Creep Classification of Grade 91 Steel
Grade 91 steel is superior to many other industry used materials for its increased resistance to many properties, including creep. These are dependent on the formation of a specific martensitic lath microstructure developed through correctly performed heat treatments. This structure is easily altered during installation, maintenance, and routine operation. The degradation of this microstructure to equiaxed grains with an increased mean diameter can lead to significantly reduced property strengths and so to an unexpected failure . Current damage assessment methods over-estimate the expected lifetime of components made of this material as they do not take into account the microstructural changes, and so property changes, over time.
Damage Assessment Method
Surface replication is the widely used technique for assessing the level of creep damage; it is able to detect defects much earlier than others. It is often performed on areas thought to be most damaged or at risk; in particular welds as they are in locations of particular stress concentration and often failure. The method is briefly described here :
- Area is grinded to high metallographic standard
- The area is treated to remove all traces of cold work through polishing / etching
- Surface is applied with acetone and acetate foil
- The foil conforms to surface shape while drying
- The foil is then removed with the image of etched surface impression
This process then allows the structure of the material to be analysed off site, classified and future action to be taken is decided.
Figure 1 – Graph showing typical creep curve model of strain against time 
Previous Classification Models
The formation of small cavities on grain boundaries in the microstructure which can evolve to larger sizes, coalesce and form cracks is creep failure. The current creep classification model for materials, the Neubauer model, shows three main stages: primary, secondary and teritary, as shown in figure 1. This model is used to assess the state of damage in the component using replica mtallography. The primary stage is defined by the continually decreasing creep rate until it becomes constant in the secondary stage- also known as the steady state. Tertiary the strain rate increases rapidly before an eventual rupture.
The stage the material was measured to be at then determined the future action to be taken. Materials deemed to be in the secondary stage required just observation, approaching the tertiary stage would either require inspections to be fixed or for a small repair, while in the tertiary stage options include limited service until repair or immediate repair. 
New Classification Models
Figure 1 shows the stages and development of the creep rupture mechanism in grade 91 steel. Many studies of the creep properties have been conducted with conclusions that both creep strain and creep time of tertiary creep stage far exceeds that of secondary. This is vastly different to the Neubauer model as the secondary ‘steady state’ phase is almost irrelevant when compared to the tertiary stage- it is in this stage that the main damage and rapid growth of voids/cracks occurs in grade 91. This significantly reduces the time available for defects to be detected, and even when they are, harder to predict the estimated lifetime remaining (as it is no longer in the region of constant creep rate as modelled by Neubauer and so no longer is simple extrapolation of data points viable). 
Part of the difficulty in identifying defects in early stages of creep is that many of the voids in the microstructure appear to be sub-surface and so not picked up using surface replica techniques. It is for this reason that some reports state strain gauge testing must be used instead of metallography to provide an earlier warning of creep damage , and research is being conductued on producing an industry wide deployable sensor using the electromagnetic spectrum to detect internal changes in microstructure. 
To continue the use of metallography techniques, many papers claim to have equations which can successfully predict the lifetime remaining of a component of grade 91 steel, however as of yet there seems to be no consensus among them. Many put forward extrapolation equations with specific calculated parameters to account for the non-steady creep rate; providing certain stable conditions such as a maximum temperature to ensure no phase changes occur [8-12]. In particular, a method of using calculated C* values to predit the creep crack growth has been found to correlate well against experimental data of P91 .
Figure 2 – SEM photographs showing void development in P91 
- Grade 91 steel does not follow the creep curve as modelled by Neubauer, as the time in the secondary steady-state phase is greatly reduced and instead the tertiary stage is the significant phase in which voids grow and coalesce.
- One potential issue with using surface replication techniques for grade 91 componenents is the location of voids and cracks forming sub-surface.
- The second problem with surface replication includes the time at which smaller defects are noticed is greatly reduced on grade 91 components – meaning fracture could occur between inspections.
- The estimated lifetime prediction of components of grade 91 cannot be predicted using extrapolation as with the Neubauer model. Instead, new methods accounting for the non-steady state are emerging.
- In particular is the C* method described by S. Mohammed is of worth further investigation into.
- Li, H. & Mitchell, D. Microstructural characterization of P91 steel in the virgin, service exposed and post-service renormalized conditions, 2013. [Cited 26/07/2016] http://ro.uow.edu.au/cgi/viewcontent.cgi?article=2837&context=eispapers
- ETD, Use of replication and portable hardness testing for high temperature plant integrity and life assessment, 2011. [Cited 26/07/2016] http://fr.slideshare.net/hussainetd/use-of-replication-and-portable-hardness-testing-for-high-temperature-plant-integrity-and-life-assessment
- C.Furtadoa, I. Le Mayb, High Temperature Degradation in Power Plants and Refineries, Vol. 7, No. 1, 2004. [Cited 26/07/2016] http://www.scielo.br/pdf/mr/v7n1/19336.pdf
- P.R. Palaparti,E.I Samuel, B.K. Choudhary, M.D. Mathew, Creep Properties of Grade 91 Steel Steam Generator Tube at 923K, 2013 [Cited 22/07/2016] http://www.sciencedirect.com/science/article/pii/S1877705813005754
- Hobbs, M. Echivarre, C. Jania, S. Whitson, Creep Performance and Microstructural Characterization of the Type IV Region in Grade 91 Steel Weldments, 2015 [Cited 19/07/2016]
- Maharaj, J.P. Dear, A. Morris, Imperial College London; A Review of Methods to Estimate Creep Damage in Low Alloy Steel Power Station Steam Pipes, 2009 [Cited 22/07/2016] https://spiral.imperial.ac.uk/bitstream/10044/1/14137/6/Strain_45_4_2009.pdf
- Liu; J. Wilson, M. Strangwood, C.L. Davies, A. Peyton, J. Parker, Electromagnetic evaluation of the microstructure of Grade 91 tubes/pipes, 2015 [Cited 22/07/2016] http://www.sciencedirect.com/science/article/pii/S0308016115000691#bib6
- Ankit , Remaining Creep Life Assessment Techniques Based on Creep Cavitation Modeling, Metallurgical and Materials Transactions A, May 2009 [Cited 20/07/2016] https://www.researchgate.net/publication/225479913_Remaining_Creep_Life_Assessment_Techniques_Based_on_Creep_Cavitation_Modeling
- Pohja, S. Holmström, H.Y. Lee, Recommendation for Creep and Creep-fatigue assessment for P91 Components, MATTER, 2016, [Cited 19/07/2016] http://publications.jrc.ec.europa.eu/repository/bitstream/JRC94508/jrc94508_matter_4-6_creep-fatigue.pdf
- Wilshire, P.J. Scharning, A new methodology for analysis of creep and creep fracture data for 9-12% chromium steels, International Materials Review, [Cited 21/07/2016]
- Holmström, C. Jia-Chao, Recommendation for the negligible creep domain for P91, MATTER, 2016, [Cited 22/07/2016] http://publications.jrc.ec.europa.eu/repository/bitstream/JRC94451/jrc94451_matter_5-3_recommendation_nec_p91%20ed.pdf
- Mohammed, Experimental and finite element studies of creep crack growth in P91 and P92 weldments; University of Nottingham, 2011 [Cited 19/07/2016]