Old...but how old?

TWI Bulletin, January - February 2010

Life extension using fitness for service assessment

Dr Philippa Moore is a senior project leader in Fracture Integrity Management within the Structural Integrity Technology Group at TWI. Philippa runs projects involving fitness-for-service assessment, fracture mechanics testing, and welding engineering, particularly for the oil and gas industry. She is TWI's Institution Representative within the MARSTRUCT Network of Excellence in Marine Structures.

Steve Smith is a project leader in Fracture Integrity Management; part of TWI's Structural Integrity Technology Group. He joined TWI in 2008 after completing his Masters at Imperial College London. His activities include mechanical testing, fracture assessment and determination of flaw acceptance criteria for welded structures, with emphasis on the oil and gas sector. He is part of the Welding and Joining Society Younger Members Committee and is an associate of the Royal School of Mines.

Using established equipment beyond its design life is an increasingly popular alternative to costly replacement. It all depends on the accuracy of your fitness-for-service assessment.

Many industries need to operate equipment beyond their original design life, to repair or re-use equipment, or to assess their remaining life. Operators of pressure equipment in the downstream oil and gas, refineries and chemical industries have traditionally removed and repaired the ageing damage they found. However, as Philippa Moore and Steve Smith explain, a fitness-for-service (FFS) assessment carried out in accordance with a recognised procedure such as BS7910 (2) is a viable and potentially beneficial alternative. TWI already has a good and established reputation for FFS assessment for the upstream oil and gas industry, such as fabrication of pipelines, and life extension of offshore structures.

Using three case studies based on TWI's experience, this article demonstrates what FFS assessment can achieve, even when limited data is available, and will give confidence to those new to FFS methods to apply the principle to ageing equipment in a wider range of industries.

As well as showing what can be achieved with only limited (or historical) data, it also demonstrates how much further the methods can be taken when high quality mechanical test data can be provided. Whether through equipment re-validation, or approving the structure's remaining life, FFS assessments can be a beneficial alternative to the scrapping of equipment.

Fitness-for-service assessment of propane storage sphere

A propane storage sphere (Fig.1) was installed as part of an expansion project in Saudi Arabia. During a scheduled outage, visual inspection revealed a defect in one of the sphere's butt welds that had not been detected during fabrication, in the form of a 5mm deep and 20mm long lack of side wall fusion along the weld fusion line at the heat affected zone (HAZ). Before the vessel could be allowed to go back into service, its defect tolerance had to be demonstrated so as to ensure the sphere would not fail by brittle fracture or plastic collapse.

The minimum Charpy impact energy measured at the design temperature of -45°C was 10J, from weld specimens extracted during the original installation of the sphere. Charpy impact values can, if required, be used to estimate a conservative level of fracture toughness in terms of a stress intensity factor, K_mat using Annex J of BS7910. In this case, 10J Charpy gives a conservative estimate of K_mat of 34.7MPa√m at -45°C.

The sphere was pressurised to give a membrane stress of 160MPa, and an initial assessment using the estimated fracture toughness, along with the specified minimum yield strength (SMYS) and tensile strength (SMTS) for the ASTM A537 steel (345 and 485MPa respectively), gave an assessment point for the defect that was outside the 'acceptable' area on the failure assessment diagram (FAD), as illustrated in Figure 2. An FAD shows the proximity to brittle fracture on the Y-axis and proximity to plastic collapse on the X-axis, with a failure-assessment line plotting the interaction of the two failure modes.

An assessment point that lies inside the line is predicted to be safe from brittle or ductile failure, and is tolerable. An assessment point outside the curve means that the flaw cannot be shown to be tolerable. The over-conservatism of the assessment using only the barest estimates of mechanical properties means that the storage sphere would require weld repairs, along with significant delays and costs resulting from that repair.

However, when an FFS assessment was performed using data obtained from a fracture mechanics test programme, consisting of measuring tensile properties and fracture toughness values which were representative of the sphere butt welds, the assessment point for the defect fell within the failure assessment line (Fig.2). Fracture mechanics tests at -45°C measured a fracture toughness K_mat of 160MPa√m, much higher than the value predicted by the Charpy estimation. The results of tensile tests were also higher than the minimum specifications, as expected, giving YS and UTS of 398 and 580MPa respectively. The assessment was able to show that the storage sphere was fit-for-service sphere despite the presence of the lack of fusion flaw. Fortunately, in this case weld material was made available and carrying out a small programme of mechanical testing allowed unnecessary repairs to be avoided.

Comparing the two assessment points in Figure 2 shows just what a powerful effect on the FFS assessment representative materials data can be when there is a necessity to illustrate the defect-tolerance of structures, whether already in service or still to be commissioned. The assessment using Charpy data is over-conservative by a margin of 1.8 relative to the assessment using fracture toughness data. Other work has also shown that compared to assessments that use fracture mechanics data, Charpy estimates give over-conservative 'safety' margins of 1.75 to 3.5. In some cases, however, this is nonetheless sufficient to demonstrate fitness-for-service in a simple manner.

Life extension of internally corroded flow lines

A series of hydrocarbon flow lines was fabricated back in the 1970s to serve as part of an oil and gas platform's separator manifold. During fabrication of the lines, non-destructive examination procedures were put in place to ensure that weld defects outside those permitted by the weld flaw acceptance criteria were removed. These flow lines were given a design life of around 30 years based on the level of carbon dioxide corrosion which was considered to occur during service. Over time corrosion rates were established and the remaining wall thickness of the lines was measured at five year intervals in order to monitor safely and maintain the integrity of the manifold. As the design life of the lines drew closer, the platform operator required that a case be put together which would validate extending the service life of the manifold, involving a FFS assessment to justify safe life-extension.

The worst case fabrication defect that could have been present in the welds at installation was considered to be a circumferentially orientated external surface defect which was 10mm long and 3mm deep. At the occasion when the FFS assessment was performed, the internal corrosion had reduced the wall thickness of the lines to 12mm from 20mm (Fig.3). Initially, all the information the operator had at their disposal was a series of Charpy impact values measured at the design temperature of the lines (-10°C). In order to be conservative, the lowest of the impact energies, in this case 5J, had to be used to estimate a value of toughness, K_mat, of 27.4MPa√m. The specified minimum tensile properties for ASTM A106 Grade B pipe material of 240MPa yield and 415MPa UTS were also used at this stage. The generic FAD shown in Figure 4 shows that with a reduced wall thickness of 12mm, this worst case weld defect could not be shown to be tolerable as it falls just outside the failure-assessment line.

Fig.3. An example of the wall thinning resulting from internal corrosion of the flow lines, both

Subsequent fracture toughness and tensile tests were therefore recommended to gain more representative material properties. Some pipe material was removed from service and the test programme was kept to a minimum to save time and money. The full stress-strain curve obtained through the tensile testing measured a yield strength of 306MPa and was used to derive a material-specific FAD which accounts for work hardening effects in the material. The measured CTOD value of 0.42mm (equivalent to K_mat of 163MPa√m) was used as a replacement to the lower bound estimated fracture toughness.

The effect of even a modest test programme can be clearly seen in Figure 5 where the assessment points at wall thicknesses 12, 10, 8, 6 and 4mm are all shown to be able to tolerate a 3mm by 10mm surface flaw. This study showed that a significantly reduced wall thickness would be acceptable when considering the structure's ability to tolerate the worst case weld defect permitted during fabrication. As a result, the proposed case for life extension of the flow lines was accepted on the basis of their FFS and the manifold remained in service past its initial design life without risk of failure.

Failure investigation of a heat exchanger

In December 2001, a leak from the shell of a heat exchanger was detected after start-up following an unplanned outage of an ammonia plant see Figures 6 and 7. Preliminary metallographic investigation of the failed heat exchanger revealed that the failure mode was brittle fracture resulting from an internal surface-breaking defect. A third-party review was required to investigate the reason for failure of the vessel.

The heat exchanger was subjected to pressure and temperature changes in service. Instrument traces recorded during start-up were supplied in order to determine the environmental conditions at the point of failure, to perform an FFS assessment. From this data, it was determined that the temperature of the heat exchanger had been 50°C during failure at start-up. Prior to installation of the vessel, Charpy specimens had been extracted from spare fabrication material and tested at ambient temperature to ensure that the material was in compliance with DIN 17155 which requires minimum impact energy of 31J at 20°C.

The specimens extracted from this fabrication material gave a minimum Charpy value of 95J at 20°C and thus the material was in compliance. For purposes of the assessment, this Charpy value was used to calculate an estimated toughness at the approximate temperature of failure. The resulting estimated toughness at 50°C was 178MPa√m and sectioning of the vessel's fracture surface revealed the dimensions of the internal defect immediately before failure was 25mm long and 6mm high. Tensile properties (277MPa SMYS and 536MPa SMTS) were taken from DIN 17155 for 15Mo3. The assessment performed using this data predicted that the flaw would have been safe from brittle fracture.

It was immediately apparent that if all these assumptions were correct, the heat exchanger should not have failed, which raised questions about the accuracy of the input data used in the assessment to explain the failure. Finite element analysis confirmed that the stresses assumed to act on the vessel during start-up were reasonably accurate. Further materials testing was conducted to account for potential errors in the materials data.

Charpy specimens and fracture toughness test specimens were machined from the heat exchanger itself, close to the location of failure. These were tested at 50°C to determine the fracture toughness of the material which had seen service rather than that which was in the 'as-received' state. The lowest Charpy value obtained from these tests was 14J, which was much lower than the 95J measured from Charpy tests at fabrication, and indicated that embrittlement during service had occurred. Chemical analysis showed that the levels of aluminium and nitrogen found in samples of the heat exchanger were sufficient to make the material susceptible to strain ageing embrittlement. Fracture toughness tests measured CTOD values as low as 0.1mm. The assessment was re-performed with this new data and showed that the assessed defect, after the given service conditions, was sufficient to lead to failure by brittle fracture, as illustrated in Figure 8.

This failure investigation shows the importance of using good quality representative materials data when performing a FFS assessment, and its benefit for understanding failures. If the initial Charpy specimens had been strained and aged to simulate the effects of service then the operators may have been aware of the potential embrittlement mechanisms, and allowed better engineering of the heat exchanger and to avoid failure in this instance. Using FFS assessment to understand this failure helped the operators to prevent similar failures in other potentially susceptible equipment.

Conclusions

The case studies presented in this article show the flexibility of the fitness-for-service assessment methods given in BS7910, not just to assess the significance of flaws and defects found during service, but also for avoiding unnecessary repairs and to justify life extension. The methods can also be used to help understand failures in order to prevent similar mechanisms affecting other equipment in service.

The methods can be applied using literature data, specification properties or estimates of fracture toughness, in the absence of any other data, although the results risk being over-conservative by a significant factor. When actual materials properties data are used in the assessment, the real power of FFS methods can be exploited.