John Harrison obtained a 1st class honours degree in mechanical sciences from Cambridge University in 1957. After a period working for a firm of civil engineering consultants he returned to Cambridge to attend a postgraduate course in structures and strength of materials. Joining TWI in 1962, he worked on fatigue of welded structures, concentrating in particular on the significance of weld defects. He obtained his PhD from Cambridge University for work on this topic.
He then became responsible for TWI's fracture research, eventually being promoted Head of Engineering Research, covering design engineering, fatigue, fracture and non-destructive testing. In 1985, the Engineering and Materials Group was formed, comprising the Design Engineering, Fatigue, Fracture, Non-Destructive Testing and Materials Departments, and John was appointed Group Manager. The Group employs a total of 90 staff of whom 30 are university graduates and 15 have Chartered Engineer status.
John Harrison has published over 50 papers on fatigue and fracture and has been extensively involved in national and international committees. In 1988 he resigned Chairmanship of Commission XIII of the International Institute of Welding which deals with the fatigue behaviour of welded structures and components, a position which he held for 15 years.
Engineering failures often occur as the result of a combination of circumstances. Seldom, however, can the designer and manufacturer of equipment have created such an unfortunate combination as readily as they did here, as John Harrison describes.
When several years ago the driving shafts of a number of winches failed on a semi-submersible pipe-lay barge, TWI was called in to investigate. The barge itself was a marvel of modern engineering, built at a cost of about $30M, and designed to be able to continue to lay pipe in severe weather conditions. Lay barges use a spread of anchors to control their position. Their only motive power comes from the winches which advance the barge by pulling on the forward anchor cables and slacking off on the after ones. Work boats are constantly used for shifting the anchors as pipe laying progresses. The barge had a spread of 14 anchors with associated winches. Like everything else on such barges, the winches were big, each capable of exerting loads of over 125 tonnes, with 1.0m diameter barrels, 4m diameter check plates and 300mm diameter driving shafts.
Shortly after starting to lay its first pipeline, one of the barge's winch shafts failed. Because the insurance cover required the barge to have ten operating winches and because the failure had occurred so rapidly, it was decided to abandon the pipe and return to sheltered waters whilst the cause of failure was investigated. During the abandonment of the pipe a second shaft failed, and on arrival at base it was found that a third had also cracked.
Examination of the fracture surface indicated a torsional failure of extreme brittleness. Figure 1 shows the shaft sited in the hub of the so-called 'bull ring' and with the key in position. Figure 2 is taken after the shaft had been removed from the bull ring and shows helical cracking caused by torsional failure.
The material was subjected to Charpy V notch impact and fracture toughness testing. One of the Charpy specimens broke into three pieces with the notch playing no part in the fracture! Charpy specimens were 100% crystalline up to +120°C and 50% crystalline at +190°C. Even at +250°C, the absorbed energy was less than 40J. A Charpy transition curve is given in Fig.3.
Fracture toughness tests carried out at +15°C gave K Ic values of 930-1200 N/mm 3/2 (or 30-38 MPa √m for those more familiar with this unit). Effectively the material was as brittle as a proverbial carrot. These were the lowest values of fracture toughness that TWI had ever measured at that time.
The chemical analysis was found to be out of specification and the heat treatment had clearly been in error. The mean grain size was greater than 160µm or larger than ASTM No 2 - coarse by any standards.
The deficiencies in the material were compounded by the design of the keyways. These were 75mm wide by 25mm deep, but their outstanding feature was the sharpness of the corners. These were virtually right-angles. When magnified on a shadow-graph the radius was about 0.02mm (less than one thousandth of an inch). That takes some machining and must have required the machinist to re-sharpen his milling cutter very frequently. The stress concentration factor due to this very small radius is estimated at 15 to 20. For shafts of this diameter, American and British Standards would call for keyway radii of about 6mm and, had such a radius been provided, the stress concentration factor would have been much lower, just over 3.
The cause of failure was clearly identified as a very brittle material associated with a high stress concentration.
After our studies, TWI's Dr Richard Dolby and the writer were asked to attend a meeting at the offices of the barge owner. The designer's main contribution concerned the fracture surface. In his words: 'Look what the loading has done to that material - it's turned it crystalline!'
He would have been a wiser man if he had studied the writings of Rankine [1] published in Volume II of the Proceedings of the Institution of Civil Engineers, 1843. Rankine was studying the reasons for premature failure of forged railway axles. He had surmised that it was due to the sharpness of the corner at the shoulder of the axle. He carried out tests using a drop forge as a testing machine and found that axles with sharp corners invariably failed on the first blow of the hammer, whereas those with more generous radii always survived at least five blows! Furthermore, he examined the material of the shafts in a microscope before and after testing and could find no indication that the loading had in any way altered what he called its fibrous nature. In the weeks that followed, TWI provided considerable assistance to the barge owner in selecting materials for the replacement shafts which were installed in reasonable time. In the intervening years the barge has given sterling service.
