Structural Integrity Technical Group Meeting

Recent Developments in Structural Integrity Methods

Notes of the meeting held on 10 December 2009 at TWI at Granta Park, Great Abington, Cambridge.

Welcome and Introduction

The chairman, Dr Bob Ainsworth (British Energy) welcomed everyone to the meeting and gave a brief introduction. He outlined the programme for the day and directed proceedings for the duration of the meeting. A summary of the presentations is given below and the overheads used by the speakers are attached to these notes where they have been provided.

An overview of BS7910 developments

Dr Henryk Pisarski of TWI began the day with an overview of the need for the development of fitness-for-service (FFS) assessment procedures, the history of BS790 developments and its future developments. In respect of the history of BS7910 developments, the current procedure BS7910:2005 plus Amendment 1 is dated to its earliest predecessor, PD6493:1980, a CTOD design curve based FFS procedure developed in response to the need for FFS assessment of pressure vessels. Between them were PD6493:1991 (with the most pronounced feature being the replacement of the CTOD design curve approach with the R6 failure assessment diagram (FAD) approach), and BS7910:1999 plus Amendment 1 which incorporated new contents including creep, parent metal/weld metal mismatch effects, and libraries of stress intensity factor, reference stress and residual stress profiles. It is envisaged that a new version of BS7910 will be published in 2012, incorporating substantial parts of the European FITNET FFS procedure. Dr Pisarski then made a comparison between BS7910 and FITNET assessment procedures, with an example to illustrate the discrepancies between the two procedures and the trend of obtaining a less conservative assessment from Option 1 (single-point value of fracture toughness analysis) to Option 5 (constraint-based analysis) in the FITNET procedure. He went further to give an introduction of the strain-based assessment approach which was proposed primarily for the assessment of offshore pipelines in which plastic straining is often experienced. Dr Pisarski concluded his presentation by stating that BS7910 is continuous development to provide a general framework for the assessment of flaws to ensure safe structures that are fit-for-purpose and meet the needs of industry.

Recent developments when considering combined primary and secondary stresses in fracture

Following Dr Pisarski's presentation, Mr Peter James of Serco gave a talk on a project aiming to provide deeper understanding, and hence an improved technique, for the treatment of combined primary and secondary stresses. A simplified method was proposed to estimate combined crack driving force by introducing an equivalent flaw size ã (relating SIFs to reference stress) and a g() function which defines the behaviour of secondary stress with plasticity. The method was demonstrated to provide an improved estimation to the R6 method and show a significant decrease in conservatism. The presentation also indicated that all methods are equally applicable to both thermally induced (long range) and residual (local) secondary stresses. The g() function based on an Option 2 FAD was shown to be in the best agreement with the FE results under gross plasticity, with support from examining a circumferentially cracked cylinder subjected to thermally induced bending stress and weld residual stress.

Modification of the failure assessment diagram for non-sharp defects

Mr Anthony Horn of Corus focused his presentation on the development of a new engineering assessment methodology for steel components containing non-sharp defects to avoid the over-conservatism incurred by application of the assessment procedures which treat defects as a sharp-crack. The Weibull Stress approach was adopted to calculate the probability of cleavage fracture based on the results of finite element analysis. The constraint-based FAD approach in FITNET/R6 has been extended to account for non-sharp defects, with modified FADs. Four steps of modifying a sharp-crack FAD for a non-sharp defect were proposed, in sequence by correcting the J_e (elastic J-integral) of a crack to that of a notch, correcting the plastic limit load of a crack to that of a notch, correcting the K_mat of a crack to an effective K_mat for a notch, and modifying failure locus.

Fracture based structural integrity in welded structures at elevated temperatures

Professor Kamran Nikbin of Imperial College started his presentation by briefly going through the background to assessment codes relevant to elevated temperature applications, from ASME and RCCM in 1950s and 1960s to BS7910:2005 and R5 in 2007. The fundamental parameters and criteria of characterising creep crack initiation (in practice, an initiation crack size can range from between 0.05 to >1mm) and crack growth were interpreted and commented. The three methods of determining C* (a parameter used for characterising creep crack growth) were described, respectively in relation to experimental determination, estimation from reference stresses, and using uniaxial creep data. Examples of creep crack growth testing (both in parent metal and HAZ material) and data processing for correlation were also given, in association with pipes and notched bars. Finally, recommendations were highlighted, such as 'standardise testing methods/geometries' and 'standardise the K/_ref/C* solutions for generic components'.

Strain based ECA tools to determine the strain capacity of girth welded pipes

In his presentation, Dr Afshin Motarjemi of DNV first gave a brief introduction about DNV. The need of strain-based design was then addressed, largely in relation to the increasing use of high strength steel and clad pipes in deep water exploitation of oil and gas. He then demonstrated that the tensile strain capacity is one of the limit states in the strain-based design of pipelines. The methods used in strain-based design were outlined, including the extension of stress-based design procedures for strain-based design with the guidance provided in DNV OS-F101 and DNV RP-F108, whose main features are the use of SENT specimens for installation phase and the use of SENB specimens for the operational phase; the use of the crack driving force approach to relate between the crack driving force and the remote longitudinal strain, taking into account defect size, defect location, material tensile properties, weld strength mismatch, and weld geometry; and the use of curved wide plate testing and full-scale testing of pressurised pipes for establishing the tensile strain capacity for strain-based design.

Development of strain-based failure assessment diagrams

Dr Bob Ainsworth of British Energy went through the background, definitions and shape of a strain-based FAD, treatment of secondary strains, validation and R6 development. A strain-based FAD is defined with K_r as a function of D_r (reference strain ratio), namely K_r = f*(D_r). The function f* was derived in terms of the equivalence of J-estimation methods and FADs. Three hierarchies of strain-based FADs were outlined: Option 1 (an analogy to stress-based Level 2A (generalised FAD), Option 2 (an analogy to stress-based Level 2B (material-specific curve) and Option 3 (directly from finite element J elastic and elastic-plastic solutions). Treatment of secondary stresses within the strain-based FADs was also addressed, with the methodology for deriving V (a factor used to quantify the plasticity interaction effect between primary and secondary stresses in the stress-based FAD approach) being adapted to the strain-based FAD. Some numerical examples (surface cracks in plate and semi-elliptical cracks in cylinders) have been used to validate the effectiveness of the derived strain-based FADs. A new R6 Section III.16 has been drafted to reflect the developments of a strain-based FAD which is currently under peer review by TAGSI. Dr Ainsworth summarised his talk by highlighting that stress-based FADs can be modified to derive strain-based FADs with a hierarchy of options and the methods will be included in R6 in future when the external peer review is complete.

Reliability analysis of defect-containing structures using BS7910 partial safety factors

Dr Liwu Wei of TWI made a presentation on the use of BS7910 partial safety factors (PSFs) for reliability analysis of structures containing defects. The need of PSFs and the derivation procedures for PSFs were broadly introduced. The PSFs recommended by BS7910 and API579-1/ASME FFS-1 were compared, highlighting that despite the similarities in derivation of PSFs between these two procedures, it is difficult to make general comparison due to some different probabilities of failure (P_f) employed in the two procedures . For a specific case corresponding to a P_f of 10^-3, larger PSFs are generally proposed in BS7910, implying that BS7910 could be more conservative than API579. The probabilistic fracture mechanics analysis of a through-thickness crack and a surface semi-elliptical crack in a plate was presented to illustrate the likelihood of causing some over-conservatism when using the recommended PSFs in BS7910; a target P_f value from BS 7910 could be four orders of magnitude higher than that from a PFM based reliability analysis in the plastic collapse region. Dr Wei also demonstrated that in a general trend for each of the four PSFs recommended in BS7910, the failure probability decreased as the assessments changed from the elastic fracture region to the plastic collapse region.

Modelling of Welding residual stresses

The final presentation of the day was given by Professor John Bouchard of The Open University on modelling of welding residual stresses. He went through the fundamental steps, from loadings, boundary conditions, properties, idealisation, constitutive model, to validation with experimental measurements. This was demonstrated with a NETe round robin project of modelling residual stresses in a bead-on-plate weldment. The weld residual stress modelling guidelines published R6 Revision 4 (Section III.15: Calculation of Residual Stresses in Weldments) were browsed through. Nine steps of calculating residual stresses were presented, with a highlight of the four most important steps, which are 'define analysis objectives', 'collect analysis input data', 'decide modeling approach', and 'analysis validation'. Professor Bouchard raised cautions about all isolated results claiming to characterise the state of residual stress in a weldment. The presentation was concluded with a list of guidance documents on modelling of welding residual stresses.

Closing Remarks

After a general discussion on the topics discussed, the chairman thanked the speakers and all the attendees before bringing the meeting to a close.

Future Meeting

The next meeting of the technical group is likely to be held in the autumn of 2010 with a hot topic relating to structural integrity. Further information will be circulated closer to the time.