A cautionary tale on propagating brittle fracture in pipelines

A cautionary tale is comprised of three essential parts: an act, location, or object is said to be dangerous; someone disregards the warning and performs the forbidden act; and, as a result, is subjected to an unpleasant fate.

The German author Heinrich Hoffmann was famous for his cautionary tales in his book “˜Struwwelpeter’ [1] where he told such stories as “˜Flying Robert’ and “˜The Dreadful Story About Harriet and the Matches’. In Harriett’s case the taboo was playing with matches and the unpleasant fate that awaited her when she disregarded the warning is illustrated graphically in Figure 1.

The danger in this cautionary tale about pipeline design is propagating brittle fracture. Two kinds of long propagating fracture can occur in pipelines. Both kinds can extend for hundreds or even thousands of metres. Brittle fracture occurs by a cleavage mechanism at speeds of around 1,000 m per second with a crystalline appearance and near zero levels of ductility. Ductile fracture can extend rather more slowly at several hundred metres per second in a shear mechanism with substantial plastic deformation, and with a fibrous fracture surface appearance.

Any pipeline engineer would be keen to avoid having their new pipeline fail spectacularly the way the SS Schenectady did when it broke in half before it even went to sea, shown in Figure 2.

Closer to home for the authors, the King St Bridge in Melbourne, Australia suffered a brittle fracture on a frosty morning in 1962. The fracture was initiated by heat affected zone hydrogen assisted cracking at a transverse fillet weld on a cover plate end [2].

Brittle fracture, unlike ductile fracture, can occur at local stress levels well below yield and is driven entirely by elastic strain energy. Brittle fracture can be driven by elastic strain energy alone, which means that crack arresters cannot be used as a practical design method to control brittle fracture. The only controls are either adequate material toughness or the use of a low design stress, for example below 85 megapascals, as provided by AS 2885 Part 1 (AS 2885.1)[4].

Brittle fractures in gas pipelines

In the 1950s and 1960s there were a number of long propagating brittle fractures in gas pipelines. There was a 5.6 km brittle fracture of a 609 mm pipeline in Venezuela in 1958 and
a 13 km brittle fracture of a 762 mm pipeline New Mexico in
1960 [5, 10].

A full scale brittle fracture propagation test (called an Athens test) in 1965 on DN150 line pipe clearly demonstrated that brittle fracture propagation was not limited to large diameter pipes. Of the five tested joints, all presented cleavage fracture [5].

The pre-1960s brittle failures led the American Gas Association (AGA) to undertake a program of research that led to the introduction of the drop weight tear test (DWTT), which has since been incorporated into standards like the American Petroleum Institute (API) 5L and AS 2885.

The DWTT

Unfortunately, however, while AS 2885.1 and other national standards require DWTT on small diameter pipe, for reasons that are not clear, ISO/API 3183/5L does not require it for pipe sizes less than DN500. Although it is stated that “…sufficient shear-fracture area…is
an essential pipe-body property to ensure the avoidance of brittle fracture propagation in gas pipelines,” DWTT is only by agreement for sizes of DN500 and above.

This is a serious problem because there have been cases of brittle fracture in small diameter pipe [6], and to quote a mantra sometimes used in the Standards Australia Petroleum Pipelines Committee ME38, the API 5L exemption from DWTT for small diameter pipe is “a law of man, not a law of nature”. There is no basis in fracture mechanics for the exemption.

Can DWTTs be performed on small diameter pipe?

For several years now, the APIA Research and Standards Committee (RSC) has been unsuccessfully advocating to ISO/TC 67/SC 2/WG 16, the committee responsible for ISO-API 3183 and API 5L, for a change to reduce the threshold diameter for DWTT. This has been resisted, especially by manufacturers of seamless pipe, and by others that cite difficulties in conducting the test on small pipe due to problems with flattening and associated changes in properties that could affect the outcome of the test. In response to this, the APIA RSC and the Energy Pipelines Cooperative Research Centre (CRC) have been conducting research on size effects in DWTT and recently published a paper [7] at the Joint Technical Meeting of APIA, the European Pipelines Research Group and the Pipelines Research Council International. The results of that work will be presented further on in this cautionary tale.

In this context, the present authors concede that some product types, especially quenched and tempered seamless pipe, exhibit inherently high levels of initiation toughness, making valid DWTT results difficult to obtain, and placing large demands on the capacity and reliability of the testing machine. The same situation may happen in small diameter, thick walled pipe in general. Nevertheless, unless testing is performed, we would argue that adequate toughness to prevent brittle fracture cannot be guaranteed.

A case study showing the need for DWTT on small diameter pipe

Several years ago within Australasia, a short DN350 gas pipeline was required where the quantity of pipe was insufficient to warrant a special order and hence it was necessary to use pipe from a stockist. Very prudently, the designer chose to conduct DWTT tests on PSL 2 pipe (a designation in API 5L that has specified fracture toughness properties) available from stockists. The pipe was DN350, 12.7 mm wall thickness, certified PSL 2 and had Charpy values of 90, 112 and 130 J at 0 degrees Celsius. These values would lead one to expect good resistance to brittle fracture, however, when DWTTs were performed the results were 0 per cent shear at -10 degrees Celcius and 10 per cent and 90 per cent shear at the design minimum temperature of 0 degrees Celcius. The DWTT test pieces after fracture are shown in Figure 5.

The pipe, which looked at face value to be fit for purpose based on its certification as PSL 2 and with its excellent Charpy properties, did not comply with AS 2885.1 and constituted a serious risk of propagating brittle fracture. On this basis it could not be used and other means had to be sought to control brittle fracture in that pipeline.

At the time, a suggestion was made that the material could have been safely used if some crack arresters had been incorporated. As illustrated previously in this paper and shown in Figure 4, practical designs of crack arresters do not work for brittle fracture where there is negligible plastic deformation and no significant crack opening that can be contained and prevented by the crack arresters. Moreover, since the pipeline was to be located in a high consequence area, it was required to meet the no-rupture and limited release rate provisions of AS 2885.1, and this would also have ruled out the use of a design where brittle fracture control was based on the use of crack arresters.

Another cautionary tale

Very recently, a company operating within Australasia placed an order for a large quantity of small diameter pipe from a developing country. Although the specification included a requirement for DWTT, the supplier excluded that requirement on the grounds that API does not require this test and the manufacturer does not therefore provide for it. Given that this situation only became apparent after placement of the order, the purchaser agreed to the exception and began to consider options for brittle fracture control based on methods other than DWTT.

One such option is the development of a correlation between DWTT and Charpy shear fracture appearance. There is evidence, however, that the empirical correlation method presented by AGA [8]
no longer works on modern materials [5, 9]. The transition temperature shift and thickness at which the Charpy V-Notch and DWTT lead to equal 85 per cent shear area are material dependent. Moreover, if the correlation didn’t turn out to work in this particular case, the purchaser would have been faced with a large quantity of delivered pipe not fit for purpose and no redress against the supplier.

Very fortunately, however, the manufacturer came independently to understand that the pipeline made from material without the AS 2885 mandated DWTTs could not have been licensed in Australia because AS 2885 is accepted by regulation as a single and sufficient set of requirements for pipeline safety, and so, to avoid that outcome, the manufacturer volunteered to undertake the DWTT testing and to meet the 85 per cent shear fracture appearance requirement. This is another demonstration that DWTTs can be successfully performed on small diameter pipe, and was a happy outcome to that cautionary tale.

The results of the Energy Pipelines CRC’s research

Energy Pipelines CRC research was conducted on X42 and X70 pipe with DN150-DN400 diameter and 4.8-12.7 mm wall thickness.

Both flattened and gull-winged specimens were tested to ascertain the effect of the plastic deformation involved in flattening on the test results. The specimen and the gull-wing shape are shown in Figure 6.

The uniformity of the gull-wing shape was not as good in the DN150 pipe as the DN300. Nevertheless, satisfactory DWTTs were still able to be performed at DN150, thus indicating that the AS 2885 requirement of doing DWTT down to DN300 is eminently feasible and indeed could be lowered.

This is an important conclusion that overcomes the objections of the ISO/API committee that there was not sufficient evidence that the test could be satisfactorily performed at these small diameters, even down to DN150. Another important conclusion was that small amounts of buckling are not an impediment to obtaining valid results. This is extremely important because currently API RP 5L3 requires that if buckling occurs the results are not valid and replacement tests shall be conducted. Interpreted literally, this requirement would make almost all DWTTs conducted on the upper shelf invalid.

A further valuable outcome was a suggestion that instead of the traditional approach of conducting tests starting at high temperatures and moving progressively lower in order to find the transition temperature, it may be more valuable and practical to start at low temperature and progress upwards as shown in Figure 7. With this approach, especially if the API RP 5L3 no-buckling rule is to be religiously observed – and we expect that it rarely is – then the fracture appearance transition temperature can still be established with considerable reliability.

Further work

The outcomes presented here are based on the experiments on pipe up to 12.7 mm thickness. Further experiments on small diameter pipe with larger wall thickness should be undertaken. Further work is required to better standardise the DWTT method in API RP 5L3 – especially to set rational limits to the amount of buckling and to harmonise the lower limit of diameter to which DWTT should be applied – so as to remove inconsistencies between national standards operating in various jurisdictions and to ensure the requirements are technically robust.

There is also a need to find a way of guarding against brittle fracture propagation in cases where it is impossible for one reason or another to conduct valid DWTTs. Present correlations between DWTT and Charpy shear fracture appearance developed on old steels are no longer reliable. This is especially the case on small diameter pipe where, because of curvature, Charpy tests cannot sample the full wall thickness. This leads to a significant transition temperature difference between the Charpy tests and the full wall DWTTs.

There is also a need to underpin the current findings of DWTTs on small diameter pipe with some full scale tests along the lines of West Jefferson tests.

References
1. Cautionary tale, accessed June 2013 . 2. Hoffmann, H. Struwwelpeter Figure 1 taken from cited Wikipedia web page. English translation published by Dover Publications New York 1995 3. Report of Royal Commission into the Failure of Kings Bridge A.C. Brooks, Government Printer Melbourne Victoria 1963. 4. Fearnehough, G. D., Fracture Propagation Control in Gas Pipelines: A Survey of Relevant Studies, International Journal of Pressure Vessels and Piping, Vol. 2, No. 4, October 1974, p. 257-282. also JIG Symposium on Crack Propagation in Pipelines, Newcastle, UK, March 1974. 5. AS 2885.1-2012 Pipelines – Gas and Liquid Petroleum Part 1 Standards Australia. 6. Cosham, A. et al “Don’t drop the drop weight tear test” J. Pipeline Eng. 2nd Q, 2010. 7. Wu, Y et al, “Drop Weight Tear Tests on Small Diameter X70 Line-Pipe” Joint Technical Meeting of APIA, EPRG, and PRCI May 2013, Sydney. 8. R. J. Eiber and B. N. Leis, “Fracture control technology for natural gas pipelines” circa 2001, Report No. PR-003-00108 to PRCI,” 9. John Piper personal communication. 10. Anon., “Rupture Will Not Delay Transwestern’s Line”, Oil & Gas Journal, Volume 58, No. 17, April 25, 1960.
Acknowledgements
This work was funded by the Energy Pipelines CRC, supported through the Australian Government’s Cooperative Research Centres Program. The funding and in-kind support from the APIA RSC is gratefully acknowledged, as is the contribution of Drs Cheng Lu and Guillaume Michal. Valuable comment was also provided by Bob Andrews, Andrew Cosham, and Brian Rothwell.

Send this to a friend