Session: DA-12-02 Fracture 2-Fracture Prediction and Evaluation

Paper Number: 133078

133078 - Low Temperature Integrity of Carbon Steel Pressure Piping: An Assessment Based on Probabilistic Fracture Mechanics and Full-Scale Testing 

Good engineering practice usually requires that carbon steel pressure piping for low-temperature use should be Charpy-tested at the service temperature. For example, an A333 Grade 6 material meeting a Charpy requirement of  18J at -46C can be safely employed in the as-welded condition at a temperature down to -46C. Austenitic stainless steels are usually employed at colder temperatures. Nevertheless, there is some evidence that operation of carbon steels at temperatures below the Charpy test temperature can be shown to be safe, particularly for thinner-section piping and lower applied stresses. This work explores that possibility, using a mixture of mechanical testing and Engineering Critical Assessment (ECA), including the use of Probabilistic Fracture Mechanics (PFM) to model the variations in materials properties. 
The experimental part of the program consisted of small-and large-scale testing of 12.7mm (nominally 0.5 in) thick girth welds made in a seamless pressure piping material compliant with ASTM A106/API X42. Welds were made in the laboratory using a procedure and consumables representing normal industrial practice. The welds were qualified to ASME B31.3, and also subjected to additional tests such as low-temperature fracture mechanics testing (critical J/CTOD) in order to determine the statistical distribution of fracture toughness using the Master Curve concept (now incorporated into the UK flaw assessment procedure BS 7910).
Three full-scale tests were then carried out on girth welds containing a large artificial flaw. The concept was to test the pipes under tension at a temperature somewhat below the Charpy transition temperature, T27J, and to record the applied stress associated with a brittle fracture event. Although the materials had been selected to be as close as possible to the specified minimum Charpy requirements (T27J=-10C), the actual Charpy properties of both the parent pipe and the weld metal exceeded the requirements by a large margin, with T27J=-72C achieved in the parent metal and T27J=-123C in the weld metal.  Consequently, in order to achieve the goal of triggering brittle fracture in the weld metal, the full-scale tests needed to be carried out at temperatures between  -120 and -160C. In spite of these very low test temperatures, it proved extremely difficult to induce brittle fracture in the large-scale tests, one of which failed by necking remote from the crack plane and another from an undetected fatigue crack in the parent metal. Even so, the headline indications are that that the welded joints, each containing a large fatigue crack, could be loaded to 100% of SMYS at T27J, to 78% of SMYS at T27J-17C and to 43% of SMYS at T27J-37C, all of which suggest that the safe working temperature for carbon steels could be extended to below T27J under the appropriate circumstances. 
Results of the full-scale tests were interpreted in terms of both deterministic and probabilistic ECA, based on the UK flaw assessment procedure BS 7910 (‘Guide to methods for assessing the acceptability of flaws in metallic structures’). These assessments suggested a very high probability of failure of the full-scale tests, sometimes at odds with the observed results. Part of the discrepancy between the ECA and the observed behaviour of the test specimens is likely to be due to the fact that only fracture toughness and tensile properties were treated probabilistically in this work, with other variables treated as deterministic. In practice, whilst some inputs (for example, flaw size and applied stress at failure) can be justifiably treated as single values, welding residual stress is both very relevant to brittle fracture and highly variable. Because of the paucity of data specific to the type of girth welds tested, a probabilistic interpretation of welding residual stress was not possible.  The characteristic upper bound residual stress profiles given in Annex Q of BS 7910 were used instead, leading to an overestimate of the probability of failure. 

Presenting Author: Geoff Evans BP

Presenting Author Biography: Geoff is a mechanical engineer, with over thirty years’ experience in the upstream, refining and chemical industry. He has worked on a wide range of mechanical design and construction problems in Europe, U.S.A., Middle East, Africa and the Far East. Experience includes detailed stress analysis of pressure vessels, tanks, industrial noise and vibration control of pipework systems using both British Standard and ASME (American Society of Mechanical Engineers) design codes. These projects have involved both the design of new plants and the assessment of existing equipment to establish suitability for continued service, using codes including BS7910 and API 579.

He is responsible for the BP Engineering Technical Practice for pressure systems and represents BP on ASME, ISO and EEMUA piping committee. He is also a member of a Joint Industry Committee responsible for developing guidelines for the assessment of piping vibration and fatigue.

Authors:

Isabel Hadley TWI Ltd
Siak Manteghi BP
Geoff Evans BP
Matthew Haslett TWI Ltd
Yin Jin Janin TWI Ltd