J Draper, Safe Technology; D Green & D Worswick, Serco TAS, UK
Engineering components operating hot and cyclically will experience creep damage and fatigue damage which can initiate cracks and ultimately lead to component failure. Fatigue damage was first recorded and investigated at the begining of the railway industry and continues to be extensively studied today.
Creep deformation in ice and rock under geological forces has been noted for centuries. However, creep damage in engineering materials became of significant interest with the advent of power stations since the thermal efficiency of thermodynamic cycles is directly related to the maximum operating temperature.
Creep damage and fatigue damage can occur separately and progressively during plant operation. Fatigue damage is related to the magnitude and number of load cycles whereas creep damage is related to steady load levels and operating temperature. There are commonly occurring circumstances under which fatigue damage and creep damage can interact leading to the sudden onset of cracking and much reduced component endurance.
Current approaches are inadequate
A common approach to managing sudden degradation due to creep-fatigue damage is to limit the maximum operating temperature of components, thus negating creep effects. For example, ASME codes (ASME, NB 3000) limit the operating temperature of certain classes of carbon steels and low alloy steels to 371 °C and to certain austenitic stainless steels and nickel-based alloys to 426 °C.
This approach is not adequate for future needs. Recent changes in the operating rationale of power stations have occurred because of industrial deregulation. This has meant that many power stations have changed from baseload generation to peak load generation so that high temperature components such as gas turbines and boilers are now operating beyond their design basis.
Notwithstanding the long history of creep and fatigue research, the complexity of the phenomenon and industrial changes driven by globalization, deregulation and environmental issues mean that there is an urgent need for validated creep-fatigue endurance methodologies linked to modern methods for stress and strain analysis. This need will only increase in the future. Since the technical problem is now occurring in a wider range of industries, the methods should be robust for use by non-specialists.
The strain based approach
Over the past 40 years, extensive research has been undertaken collaboratively by various companies in the UK power generation sector, concerned with the understanding of thermal-mechanical, creep-fatigue damage mechanisms. Serco Technical and Assurance Services (TAS) which traces its origins back to the UK Atomic Energy Authority has been continually involved with these programmes and has contributed extensively to the research.
The need was driven by incompatibilities between the type of failure observed in laboratory material tests and the failure of plant components, allied to the need for realistic estimation of thermal-mechanical fatigue damage for safe design and operation of nuclear plant components. This led to the development of a component and material specific ‘strain’ based procedure, as an alternative to the more traditional ‘time’ based British Standard and ASME approaches. The UK strain based development is the basis of the R5 assessment procedure used extensively in the UK nuclear power industry1.
An essential element of the strain-based approach is the understanding of damage mechanisms associated with elevated temperature thermal-mechanical, creep-fatigue failure of laboratory material and features tests.
Fatigue concerns the initiation and growth of cracks at free surface. Creep concerns the initiation and coalescence of voids and micro-cracks to form macro-cracks along grain boundaries, and as such affects the bulk of the material (Figure 1). At low strain ranges creep cracking can occur alone, while at higher strain ranges surface fatigue cracks can alone. Alternatively, surface fatigue cracks can trigger the coalescence of creep induced grain boundary micro-damage into grain boundary macro-cracks. Thus thermal-mechanical, creep-fatigue interactions can occur.
A thoroughly modern solution
This understanding of material damage mechanisms at a fundamental level has been reconstituted into fe-safe/TURBOlife analysis algorithms to estimate of component thermal-mechanical, creep-fatigue endurance2. Thus the fe-safe/TURBOlife creep-fatigue damage software recognizes and accounts for aspects such as multi-axial loading, component size and geometric features, cyclic hardening and strain rate effects which are known to influence thermal-mechanical fatigue failure of components.
These aspects may not be adequately accounted for in the more traditional phenomenological approaches which use laboratory tests to characterize the behaviour of uniaxial material test specimens, which are then assumed to transfer directly to component assessment. Any cyclic history can be inputted into the software, either repeated cycles or random loading, resulting in realistic estimates of component damage. Thus the software solution can be used for component design, forensic investigation of component failures, safety case justifications, life extension studies, change of operating regime studies, etc.
The nuclear option
Nuclear boiler components incorporated into the UK’s Advanced Gas Cooled Reactors operate under variable condition due to start-up/shut-down cycles with extended periods of high temperature steady load during normal operation. Such components are high integrity and require an adequate demonstration of margins against creep-fatigue cracking in support of the station safety case and for justification of station life extension. One aspect of demonstrating high integrity concerns the calculation of creep-fatigue damage because of historic operating conditions and anticipated operating conditions in the future.
A series of ten, stainless steel, full penetration, surface dressed cruciform butt welds (Figure 2) were tested in four point bending with fully reversed loading and incorporating hold periods in the fully loaded condition. The specimens were constructed of plate 26 mm thick.
Figure 2: Full penetration, stainless steel, dressed cruciform butt weld typical of high temperature applications in nuclear boilers
The tests were performed at 550 °C, which is in the creep regime, with four different surface strain ranges of ±0.25 per cent, ±0.4 per cent, ±0.6 per cent and ±1.0 per cent, and three different hold times of zero hour, one hour and five hours. For each test the location of crack initiation and endurance to crack initiation was noted.
On the basis of considering fatigue damage only, cracking would be expected at the position of the maximum stress concentration factor for the entire joint. Because the weld surface is dressed, the maximum stress concentration factor did not occur at the weld toe but on the surface of the weld at a distance of some 5 mm from the weld toe.
Of the ten test results, crack initiation for two tests occurred at this position of the maximum stress concentration factor. One failure was at the weld toe and seven failures were in the plate away from the weld toe. Thus the occurrence of creep damage during the hold periods influenced not only the cyclic endurance of the joint but also the cracking location.
Detailed creep-fatigue endurance assessment of the cruciform tests was performed using the fe-safe/TURBOlife software. Aspects of the analysis which are enabled by the use of the software were as follows:
- stress and strain analysis was conducted by elastic finite element, with the necessary plastic strain and creep strain determined the software;
- separate tensile, fatigue and creep properties for the weld and plate material were used;
- cycle by cycle analysis was performed accounting for cyclic hardening which progressively changes the material stress-strain response during cycling, even though the loading cycle remained constant.
From these studies it was concluded that:
- the failure locations for all tests were correctly predicted;
- the cyclic endurance for all tests was conservative but not unduly so;
- cycle by cycle analysis accounting for cyclic hardening is a very important requirement for obtaining good predictions;
- the effort required to perform assessments is much reduced by the use of the software.
Oil fired boiler components
With the liberalization of the electricity market and the introduction of the electricity pool in Singapore, the competition among the power companies to produce electricity is increasingly intense. One of the key factors for them is to ensure high power plant availability and reliability.
For conventional simple-cycle power plants, the relatively short inspection interval of boilers is one of the dominant factors that impact the overall plant availability. Plant inspection is important for identifying incipient failure, maintenance planning and giving assurance of component integrity. However, inspection needs to be correctly targeted and timely, based on the consideration of the risk of component failure.
The Ministry of Manpower (MOM) is the safety regulator for boilers and pressure vessels in Singapore. MOM operate a statutory one-year inspection period for boilers. Applications to extend inspection intervals beyond one year require a written scheme to justify safe operation over the extended period.
Tuas Power operates Tuas power station which incorporates 2 x 600 MW simple-cycle generating units. Tuas Power has put in place a comprehensive scheme and programme of due diligence to guarantee safe operation of both 600 MW oil fired boilers. This scheme has gained approval from the MOM to extend the boiler statutory inspection period to three years. One aspect of the written scheme is that the boilers are equipped with a Boiler Life Monitoring System (BLMS) for real-time condition monitoring of high temperature creep-fatigue damage of critical boiler components3.
Creep damage is assessed on the basis of pressure driven creep using a creep-rupture assessment procedure where the material’s behaviour is described by a parametric equation, relating temperature and stress to rupture life. The primary online inputs required to establish life usage are temperature and pressure, which in the BLMS system are taken from sensors mounted at positions of interest.
Cyclic damage develops in the bore of thick-section components because of thermal gradients established through the wall during operating transients. Fatigue damage is concentrated in areas of stress concentration at internal surfaces, such as the ligaments between tube stubs in headers.
Damage transients are generally those where hot or cold steam (or water) enters thick section components, thereby producing a sufficient down-shock locally in the bore that a high strain fatigue cycle is generated. If the component is operating below the creep temperature range, the damage will accrue by high strain (low cycle) fatigue. If it is operating within the creep temperature range, damage will accrue by creep-fatigue.
The algorithms used for this creep-fatigue damage system are earlier versions of the fe-safe/TURBOlife algorithms based on the UK strain-based methodologies. Individual components are ‘customized’ using finite element analysis so that stress, strain and cyclic loading are determined on-line from primary inputs of time, pressure and temperature. Typical monitoring locations for a superheater outlet header are shown in Figure 3.
Figure 3: Superheater outlet header monitoring locations for creep-fatigue damage
Straining for success
Strain-based methods for the assessment of creep-fatigue damage during thermo-mechanical component cycling, based on mechanistic understanding of damage accumulation have been continuously developed over the past 40 years in the UK. These methods allow for influences related to multiaxial loading, nested operating cycles and material property variability within components.
Also, creep relaxation and cyclic hardening can be accounted for, which cause variations to stress-strain cycles where load cycles are nominally identical. By accounting for these influences, realistic predictions of creep-fatigue failure location and endurance can be obtained.
These methods have been successfully applied offline to component life assessment and online to power station component assessment based on actual operating history. Benefits accrue regarding the development of safety justification and increased availability through the extension of inspection intervals. Also the correlation of damage with actual plant history enables operating procedures to be optimised for maximum plant life.
1. R5, Issue 3. Assessment Procedure for the High Temperature Response of Structures, British Energy Generation Limited, Gloucester, UK, 2003.
2. fe-safe/TURBOlife. Software for the Endurance Assessment of Thermo-mechanical, Creep-fatigue component cycling, Safe Technology, Sheffield, UK.
3. Extension of Boiler Inspection Interval to Improve Power Plant Availability, Tan Kah Chay George, David Worswick & Robert James Brown. Presented at POWER-GEN Asia, Kuala Lumpur, Malaysia, September 2001.