How much longer will it last? The assessment of high temperature plant

TWI Bulletin, March/April 1995

Keith Bell read Metallurgy and Material Science at Birmingham University and went on to do postgraduate research on fatigue crack growth in nickel base superalloys. He then worked for five years with a contract research organisation specialising in high temperature materials and remaining life assessment, before joining TWI's Fracture Department in 1988. He is now a Senior Project Engineer in the Structural Integrity Department, where his work is concerned largely with the assessment of defects in structures, at both high and low temperatures.

Components which are stressed at high temperatures do not last for ever. Keith Bell describes some of the ways to determine the remaining life of plant in service.

Most design rules for process plant, such as pressure vessels and piping which are running at high temperature (500-600°C) and pressure, are based on an assumed service life of 100,000hr (although BS 5500 is a little more flexible in that lives between 100,000 and 250,000 can be selected). These rules typically use average creep stress rupture data for the parent material. Safety factors are then applied to define the allowable stress which compensates for both the variability in creep properties that is found for a given material and the presence of welds.

If the plant is then run at the allowable design stress, failure should not occur within the design life. Codes, therefore, nominally define a conservative usable life and the actual available life may not be expended in service. There are considerable economic benefits in operating plant beyond the design life, but this is not the only advantage that can be gained from a remaining life assessment:

• costly, unplanned breakdowns can be avoided;
• replacement of components can be scheduled to fit in with manufacturer's lead times;
• safety and insurance regulations can be satisfied;
• variations in operating conditions can be accounted for and new operating limits can be defined to match any required minimum desired life extension.

Remaining Life Assessment (RLA), a technology largely originated in the power generation industry, is now being widely used on many types of high temperature plant. The down-time needed for a full RLA on-site inspection and the specialist techniques of analysis used, can result in the assessment costs being high. However, it is possible to break an RLA down into several phases with increasing cost being associated with increasing sophistication and accuracy.

If a basic level RLA shows that there is a very long remaining life then further expenditure of time and money is unnecessary at this stage. The more accurate, more costly evaluations of the creep life fraction consumed can be postponed to a late date, with expenditure and effort phased to match each component as it becomes more critical. A component should be considered for RLA at some time in its service life if its failure would lead to general safety problems; long breakdowns or unavailability of the plant; or high costs of repair/replacement.

Phased approach

Preliminary investigation

The first stage of RLA involves a calculational exercise using any information on the history of the component. All available operational data ( eg process fluid and metal temperatures, inlet and outlet pressures, durations of periods of service, external and internal chemical environments) are collected. These are then compared with the design data (temperature, pressure, dimensions) and standard materials properties information (times to rupture, as given by ISO for example). ^[1] If, for instance, the plant has been running about 10°C lower than the design temperature, this can be sufficient to double the expected rupture life (the corollary being that overtemperature running dramatically reduces the available life).

At this stage one would also consider any other pertinent data (results from visual and routine NDT inspections, or any history of repairs to the plant). A further elaboration in the preliminary RLA may be to make use of the enhanced numerical analysis techniques that have become available since most of the plant currently in operation was designed. Such structural and thermal analyses may be useful in defining critical locations for more detailed examinations on-site.

Detailed non-destructive examination

The routine techniques of ultrasonics and magnetic particle inspection should continue to be used, together with dimensional measurements (changes in wall thickness and the diameter of vessels and pipes can be used to monitor strain accumulation and corrosion/oxidation effects), and some of the more advanced developments that have been used to detect and measure the early stages of damage accumulation. It is now generally recognised that in welded structures it is either the welds or heat affected zones which are going to be the life limiting features of most components. The analyses performed in the preliminary phase will have identified where welds and critical locations coincide and these areas should be subject to more detailed investigation.

Considerable research effort has been devoted to finding a non-destructive technique for characterising the degree of degradation produced by high temperature service and relating this to the remaining life of the weldments in the component. Although progress has been made in the use of specialised ultrasonics, eddy currents, resistivity and in situ hardness measurements, the main methods for damage characterisation are based on in situ metallography. In the low alloy and Cr-Mo steels that are largely used for the fabrication of high temperature plant in the power generation and petrochemical industries, the consumption of the creep life and the degradation in properties are accompanied by physical and microstructural changes that can be measured. The main methods are surface replication (to check for creep cavitation at grain boundaries and, in the later stages of life, microcracking, to measure the extent of carbide coarsening, dissolution, and spheroidisation), extraction replicas (for microanalysis of secondary phases - the carbide compensation changes with exposure) and microsampling (for more extensive chemical analysis). Details of these techniques are given by various authors in the literature. ^[2,3]

Post-exposure testing

This involves removal of sufficient material for the fabrication of test specimens and even with the various techniques of specimen miniaturisation that have been developed, a repair will be necessary before the component goes back into service. The advantages of post-exposure creep and rupture testing on the actual material are mainly in the refinement of the input data in the calculations, ie firstly, the safety factor due to large variability in measured creep properties can be eliminated (it is typical for the scatterband in stress rupture results to be ± 20% of the line used to define the allowable stress), and secondly, the effects of uncertainties in the prior service can be overcome. The disadvantages, apart from the need to repair and the higher costs of such specialised testing, are that it is not always possible to extract the material from the most critical regions ( eg welds at complex joints), and the time needed to obtain the test data, (even with acceleration of the tests by the application of temperatures or stresses higher than those found in-service).

Case studies

Moth-balled naphtha plant

TWI was approached by a Research Member to give an independent assessment of major components in a refinery that they were considering for purchase. The plant had been built in the 1960s to the ASME codes and operated for over 100,000hr in various campaigns during a 14 year life. Economic circumstances then led to the plant being shut down for several years. The plant was refurbished and run for a further period (less than one year) before finances again led to it being shut down. TWI staff visited the plant to collate the fabrication, operation and maintenance information that was available in the plant records. The site visit included visual inspection of various items of plant that were considered critical to the viability of the refinery.

From the temperature and pressure records available for the reactor pressure vessels, piping and furnaces, it was possible to build up a reasonable picture of the service exposure seen by these components. The fabrication drawings revealed the materials used, and the design operating conditions. Ultrasonic inspection records showed the vessel wall thicknesses (even following the corrosion loss due to 100,000hr service) still exceeded the specifications. It was then possible with a preliminary calculational RLA, to show that only a very small fraction of the available life had been consumed in all of the pressure vessels and most of the tubing.

One furnace was found to have tubes which had been running at temperatures in excess of the design specification. The cumulative life fraction calculation for this tubing was large compared to the other components assessed, indeed, it exceeded unity, indicating that failure would have occurred if the material had been 'lower bound' in its creep resistance. Visual inspection of this furnace revealed severe sagging and bulging of the lower rows of tubes, ie considerable creep strain accumulation. Metallographic examination during the refurbishment had revealed extensive creep cavitation damage. It was recommended that this furnace should be at least partially retubed before it entered further service.

With the exception of the one furnace described above (where problems with the layout of the burners had led to excessive service temperatures), it was thus possible, as a result of the RLA (which took less than one week), for the prospective buyers to gain considerable confidence that there was a worthwhile operating future for the plant.

Ethylene cracking furnace coils

Cracking problems had been experienced during the repair of some cast heat resistant furnace tubes which had been in service for several years. Some of these furnace tubes had been removed during the repair work, and this material was available for post-exposure dimensional and metallographic examination. A preliminary calculational approach was applied, with the limited operating temperature and pressure data available. This showed that even at the maximum operating temperature, the creep stress rupture life remaining was as great as the service life already experienced and the entire furnace did not require immediate retubing. The metallographic and dimensional data confirmed the level of creep life fraction consumed. It was considered that the life limiting factor in service was likely to be the rate at which oxidation and carburisation were reducing the effective wall thickness. A retubing schedule was developed on the basis of an estimated rate of microstructural change determined from the metallography. No post-exposure creep testing was required.

Some 2.25%Cr-1%Mo furnace tubes had been removed from service during a modification of the furnace layout after prolonged service (more than the design life of 100,000hr).

The calculational approach suggested that the tubes were approaching the end of their available lives, based upon the standard creep rupture data available. These tubes were made available to TWI for RLA by post-exposure creep testing. The wall thickness of the tubes was insufficient for the fabrication of conventional creep specimens and rods of test material were electron beam welded on to sections of matching composition for the larger diameter threaded ends to the specimens (see Fig. 6). A small-scale programme of accelerated stress rupture tests was performed and it was found that under the design conditions of temperature and stress there was still a very significant creep life remaining and retubing could be postponed for several years.

Some other examples of case studies in high temperature RLA on petrochemical plant are available. ^[4] The procedures for high temperature defect assessment are discussed in the conference proceedings. ^[4] These procedures have been formalised recently with the issue of a British Standard Published Document PD6539: 1994 ^[5] and work to create a piece of software to assist in calculating the significance of defects in high temperature plant is in progress at TWI.

References