatigue cracking would be more prevalent in these strands which directly experience contact with the sheave. The fractured ends were cut off, and ultrasonically cleaned in de-rusted and de-greasing agents, before being examined at low magnification using stereo binoculars, and at high magnification in a scanning electron microscope (SEM).  Typical SEM observations of the fracture surface failure modes are given below at both a low and a high magnification, together with information on the number of wires in the sample which exhibited similar fractures to the mode shown.

Type 1: Tensile cup-and-cone fracture – 3 occurrences (10%) – 1 wire in outer strands; 2 wires in inner strands (Figures 4 and 5).

Type 2: Flat twisted failure – 2 instances (6%) in outer strands (Figures 6 and 7).

Type 3: Flat semi-elliptic regions present – 26 cases (84%) – 17 in outer strands; 9 in inner strands (Figures 8 and 9).

Fatigue crack initiation and growth is the explanation for the presence of the small semi-elliptic regions seen on many of the selected wires. They are associated with the flattened regions, but fatigue striations cannot be resolved in this fine scale quenched and tempered microstructure. There are subtle differences between the appearance of the MVC (microvoid coalescence) and fatigue fractographs. Such defects would reduce the breaking load of the individual wires considerably, but note that the remainder of the fracture surface shows evidence of ductile fracture, indicating that the toughness of the individual wires remains high. The interpretation of this data, however, is that the existence of fatigue cracks is likely to have led to a greater reduction in load capacity of the rope at this location than was recorded during the tensile testing.

It is now worth considering what part rope bounce might have played in this failure. There are two possible effects of rope bounce, depending on whether the rope stays in the groove, or whether it jumps out of the groove.

• Rope stays in the groove: Any bounce of the load would add a transient impulsive force. In extreme cases, this may double the load being lifted, giving a possible upper limit to the apparent load of perhaps 50-60 kN. If the measured breaking load for the piece of wire rope had been around 50-60 kN rope bounce may have caused the failure to occur. However, the actual breaking load was measured as 232 kN.

• Rope jumps the sheave: The original report on the failure stated that the load did not drop because the rope jammed between the sheave and the support bracket. The interesting question is whether this jamming occurred after the rope broke or whether it was a direct contributor to the failure. Such a severe effect of rope bounce would imply poor lifting practice but it is possible and, if it had occurred, would cause high loads on the rope if the crane operator did not detect it and there was no trip mechanism installed to prevent continued winding in such an event.

Summary and Conclusions

The information gleaned so far from this case study is summarised below:

1. Overall, the strength of the rope was reduced by the presence of fatigue cracks – this is evidenced by the observed tensile strength of 232 kN compared with the manufacturers stated breaking load of 332 kN. The section that broke may have had a lower tensile strength than this value, as the relative number of wires containing fatigue cracks is higher.

2. The observed breaking load would still have been very much higher than the stated load being lifted at the time of failure (some 25-30 kN). Thus the rope must have failed through application of an overload relative to its current strength level. The failure was not solely due to the presence of fatigue cracks (whose existence is fairly normal in wire ropes and explains the requirement for regular maintenance and high factors of safety).

3. The cause of this overload is not clear, but bouncing of the load might have allowed the rope to jump from its groove and jam between sheave and boom during winding. If this state of affairs could exist for a short time undetected, and the rope winding was continued, a very significant overload could be applied to the rope.

4. The cause of the fatigue cracks needs clarification. As noted above, they can initiate as a result of bending stresses induced by too small a sheave diameter. The recommended diameter is 18x rope diameter which equals 432 mm for the resent rope. The actual sheave diameter was 520 mm, which should have been sufficient. As the sheave was older than the rope, however, it is possible that wear of the sheave groove has had an influence. Fatigue cracks occur at deformed surface regions where ductility has become exhausted, particularly if surface damage from abrasion occurs. This would be exacerbated by any decrease in sheave groove diameter, which could occur by wear during service, and by poor lubrication practice (which was apparently the case).

The conclusion to be drawn from this investigation is that the presence of fatigue cracks has lowered the breaking load of the rope by >30%. However, the breaking load would still have been very much higher than the stated lifting load of 25-30 kN. The fractographic work has indicated that ductile fracture occurs over some part of the fracture surface in all wires, demonstrating that the rope metallurgy is up to specification. The most likely cause of the fracture therefore seems to be rope jamming between sheave and boom, probably due to bounce during lifting.

In insurance terms, poor maintenance is not a prime cause of the failure, which would have been “sudden and unexpected” when it occurred. Cover should exist for such circumstances.