Crack arrest concepts - Part 2

TWI Bulletin, August 1987

Tony Willoughby

Tony Willoughby, MA, PhD, DIC, MWeldI, is in the Fracture Department of the Engineering and Materials Group.

This article describes the results of a recent experimental programme at The Welding Institute to investigate the crack arrest capabilities of 9%Ni steels.

Based on a paper presented at EWI/TWI North American Seminar on 'Engineering performance of welded joints' 1986 October, Columbus, Ohio, USA.

Crack arrest in 9%Ni steel

The Welding Institute has recently completed a study of the crack arrest properties of 9%Ni steel and weldments, as used in the construction of cryogenic storage tanks containing liquefied natural gas (LNG). This project formed part of an international programme sponsored by the Gas Research Institute of Chicago, with additional work being carried out in the USA and Japan.

Considerable effort in the TWI project was aimed at establishing crack arrest properties, using structurally representative double tension tests, with the test plate measuring approximately 1m long by 600mm wide. The majority of these tests was carried out at -162°C, the normal temperature of LNG. Tests on parent steel were accomplished by guiding the crack down a brittle electron beam melt run, 100mm long, before allowing it to run into the parent material. Tests on weldments were carried out in the same way, except that on leaving the EB melt run the crack propagated into the HAZ of the test weld, which was parallel to the direction of crack propagation. In addition to the double tension tests, a range of small scale tests including Charpy, DWT, DWTT and CCA were carried out on parent materials and weldments.

The materials used in the programme comprised a total of nine different quenched and tempered parent steels, five of which were examined in detail, and four weldments. The parent steels were designated with the letters L, M or H,for low, medium and high impact toughness respectively and by the thickness in mm. Thus L25 steel was the low toughness, 25mm thickness material. All the steels examined were commercial heats, except for the L25 which was produced specially for the programme, so as to be near to the minimum Charpy requirement in ASTM A553 Type 1. The welding consumables consisted of two nickel based (Inconel type) fillers, welded at two different heat inputs.

Figures 18 and 19 compare the results of Charpy V notch impact tests and drop weight tear tests with K _a values measured from CCA tests, for parent steels. It is apparent from Fig.18 that Charpy energy and K _a show comparable trends, but that there is an effect of thickness. This is not surprising since the CCA specimens, which were not sidegrooved, used the full section thickness whereas the Charpy specimens used a constant thickness of 10mm. The DWTT, which may be likened to a full thickness Charpy, should remove this thickness dependence from the correlation. Figure 19 shows that this is to some extent true, but that there is an anomalous result for the L25 steel.

Early on in the programme it became apparent that the majority of the steels exhibited exceptionally good crack arrest toughness, even at -162°C, so large scale testing was concentrated upon the L25 steel, which showed the lowest toughness. Five double tension tests on the steel were performed, in three of which the crack propagated completely through, and in the remaining two it arrested. A total of 15 other double tension tests, seven on the parentsteels and eight on weldments, were carried out, and in 14 of these crack arrest occurred. In the 15th the crack appeared to arrest for a period of 18msec, and then jumped sideways into the softer weld metal and tore through it at a relatively low speed. The weld metals used were of lower strengths and relatively low tearing resistances compared with the parent steels.

Figure 20 compares the estimates of K _a obtained from CCA tests on L25 and M28 steel with the statically calculated stress intensity factors at the point of propagation into the parent steel in the double tension test. Agreement is good, especially for the L25 steel. This would suggest at first sight that static calculations of K are sufficient for predicting the behaviour of wide plate tests.

Dynamic finite element analyses were carried out on several of the wide plate tests, and the results showed that the discrepancy between dynamic and static estimates of K differed by less than 10%. This was probably a result of the design of the specimens which minimised the effect of stress wave reflection on the moving cracks. It does not imply therefore, that static analysis is always adequate. Dynamic analyses of the Ks at arrest in the CCA specimens were not attempted in the TWI project. As described in the previous section, however, the static K a should provide a conservative estimate.

At normal design stresses of approximately 200 N/mm ² (30ksi) present in LNG tanks, all the parent steels except L25 were demonstrated to be capable of arresting brittle cracks with half lengths of the order of 100mm. The L25 steel, when welded with Ni basedconsumables, gave HAZs with relatively poor Charpy properties (Cv ≅ 10-15J at -196°C). However, it proved impossible to run brittle cracks down these HAZs, indicating that they should not jeopardise thesafety of tanks made from such materials.

The main conclusions of this project were that the LNG tanks currently in service should have adequate resistance to rapid crack propagation, and that tanks made from steels today should be extremely safe. Materials of relativelylow crack arrest toughness such as the L25 steel could be rejected on the basis of small scale tests such as DWT and Charpy. The criteria would of course be valid only for steels of similar type to those tested.

Future work

Considerable research remains to be done before methods relying on the arrest of unstable cracks become as generally applicable as those depending on the prevention of initiation of unstable fracture. Improvements in small scaletests and their interpretation are urgently required. Crack arrest procedures are being developed for arresting small through thickness cracks in large structures, where the restrictions of linear elastic fracture mechanics are capableof being met, or where the materials are on the upper shelf of toughness, so that energy balance considerations appear paramount. In the intermediate situation, where crack size and section dimensions are of the same order,elastic-plastic fracture mechanics becomes important but has received relatively little attention.

Another area of concern is that of weldments. Most work to date has concentrated on parent steel for two reasons. Firstly, crack arrest in parent steels is a necessary prerequisite for a crack arrest procedure, and secondly, it iscommonly believed that brittle cracks which initiate in welds tend to run into parent steel. The latter belief, however, is not supported by evidence of service failures, as discussed in ref. ^[23] Cleavage cracks can run along brittle heat affected zones, if the weld is stressed transversely, particularly when the weld toe is at a stress concentration. Crack arrest in weldments is thus anessential requirement for a complete crack arrest design.

A current Group Sponsored Project at TWI, titled 'Crack arrest in modern C-Mn steels', is aimed at developing improved correlations between small scale and large scale tests for base metals, using steels of interest to the cryogenicstorage tank, offshore and shipping industries. Recent development in steelmaking practice have resulted in steels being produced with exceptional toughness as measured by conventional tests such as Charpy impact. These measurements,however, may bear little relationship to the crack arrest performance. It is intended to follow this work with a project to examine the crack arrest behaviour of weldments in similar materials.

References