Abstract
Fossil fuels continue to be the primary source of energy in the world. The worldwide demand for clean and affordable energy will continue to grow, and a strong emphasis has been placed on increasing the efficiency and reducing the carbon footprint of new and existing fossil fired power plants. Throughout Asia, Europe and the USA, this demand is being met with programmes to develop advanced materials that have enhanced high temperature creep and corrosion properties. A new class of ferritic alloys, known as creep strength enhanced ferritic steels, has been developed to meet these requirements. This article focuses on the weldability of the advanced ferritic alloys used in boilers and boiler components of ultrasupercritical coal fired power plants. This review focuses on alloy selection; welding and weldability issues, including in service weld failure such as type IV cracking; welding of dissimilar metals; and weld repair. Future articles will address the welding and weldability issues of two other classes of materials, namely austenitic stainless steels and nickel base superalloys.
Keywords
Introduction
Efforts are under way to meet the world demand for clean and affordable energy, particularly in fossil power industries, to increase the plant operating efficiencies and to reduce carbon footprints of conventional coal fired power plants and cycling heat recovery steam generator applications. This requires operating the plant at higher steam temperatures and pressures. Accomplishing this goal requires development and utilisation of advanced materials. In addition, successful development and utilisation of these materials in advanced power systems depend on their ability to be welded or joined. Therefore, welding and weldability of these new materials become key issues. Weldability plays a critical role in the selection of materials, in particular for the advanced coal fired power plants.
Over the last 50 years, coal fired power plants have evolved from a subcritical plant, which operates at a steam temperature of 540°C and at a steam pressure of 16·5 MPa with a normal efficiency of 35% HHV [the higher heating value includes latent heat of vaporization of water in the combustion of coal. In this paper, plant efficiency is always noted in terms of HHV.], to the present ultrasupercritical (USC) plant, which is capable of operating at temperatures >593°C, a steam pressure >24·8 MPa and an efficiency up to 42%. Now, efforts are under way in the USA, Europe and Japan to increase the efficiency of USC power plants to >45% by increasing the steam temperature and pressure; these parameters are representative of so called advanced ultrasupercritical (AUSC) technology. Efforts for alloy development for higher efficiency USC power plants have focused on the optimisation of current creep strength enhanced ferritic (CSEF) alloys and the modification of traditional chromium–molybdenum (Cr–Mo) steels. These alloys have excellent high temperature creep strength and excellent physical and corrosion properties. Conventional alloy design principles are used to enhance creep strength.1– 3 Recently, within the realm of alloy design, Bhadeshia4 has reviewed a currently available quantitative methodology to attempt to design new heat resistant steels and welding consumables.
In addition to alloy development, many welding and joining issues require attention to address the material needs of USC power plants. Unlike the other structural materials, CSEF steels require special care and attention during welding if they are to retain their high temperature properties. They need a very precisely controlled welding procedure, and susceptibility of these alloys to type IV cracking [type IV cracking is creep cracking (also known as “midlife cracking”) that occurs during service in the heat affected zone of ferritic steel weldments] needs to be understood and mitigated. In this paper, power plant classification, materials selection, welding and weldability issues, weldability of USC power plant materials and topics including weld repair and welding of dissimilar metals are reviewed, and a prognosis for welding them successfully is offered.
Welding and weldability
Welding is a critical and complex fabrication technology used in the construction of energy systems. The term ‘weldability’ has no universal definition. The American welding society handbook defines weldability as ‘the capacity of a material to be welded under the imposed fabrication conditions into a specific, suitably designed structure and to perform satisfactorily in the intended service’.5 Although adequate, that definition does not address all of the issues related to material behaviour when exposed to the weld thermal cycle.
Recent discussions of the authors with others in the field have led to a new way of looking at weldability. To put it simply, ‘weldability is a measure of the ease with which a metal or an alloy can be welded or joined without degradation that is detrimental to the weldment microstructure or properties during or after welding and for the duration of intended service’.6 The definition is inclusive of most of the problems encountered by the welded structures (e.g. hot and crater cracking, cold or hydrogen cracking, reheat cracking and in service problems such as type IV cracking). Undeniably, the term ‘weldability’ remains a highly subjective term, whose definition may be specific to an individual or institution.
During welding, the original microstructure of the material is destroyed by melting, which creates a liquid pool that solidifies upon cooling. This region is called the fusion zone (FZ). Beside the FZ is the heat affected zone (HAZ), a region of the metal that experiences heat but does not melt. The rest of the material, which is not affected by the welding heat, is the base metal.
Figure 1 is the schematic diagram showing the development of various subzones in the HAZ of P(T)91 weldment during welding and their approximate correlation to the calculated P(T)91 equilibrium phase diagram7 (see Table 1 for alloy compositions of CSEF steels and the designations used and Table 2 for CSEF boiler material applications).8, 9 Similar regions exist for all of the CSEF steels. In particular, the figure shows the various submicrostructural zones due to the thermal gradient experienced by the HAZ. Often, most of the welding and weldability problems encountered during welding are not only associated with the structures of the FZ and the HAZ, but also due to the gradients in structure, which are of critical importance in focusing creep strain, for example in the type IV cracking phenomenon. Also during welding, thermal and solidification shrinkage stresses develop, and most of the stresses remain in the weldment as residual stresses and may affect the performance of CSEF weldments in the following ways:
Schematic diagram of various subzones in HAZ of P(T)91 weld and its approximate relation to calculated equilibrium phase diagram7 (courtesy of P. Mayr)
Composition in of candidate CSEF steels/wt-%
ww*: postweld heat treatment (PWHT) required. Note that the requirement of PWHT in waterwall applications introduces complexity that may be difficult to address.
residual stresses in the weldment in the presence of a corrosive atmosphere may increase the potential for stress corrosion cracking (SCC). As welded grade 91 components that were exposed to uncontrolled corrosive atmospheric conditions were found to be affected.10 Additional issues, attributed to SCC, have been reported during start-up of new supercritical (SC) units employing grades 23 and 24 in waterwall panels11, 12
the presence of residual stress and a triaxial stress state may induce reheat cracking in components made from grade 2313– 16
the presence of residual stress, combined with improper preheat in all CSEF materials, may lead to issues resulting in hydrogen induced cracking.17– 20
Several other factors must be considered before welding [e.g. joint design, selection of a joining process, joint restraint, postweld heat treatment (PWHT), procedure qualification]. Typically, commercial alloys are selected on the basis of mechanical properties or corrosion resistance. Generally, welding and weldability of the selected material are considered or evaluated at the end of the material selection cycle; such negligence of welding considerations may result in significant problems in the construction of a component.
Weldability of a material is often determined by a number of metallurgical and non-metallurgical factors, such as composition, microstructure, properties, process parameters and restraint. In actuality, to determine the weldability of an alloy, a testing procedure that duplicates the conditions that the material experiences during weld fabrication should be employed. In most cases, it is not feasible to demonstrate weldability with a full size mock-up component with the appropriate materials, welding consumables and restraint. Therefore, a number of weldability tests have been developed to appropriately characterise the behaviour of the materials during welding and are described in the literature.21– 23 Weldability tests have been designed to evaluate the weldability of metals and alloys during fabrication (e.g. hot cracking tests, implant tests, laminar tearing tests) and during service (e.g. tensile and ductility tests, fracture toughness tests, creep resistance, fatigue and corrosion resistance tests).23 Despite the abundance of weldability tests, no single universal test exists that can determine the weldability of materials. One possible solution would be to develop an integrated computational materials engineering model with the industrial software tools that are currently available (Table 3).24
Software tools for integrated modelling24
Coal fired power plants
The development of advanced fossil energy technologies for increased efficiency and a reduced carbon footprint requires the development of materials and structures that can withstand an increase in operating steam temperature and pressure. In its most simplistic description, a boiler carrying high pressure steam consists of tubes (i.e. small outside diameter and thin walls) and pipes (i.e. large outside diameter and thick walls). Pipes can be subdivided into headers that collect steam and piping that transports hot steam to the turbine. Tubes consist of furnace wall tubes and superheater/reheater tubes. Figure 2 shows a typical header consisting of thick walled pipe penetrated by a number of tubes.25
Photograph of typical header25 (with permission of Springer Science)
The terms ‘subcritical’, ‘supercritical’ and ‘ultrasupercritical’ refer to the temperature and pressure of a power plant's working fluid (steam). Steam conditions can be illustrated in a phase diagram for water.26 Figure 3 (not to scale) shows phase equilibria for water with various phase fields (steam, water and ice). Solid lines A, B and C show where two adjacent phases coexist. It also shows the values of pressure and temperature at which water and steam are in equilibrium with each other. Point D in Fig. 3 is the triple point (i.e. the temperature and pressure at which the three phases, namely water, ice and steam coexist). The normal boiling point of water in Fig. 3 is at 100°C and 0·1 MPa; at that point on curve C, water and steam coexist. Increasing pressure and temperature above that point causes water and steam to coexist as two phases in the subcritical range. Point E on curve C, which occurs at 374°C and 22·1 MPa, is called the critical point of water (Fig. 3), above which the phase boundary between water and steam terminates and distinct steam and water phases do not exist. Increasing pressure and temperature above that point pushes steam into SC range. Further increase in either pressure or temperature would produce USC steam. The term ‘advanced ultrasupercritical’ has been recently used to delineate more aggressive steam conditions from those representative of ‘USC’ although there is no phase change in the steam.
Phase diagram for water system26
Coal fired power plants are classified according to steam conditions and plant efficiency.27 Increasing steam temperature and pressure is desirable because doing so increases efficiency (i.e. it reduces the amount of coal required to achieve an equivalent amount of energy) and therefore reduces the emission of carbon dioxide. Table 4 shows different types of power plants, their operating conditions and their efficiencies. The subcritical power plants typically operate with a steam temperature of <540°C and a pressure of 16·5 MPa, resulting in a thermal efficiency of 35%. An increase in temperature and pressure moves the operating conditions into the SC range. Typical operating conditions for SC power plants are a steam temperature of 565 to 580°C and a pressure of 24·8 MPa, resulting in an efficiency of 38%. The USC plants have an operating temperature in the range of 593 to 620°C and a pressure >24·8 MPa. These state of the art plants operate at an efficiency >42%. Future AUSC plants will operate at temperatures >700°C and at pressures >27·5 MPa, resulting in efficiencies of >45%.
Classification of coal fired power plants9
Candidate ferritic alloys for AUSC boilers
To achieve the efficiencies given in Table 4, AUSC power plants will require the use of materials with high creep strength, high corrosion resistance and adequate fabricability. In the opinion of materials experts, the materials required for constructing an AUSC power plant are commercially available.25 An assessment of weldability is of paramount importance and necessary for the implementation of candidate materials. In a US programme during the last decade, the Electric Power Research Institute (EPRI) has been leading an effort to apply existing materials and to develop new materials for AUSC boilers. Under the programme, the US Department of Energy and the Ohio Coal Development Office have funded a consortium of major US boiler manufacturers to evaluate and develop materials for advanced steam cycle coal fired power plants.27, 28 The programme has set target steam temperature of >760°C and a plant efficiency of >45%. Similar programmes to develop the necessary technology to achieve AUSC conditions of 700°C and a plant efficiency >45% are under way in Europe, Japan, China and India.28– 31 Regardless of the particular development effort, the singular goal of the worldwide interest in AUSC conditions is to increase plant efficiency through the utilisation of existing materials fabrication routes and/or development of new materials.
In advancing the AUSC fossil power plant technology, the first task was to identify and/or develop materials with adequate high temperature strength and corrosion resistance to withstand the operating conditions of the AUSC boiler components. The alloys selected for making AUSC boiler components such as headers and tubes have to satisfy the high temperature creep strength requirement and must have excellent resistance to fireside corrosion and high temperature steam oxidation. The materials should also have the required long term strength at the desired temperature and pressure. Selection of materials for the headers and piping may be based on the criterion of the temperature to cause rupture in 105 h at 100 MPa. Figure 4 shows the temperatures for various AUSC alloys;32, 33 Fig. 5 shows allowable stress derived by taking two-thirds of the extrapolated 105 creep rupture strength for various steels.32, 34
Average temperature for rupture in 105 h for various boiler materials32
Until a decade ago, power plants depended on low alloy ferritic steels for metal temperatures to ∼580°C. At temperatures exceeding that threshold, oxidation resistance was not adequate for low chromium Cr–Mo materials. Therefore, austenitic steels were used with temperatures >580°C. Because of their higher coefficient of thermal expansion and low thermal conductivity, austenitic stainless steels are susceptible to thermal fatigue problems in headers, main steam piping, valves and other thick section components. Furthermore, compared with ferritic materials, austenitic steels are also costly. For those reasons, with a few exceptions, the use of austenitic stainless steels has been limited to tubing.
A number of alloys have been considered for heavy section components such as pipes and headers. The CSEFs, a new family of Cr–Mo steels, have been developed to extend the operating temperature range to 620°C. A number of 2·25 wt-%Cr and 9–12 wt-%Cr steels have been developed to fill the needs of current USC coal fired power plants. Alloy P(T)91 has been in the American Society of Mechanical Engineers (ASME) code for nearly three decades, and its first commercial use was in the late 1980s. It was developed by Oak Ridge National Laboratory and Combustion Engineering in the 1970s35 for applications in nuclear pressure vessels for the fast breeder reactor programme. Although the intended application never came to fruition, P(T)91 was approved by ASTM and ASME in 1984 and saw primary interest in the fossil fired power plant industry for utilisation in tubing, piping and headers. Alloy P(T)92, a modification of P(T)91, was produced by substituting W for Mo and by the addition of controlled amounts of B and N. P(T)92 alloys are actively being utilised up to 620°C. Ultimately, the use of 9 wt-%Cr CSEF steels may be limited by oxidation resistance. To address that problem, a new 12 wt-%Cr steel has been developed and is being commercialised. That material, called VM12SHC, is limited to a wall thickness of 12·7 mm (0·50 in.) and has an advertised strength equivalent to P(T)91. Other CSEF steels have been developed to address a wide range of issues, including P(T)122, E911, P(T)23 and T24 (see Table 1). Figure 6 shows the evolution of these steels.1, 32
Elements such as W, V, Nb, B and N have been added to mainstay Cr–Mo steels to improve high temperature creep strength. The modification of the base alloy has resulted in the development of several new advanced ferritic steels that have a tempered bainitic microstructure [T(P)23 and T24] or martensitic lath structure [P(T)91, P(T)92, P(T)122 and VM12]. The bainitic or martensitic lath microstructure is further strengthened by a combination of M23C6 type carbides, MX type carbonitride precipitates and Mo and W solute atoms in solid solution.2
Creep strength enhanced ferritic steels have excellent high temperature creep strength, high thermal conductivity and a low coefficient of thermal expansion. They are also less expensive than austenitic stainless steels and thus are looked at favourably and are used extensively worldwide as the material for pipes and headers.25 At a given temperature, pressure and allowable stress, a reduction in pipe wall thickness of >2∶1 is possible if P(T)91 is used instead of 2·25Cr–1Mo material. Figure 7 shows relative wall thickness for P(T)91, P(T)92 and other ferritic alloys.36 For the same design conditions, less thermal stress will develop in thinner walled piping and tubing made of P(T)91. Reduced wall thickness also allows for reduced joint thickness and hence less weld metal deposited and less time to weld. All of these factors increase productivity. Although these alloys have excellent properties required for the USC power plant applications, failures have been reported recently in power plants after a few years in service in the HAZ of weldments.37 This premature failure is known as type IV cracking. One of the main reasons for this failure is that CSEF steels, when exposed to fabrication or repair procedures, are not capable of regaining the original microstructure that is responsible for their excellent creep properties.38 The new alloys are being utilised in steam applications up to 620°C and/or pressure up to 27·6 MPa.39
Relative wall thickness for piping and tubing made of P(T)91, P22 and other ferritic steels for same design conditions36
The basic P(T)91 steel has a normalised and tempered martensitic structure. The creep strength of the steel is mainly derived from the solid solution effects and from the precipitation of stable carbides. The principles by which high temperature, high strength steel can be designed are the following:3
precipitation hardening by alloy additions of W, Nb, Ti, C and N
solid solution hardening by additions of W and Mo
dislocation hardening
hardening due to the presence of boundaries and sub-boundaries that impede dislocation motion.
The strengthening of Cr–Mo steels is mainly due to a microstructure consisting of high dislocation density and precipitation of M23C6 carbides and MX carbonitrides along the sub-boundaries within the grains.
Figure 8 shows micrographs produced by scanning electron microscopy and transmission electron microscopy of a normalised and tempered P(T)91 steel.40 Figure 8a shows a scanning electron microscopy (SEM) image of prior austenite grain boundaries (PAGBs), and higher magnification reveals well defined laths of tempered martensite structure with the carbide precipitates visible. During long creep exposure (thousands of hours at 600°C), the tempered martensitic structure changes by precipitation of a Laves phase Fe2 (W, Mo) and a Z phase that is a complex nitride of the type Cr (Nb, V) N. The concentration of W and Mo in the alloy determines the amount of Laves phase that forms. The precipitation of Laves phase and Z phase causes the loss of long term creep strength of the base metal. This is due to the precipitation of Laves phase, which reduces the amount of matrix W and Mo and thus reduces their solid solution strengthening effects. Large particles of Z phase also form at the expense of finely dispersed VN, a phenomenon that has been largely associated with the degradation of creep properties of 10 to 12 wt-%Cr CSEF steels such as grade 122. This results in the formation of a VN free zone around the carbide. Also during long time creep exposure, the carbides and the carbonitrides coarsen along the PAGBs by the Ostwald coarsening mechanism. All of these changes lead to the loss of creep strength of the alloy after long term creep exposure.
Images (a SEM and b TEM) of normalised and tempered P(T)91 steel40 (with permission of Springer Science)
Most of the improvements in strength of ferritic alloys have come from adding or subtracting elements to the base composition and manipulating the microstructure. Recently, Abe2 has shown that it is possible to produce steels containing nanoscale VN that have better properties than P(T)91 by reducing carbon levels in 9 wt-%Cr alloys to 0·002 wt-%. He has also shown that the addition of boron to conventional 9 to 12 wt-%Cr alloys stabilises the M23C6 carbide, thus preventing coarsening of the carbide and the degradation in the properties.
The normalising temperature for P(T)91 alloy is between 1040 and 1080°C. Normalising is carried out to take carbides and carbonitrides into solution completely. The material is then cooled from the normalising temperature to room temperature to produce a fully martensitic structure. It is important for all of the austenite to be transformed to martensite. Otherwise, any retained austenite will have an effect on the properties. Table 5 shows transformation temperatures, normalisation and PWHT ranges for CSEF steels.32 After normalisation, the fully martensitic structure in the alloy is tempered. Ductility and toughness increase during tempering. Care must be exercised in all heat treatments for these steels. In particular, the tempering temperature for P(T)91 must not exceed the A1 temperature. If it does exceed the temperature, austenite will reform, which, upon cooling, will transform to untempered martensite (with a high cooling rate) or soft ferrite (with a low cooling rate). Retention of either one of these transformation products in the material is damaging to the toughness and creep properties.41 Furthermore, at temperatures near the A1 or after the accumulation of many successive heat treatments over the life of a component or improper processing, overtempering will occur and will result in degradation through precipitate coarsening and recovery at the subgrain boundaries. This will further degrade the creep strength of the material.
Transformation temperatures in °C, Normalization and PWHT ranges for CSEF steels32
MS: martensite start temperature; MF: martensite finish temperature; BS: bainite start temperature; A1: lower critical transformation temperature; A3: upper critical transformation temperature.
VM: Vallourec and Mannesmann; Alex.: Alexandrov.
The chemical composition of the alloy determines the maximum temperature for tempering and ferrite formation. Recently, it was pointed out that the ASTM chemical composition specification may be wide enough to exceed the critical temperature A1 during PWHT.38, 42, 43 Figure 9 shows graphically how the lower transformation temperature A1 could vary for six different 9 to 12 wt-%Cr alloys within the composition range specified by ASTM. Chalk et al. 43 have investigated the effect of variations in composition on the A1 temperature of P(T) 92 steel. Using thermodynamic modelling, they have developed equations to predict the A1 temperature for welding consumables as a function of composition.
Graph representation of possible ranges of A1 temperature for CSEF steels38
Knowing the microstructure of a given steel is important in understanding the influence of a weld thermal cycle on the material and ultimately its performance. The tempered martensitic microstructure of P(T)91 base metal changes drastically when the material is exposed to a weld thermal cycle. The weld thermal cycle alters the microstructure significantly, resulting in a weldment with a large composition gradient in the FZ due to solidification as well as regions in the HAZ containing coarse grained and fine grained microstructures. The HAZ is also associated with a soft zone. The resulting microstructure in the HAZ has been the source of problems and is the primary reason for the structural component failure. The premature type IV failure in the HAZ is unique to CSEF steels and thus needs special care and attention.44
Welding and weldability of AUSC candidate ferritic alloys
Consideration of the appropriate welding technologies (i.e. filler metal, welding process, PWHT method) required during the fabrication and field construction of fossil fired power plants is critical because a representative plant will contain millions of pounds of material and will need tens of thousands of welds to fully erect. Welds may be required between similar and/or dissimilar materials. The process used may be a single process, or multiple processes may be applied, depending on the requirements for fabrication or construction. Figure 10 shows the combination of materials used and the fabrication techniques and joining processes involved in the fabrication of a steam boiler component.45
Combination of materials and welding processes involved in fabrication of stream boiler component45
Utilisation of a particular material in a fossil fired plant requires extensive characterisation. Weldability evaluation of the materials for boilers, pipes and tubes has been studied using product forms representative of actual components. For example, in addition to creep strength requirements for the base metal, weldment strength is also a determining factor for material selection for boilers, tubes and pipes. Since the creep strength of the weld metal is less than that of the base metal, it is recommended that a ‘weld strength reduction factor’ used in the design of high temperature nuclear power plants be used in designing fossil power plant boiler components.46, 47 The term ‘weld strength reduction factor’ is defined as the ratio of the stress to rupture for weld metal to the stress to rupture for boiler materials for a given time and temperature. The ratio is particularly important in the construction of long seam welded piping and may also be important in regions of a component or system where a weld may be influenced by large bending stresses.
Welding of alloy P(T)91 and other CSEF steels can appear to be very difficult compared with welding of other steels, but it is relatively straightforward if the welding procedures and specifications are followed. Creep strength enhanced ferritic steels are highly hardenable, and upon cooling after welding, they undergo phase transformation that results in a fully martensitic structure. To obtain the desired properties, the welds must undergo PWHT, and an interpass temperature must be maintained.41 It is vital to ensure that CSEF steels are not treated like their lower alloy Cr–Mo cousins. Many failures in the fossil fired power plants are attributable to faulty heat treatments that resulted from the lack of training and supervision. Therefore, adequate training and supervision are critical to obtain the properties that can be realised with the CSEF class of steels. Because of the increased use of CSEF steels in current power plants and their anticipated use in future power plants, the American Welding Society (AWS) and ASME are defining the codes and guidelines for welding CSEF steel, piping and tubes, either by writing new documents or by revisiting existing documents (D10·10, D10·21; ASME's VIII, B31·1, and B31·3).41
Filler materials for AUSC power plants
Creep strength enhanced ferritic alloys can be implemented in coal fired AUSC plants if they can be welded successfully. As indicated in Table 2, CSEF steels are being proposed in atypical applications, such as the use of T91/T92 in waterwall panels. Many of the established welding processes can be used to weld alloy P(T)91 successfully.20 However, development of consumables that provide the required weld metal creep, corrosion and toughness properties is a major concern. Many consumables are available to weld P(T)91, and they are made to AWS specifications.48– 52 Tables 6 shows various consumables available for welding CSEF steels.20, 44 The composition of a given CSEF filler material is generally developed to match the base metal composition. Filler metal composition should be balanced to avoid ferrite formation and to ensure a fully martensitic structure in the weld metal. Presence of ferrite in the weld metal leads to poor toughness and creep properties. Work has shown that the filler metal with reduced Nb, N and Si improved toughness of the weld metal.53 Having investigated some experimental alloys, Sireesha et al. 54 have shown that the Si and Nb content in the consumable to weld P(T)91 should be lower than that in the base metal (higher levels of Si and Nb promote possible retention of ferrite and reduce weld metal toughness). The Si should be in the range 0·15 to 0·25 wt-%. Nb should be in the range 0·03 to 0·05 wt-%. In addition, an increase in the Ni content has been shown to have a beneficial effect on toughness.48, 53, 55 Ni lowers the A1 temperature and should be maintained in the range of 0·4–1·0 wt-% to prevent the A1 temperature from falling below the PWHT temperature, leading to austenite formation, which upon cooling transforms to untempered martensite. Hence, Ni content higher than 1 wt-% may promote austenite formation in the weld metal, and it is detrimental to its toughness properties. Additionally, the weld metal composition should be balanced such that the weld metal is fully martensitic without any austenite. Specific guidelines must be followed when filler metal is being developed for P(T)91. An extremely well controlled preheat, interpass temperature and PWHT are mandatory to ensure that the required creep rupture properties and toughness are obtained in the weldment.
Note that AWS will formally adopt ‘-B91’ to replace ‘-B9’ in order to be more specific in identifying CSEF filler materials.
Because a ‘G’ designation is specified, the CSEF steel grade must be specified to avoid confusion between materials. In this case, ‘grade 23’ may be sufficient until AWS formally adopts a ‘-B23’ designation.
Because a ‘G’ designation is specified, the CSEF steel grade must be specified to avoid confusion between materials. In this case, ‘grade 24’ may be sufficient until AWS formally adopts a ‘-B24’ designation.
Because a ‘G’ designation is specified, the CSEF steel grade must be specified to avoid confusion between materials. In this case, ‘grade 92’ may be sufficient until AWS formally adopts a ‘-B92’ designation.
Because a ‘G’ designation is specified, the CSEF steel grade must be specified to avoid confusion between materials. In this case, ‘grade E911’ may be sufficient. There are no formal plans for AWS to adopt a specific 9%Cr designation for E911.
Because a ‘G’ designation is specified, the CSEF steel grade must be specified to avoid confusion between materials. In this case, ‘grade VM12SHC’ may be sufficient. There are no formal plans for AWS to adopt a specific 12%Cr designation for VM12SHC or other 12%Cr CSEF alloy.
During the welding of CSEF steels, it is also important to preheat the base metal and to maintain an interpass temperature in order to mitigate hydrogen cracking problems. A preheat temperature of 150–200°C and an interpass of ∼350°C are recommended; the required preheat is material and restraint dependent. Guidelines regarding preheat and interpass for CSEF steels are given in Table 7.32 When required by governing code bodies, proper PWHT of all CSEF steels is critical to achieve proper tempering of martensite.56– 58 The time at temperature for PWHT depends on the thickness of the part. The PWHT temperature should also be controlled so that the formation of austenite, which would result in the formation of untempered martensite, is avoided. For a typical PWHT, the allowable ASME range for holding temperature is given in Table 7. If thickness is <5 in., it is 1 h in.−1 of hold time; if thickness is >5 in., then it is 5 h+15 min per inch of thickness. Figure 11 shows a typical weld thermal cycle experienced by the P(T)91 steel component during welding and PWHT.20 Control over thermal gradients is also critical.48 Existence of a thermal gradient in a heavy section component due to the method of heating (soak band) during PWHT may cause defective heat treatment. In such a case, a non-uniform PWHT may result if the outside surface temperature reaches PWHT temperature well before the inside of the component does. Practices such as increasing the size of the heating band, using properly sized heating pads to prevent large gaps or overlap through the circumference of the tube or pipe and using a control thermocouple under each heating pad are recommended in addition to the guidelines specified by AWS D10·10.59
Typical weld thermal cycle for P(T)91 steel component during PWHT20
Recommended preheat and interpass temperatures for common CSEF steels32
The minimum preheat recommendations in this table do not supersede existing code requirements, such as those present in ASME B31·1 for grade 91. Section 131·4·5, which addressed P5A and P5B (some now classified as P15C/E materials) requires 400°F (200°C) for material that has either a specified minimum tensile strength in excess of 60 000 lb in−2 (413·8 MPa) or has both a specified minimum chromium content >6% and a thickness at the joint in excess of 0·50 in. (13·0 mm); 300°F (150°C) for all other materials having this P number.
If there is concern to have minimal hardness (<350 Hv) without PWHT (to minimise susceptibility to SCC), then a 400°F (204°C) preheat is advised.
Note that welding has been successfully performed using the gas tungsten arc welding (GTAW) process and reduced preheat temperatures.
In the bainitic CSEF steels, interbead and layer to layer tempering can play a particularly important role in determining the properties of a weldment with respect to hardness, toughness and creep.60, 61 The properties depend on a number of variables, including heat input, fluxes, microlloying agents and the interaction of residual elements. In a multilayer weld region, varying microstructures can form due to the thermal cycle experienced by each bead from the subsequent weld passes. Figure 12 shows macrostructure of a multipass weld and possible complexity of microstructures that may exist due to thermal effects and associated HAZs that form during a multipass weld.62 Depending on the sequence in which the weld beads are laid and the properties of the multiple HAZs (orientation, whether they are continuous or discontinuous) could promote type I cracking within the weld metal. In a multipass weld, the bead shape and the welding position also have a significant effect on the weld properties.48 When the weld deposit is thin and wide, it allows for some degree of tempering of the previous deposit from the heat of welding and promotes grain refinement in the previously deposited weld metal.48 Thicker beads limit the degree of grain refinement and lead to loss of toughness in the weld metal. As to the welding position, most of the observations made are on welds produced in the flat (1G) position. A loss of toughness is observed in shielded metal arc welds (SMAW) made in vertical (3G) and overhead (4G) positions due to bead control problems and lack of tempering and grain refinement.48
Macrostructure of multipass weldment62 (courtesy of P. Mayr)
Microstructure development in the FZ depends on the solidification behaviour of the weld pool. The solidification behaviour controls the size and shape of the grains, the segregation and distribution of inclusions and porosity. Solidification behaviour and associated segregation are also key factors in the hot cracking behaviour of the weld metal. Sometimes, it is convenient to consider weld metal as a miniature casting. Therefore, parameters important in determining microstructure in casting, such as growth rate, temperature gradient, undercooling and composition, also determine its character. A comprehensive review of the fundamentals of weld metal solidification is available in the literature.63
In the case of CSEF steels, segregation of low melting residual elements such as S and P during solidification can cause hot cracking (specifically crater cracking) in extreme cases.48 This has been reported in root pass welds when welding the bainitic CSEF steel T24. Segregation of alloying elements (e.g. Cr, C, N, Nb) could further affect the subsequent solid state transformation in the steel. In thick section welding, in which the weld groove requires multiple overlapping weld passes, each subsequent weld pass subjects the earlier pass to a number of weld thermal cycles, leading to a number of HAZs within the weld. This is very important in CSEF steels, which undergo solid state transformation upon heating and cooling. Owing to the short time at temperature, a limited amount of solute redistribution may occur by diffusion.64
Important aspects of weld pool solidification are weld pool dynamics, development and steady state geometry. Weld pool shape is important in the development of grain structure and weld bead shape, size and penetration. In predicting the weld pool shape and penetration, recent theoretical developments include formation of a free surface computational model with a coupled conduction and convection heat transfer model.65, 66 The models are capable of predicting weld pool shape and penetration as well as thermal gradient and cooling rate. A good knowledge of heat and mass transfer will aid in the optimisation of weld bead size and shape control during temper bead procedure development. Solidification of weld metal proceeds in a dendritic or cellular dendritic fashion, depending on the thermal and constitutional conditions at the interface.
Solidification of CSEF alloys appears straightforward with L→δ (ferrite) and the ferrite subsequently transforming to austenite (γ), which then transforms to martensite (for T/P91, T/P92 or VM12SHC) or bainite (for T23/24) upon cooling. Owing to the partitioning of alloying elements between sold and liquid during weld pool solidification of a complex alloy such as P(T)91, extensive solute redistribution of alloying elements occurs. This partitioning of elements, also called segregation, leads to heterogeneity in composition within the weld metal. Segregation on a fine scale of the order of dendrite arm spacing (10 μm) is called microsegregation. This aspect of weld metal solidification is described in the literature.63 When the scale is larger than several dendrite spacings, it is called macrosegregation. Local variations in composition can change the local transformation characteristics of the material.
Recent characterisation and modelling of non-equilibrium microstructure evolution during solidification of P(T)91 alloy have revealed the presence of austenite in the highly segregated (C and N) interdentritic region of the weld metal.67 In contrast, in the base metal, where the composition is homogenous after normalisation, the austenite is completely transformed into martensite. This difference in the microstructure was attributed to microsegregation in P(T)91 during solidification. Although the segregation effects are present, it is difficult to reveal the weld metal solidification substructure or the dendritic structure, mainly because the weld metal goes through subsequent martensitic transformation after solidification, masking the solidification substructure. It is the variation in elemental concentrations and its reaction with a selected etchant that reveal the dendritic pattern during metallographic examination of the weldment structure. The underlying dendritic microstructure has been revealed by applying the special metallographic technique of tint etching. Figure 13 shows the dendritic solidification substructure in P(T)91 weld metal after tint etching. Solidification simulation was carried out using the Schiel–Gulliver model64, 68 and Thermocalc.69 The analysis shows that the first phase to solidify is δ-ferrite (Fig. 1) and that toward the end of solidification, the interdendritic regions were enriched in elements C and N. The enrichment by C and N leads to the stability of austenite γ in the interdendritic region. The variations in Cr concentration in the solid phase were predicted using Schiel–Gulliver method and are shown in Fig. 14. The calculations show the Cr level from the core of the dendrite to the interdendritic region varying from 8·9 to 11·3 wt-% and corresponding C level varying from 0·02 to 0·4 wt-%. Based on these composition levels, the weld metal region should show a large temperature range over which the transformation from austenite to martensite may occur. This indicates possible retention of austenite in the weld metal, leading to an additional martensitic transformation during cooling of the weld metal from the tempering temperature. The impact of microsegregation on the transformation behaviour was confirmed using thermomechanical simulations and time resolved X-ray diffraction experiments.67
Dendritic solidification substructure observed in P(T)91 after tint etching67
Calculated variations in chromium and carbon concentrations from core of dendrite to interdendritic region67
Welding processes
Several welding processes are used to weld CSEF steels. Processes such as SMAW, submerged arc welding (SAW), gas tungsten arc welding (GTAW) and flux cored arc welding (FCAW) are commonly used.48, 70 The process selected may vary depending on the specifics for each weld [e.g. the number of welds to be made, the required deposition rate, the thickness of the material, access to the weld joint and the welding position (flat, overhead or vertical)]. Care should be taken to see that the weld metal contains very low concentrations of tramp elements such as As, Sn, Sb, Pb, S and P. If S or P is at a maximum of 0·010%, crater cracking and grain boundary cracking can be avoided.70 The various processes used to weld 9–12 wt-%Cr steels are also described in the literature.48, 70 Although it is a slow process, GTAW produces the highest quality welds. The weld metal produced by GTAW has been found to have a higher toughness than the welds produced by flux based processes. Weld quality using the gas metal arc welding (GMAW) process can be difficult to consistently reproduce, and only a few fabricators are qualified to use the process. Flux cored arc welding is a high deposition rate process exceeding all other processes except for SAW. The use of the SMAW process to weld CSEF steels is well established and is the most commonly utilised process. The SAW process, used for thick section welding and in seam welding waterwall tubes to membranes, can be automatic or semiautomatic. It is a high deposition rate process, and the bead shape and size should be controlled to achieve a complete interbead tempering. In general, the bead thickness for SAW and FCAW should be no greater than 3·8 mm. This allows for tempering of the previously deposited layer.48
Basic flux is recommended for SAW of P(T)91.70 Other fluxes will reduce the C and increase O and N in the weld metal, affecting the strength level. For welding processes in which elements may be lost across the arc (such as in GMAW, SMAW or SAW), maintaining the weld metal composition control is critical for all welding processes. In particular, the levels of Nb, Ni, C and N must be controlled to result in good creep strength in the weld metal. For grade 91 steel, the weld metal should contain a minimum of 0·09 wt-%C, 0·02 wt-%Nb and 0·02 wt-%N.70 When Ti is present in the weld metal, lower N levels can be tolerated. However, Ti content should not exceed 0·01 wt-%, or TiN will form, and thus N, which is essential for high temperature strength, will be depleted in the matrix.
The composition of the weld metal should be controlled to maintain the A1 temperature so that PWHT can be carried out safely without increasing the possibility of forming retained austenite. For example, the Ni+Mn level should not exceed 1·0 wt-%. Ni+Mn in T/P91 weld metal at Ni+Mn contents >1·0 wt-% may suppress the A1 temperature <790°C (the maximum allowable PWHT temperature), and narrow the range in which PWHT can be performed safely.71 Alloy P(T)91 is a highly hardenable alloy and is subject to hydrogen cracking. In order to overcome this potential problem, a low hydrogen electrode (e.g. with a -H4 designation) should be used.70 The SAW and FCAW wires should be used with a -H4 designation and should be stored in a heated chamber. Efforts are under way to use computation modelling for optimising the process parameters. Vasudevan et al. 72 have used a genetic algorithm (GA) for the optimisation of the A-TiG [tungsten inert gas (TIG) welding is an unconventional term used for GTAW process] process to weld modified 9Cr–1Mo alloys. A-TiG welding is same as GTAW process where an activated flux is used on the surface of the steel to be welded to improve the depth of weld penetration.73, 74 The change in weld penetration is mainly due to the change in weld pool dynamics and the arc plasma characteristics brought about by surface active elements in the activated flux.65, 66 Genetic algorithm is a family of computational models based on evolutionary strategies. It is a field of artificial intelligence often used to generate useful solution to optimisation and search problems. A non-linear and complex process such as welding is ideally suited for optimisation of the process using GA to achieve desired weld quality. The authors have used a GA based model to determine the weld bead shape, welding parameters, depth and penetration and HAZ width. The model has been used successfully to generate optimised process parameters that agree well with observed values of weld bead geometry and HAZ width.
A potential new process that could join CSEF steels while maintaining their creep strength, fatigue resistance and corrosion properties is friction stir welding (FSW). It is a new and innovative process developed in the 1990s at The Welding Institute in the UK.75 The process details are shown schematically in Fig. 15 and are described in detail in the literature.76
Schematic of friction welding process showing interaction of tool pin with material76 (courtesy of The Welding Institute)
Friction stir welding involves plunging a rotating tool into the material to be joined and translating the tool along the joint line. Owing to the frictional heat generated by the rotating tool and the metal, the metal is plasticised and a metallurgical bond is created. Friction stir welding, a solid state, low heat input process, could be a solution to the problems encountered by the Cr–Mo steels during welding, in particular, type IV cracking. Thomas et al. 77 have demonstrated the feasibility of FSW to join 12 wt-%Cr steels with parent metal properties such as cross-weld tensile properties and bend tests. The FSW process offers numerous advantages for most materials; for example, it is an energy efficient solid state process, it does not use shielding gas or filler metal, the parent material composition is retained, steel thickness up to 25 mm can be welded from two sides and the weldment undergoes very little or no distortion. At the same time, it has several disadvantages. For steels and other high temperature materials, tool wear is a major problem. Welding travel speed is not comparable to conventional processes and thus not very economical. Bhadeshia and DebRoy78 have highlighted a number of issues and problems associated with the process.
Weldability issues in CSEF alloys
Type IV cracking
Failures due to cracking in weldments are classified from type I to IV, depending on the location of the crack in the weldment.79 Figure 16 schematically shows various types of cracking. Type I cracking (e.g. solidification cracks) occurs and remains in the weld metal. Type II originates in the weld metal and can extend into the HAZ or base metal. Type III is located in the HAZ, in the coarse grain region; an example would be reheat cracking. Type IV is in the intercritical HAZ (ICHAZ) region.
Classification and types of cracking in weldments79
The precise location of failures in the HAZ attributed to type IV cracking is the question and subject of many studies and has yet to be defined. A well designed microstructure and stress simulation and characterisation using neutron scattering may shed more light on this issue. The effect of type IV failure with regard to long term properties is shown in Fig. 17 80 and compared to base material using Keith Bell's analysis of type IV failures. The data for grade 91 and type IV failures are plotted using a Larson–Miller parameter as a function of stress. The use of the Keith Bell – 20% line in Fig. 17 bounds the minimum for type IV failure.81 An example of a type IV failure in a long term cross-weld creep test in E911 is shown in Figs. 18 and 19.82 In Fig. 18, the damage and failure are clearly in the HAZ, specifically in a region consistent with either the fine grain HAZ (FGHAZ) or the ICHAZ (i.e. type IV failure). This observation is further supported by the electron backscatter diffraction image shown in Fig. 19, where the maximum damage (white areas absent of grain orientation) is clearly concentrated in an FGHAZ. In addition, the failure is characterised by low ductility, with the highest degree of damage occurring subsurface near the midwall, as shown by comparison with the opposite HAZ, which remains intact. These characteristics in this simulated large specimen creep test are characteristic of those encountered in service.
Grade 91 base material creep rupture comparison to Keith Bell type IV failure analysis80
Example of type IV failure location in a cross-weld creep specimen from E911 after 13 945 h at 600°C and 120 MPa82
Example of type IV failure, location and electron backscatter diffraction image across HAZ; creep damage indicated by white regions in colour map82
The 9–12 wt-%Cr CSEF steels are prone to type IV cracking during long time service. The susceptibility of cross-weld bainitic CSEF steel to type IV failure is less clearly defined and is not as well researched.83, 84 The main reason for a type IV failure is believed to be that, once exposed to the weld thermal cycle, the alloys are not able to retain their original microstructure, which is what gives them their excellent creep properties. In order to understand this phenomenon, one has to very closely look at the complex microstructure that develops in the HAZ of CSEF steels.
The microstructure development in the HAZ of the welded CSEF steels P(T)91, P(T)92, P(T)122 steels and others has been discussed by many authors.85– 88 Figure 1 shows schematically the details of the microstructure development in the HAZ during welding as it relates to the phase diagram for P(T)91 steel. As shown in Fig. 1, the HAZ is made up of different zones with unique microstructures, depending on the peak temperatures Tp experienced by various regions during the weld thermal cycle. The different zones are as follows:
a coarse grained HAZ (CGHAZ) next to the fusion line, where Tp reaches much higher than A3
an FGHAZ, where Tp reaches just above A3
an ICHAZ, where Tp is between A1 and A3
an overtempered region, where Tp is below A1.
Figure 20 shows a hardness traverse across a P(T)91 weldment in three conditions: as welded, PWHT and creep tested.86 Figure 21 shows hardness map across an ex service grade 91 weldment.32 The hardness profile shows a dip in hardness in the HAZ, indicating a soft zone. In Fig. 21, the dark blue region in the centre of the thickness and white area near the top of the figure indicate softening consistent with the observations made in the ICHAZ. The location of the soft zone is still open for discussion because of the complex microstructural features make it difficult to find its exact location. It can be anywhere in the overtempered region, the intercritical region or the fine grain region. The soft zone in the HAZ region is the weakest link in the weldment. Premature creep failure (type IV cracking) has been observed to originate in that region and is common to all CSEF alloys.
Microhardness profile across P(T)91 weldment in as welded, PWHT and creep tested conditions86 (with permission of Springer Science)
Hardness across ex service Grade 91w. Dark blue through centre of thickness and white areas near top of figure indicate softening, consistent with observations made in ICHAZ. Representative weld was exposed at 580°C for ∼75 k h32
The evolution of the microstructures in these regions can be rationalised based on the transformation that takes place in each zone and on the stability of the precipitates in the HAZ. In the CGHAZ, the peak temperature is high enough to dissolve the carbide precipitates that effectively pin the austenite grain boundary. In the absence of these precipitates, the austenite grains can grow. In the FGHAZ, the peak temperature is just above A3, and the carbide precipitate dissolution is incomplete, leaving the undissolved particles to stifle the growth of austenite grains in the FGHAZ. In addition, a partial reversion of the microstructure to austenite occurs, as shown in Fig. 1. On cooling, the austenite transforms to untempered martensite. The temperature in the ICHAZ is not high enough to dissolve the carbides, but the temperature might be sufficient to coarsen them by Ostwald ripening. The austenite that forms in the ICHAZ is a low carbon austenite. On subsequent tempering at temperatures below A1, no secondary precipitation in the martensite is obtained from the low carbon austenite or from the overtempered ferrite. The microstructure in this region undergoes enhanced recovery to form a soft zone. The extent of these various zones in the HAZ depends on the thermal gradient in the HAZ. The thermal gradient is influenced by variables such as the material properties and heat input.
During welding, the region of the HAZ experiencing the intercritical temperature range undergoes a change in the microstructure of the original base metal.
The original normalised and tempered base metal microstructure, consisting of high dislocation density M23C6 carbide and carbonitride precipitates along the subgrain boundaries, is replaced by large subgrains, coarsening of the M23C6 carbides, and a change in the shape of the carbonitrides from needles to spheres.
The main problem that needs to be solved to address type IV cracking is to determine which of the zones in the weldment is the weakest and to locate the place where the type IV cracking occurs. To get an understanding of the phenomenon, it would be helpful if the properties of each zone with its unique microstructural features could be determined.
It is difficult to determine the properties of each zone because the HAZ in a weldment is narrow. To overcome this problem, various investigators have carried out creep rupture studies on simulated HAZ specimens,89, 90 cross-weld specimen testing and small specimen creep testing of localised regions in the HAZ.91– 93 Simulated specimens were made by isothermally heating large specimens to the required peak temperatures to reproduce microstructures of the various zones in each specimen. The samples were also made by thermomechanical simulator (Gleeble) by subjecting the samples to a predetermined thermal cycle. The creep results were useful but did not enable the researchers to resolve the issue of which zone of the HAZ is responsible for type IV cracking. Laha et al. 86 have shown by their investigation involving similar regions of HAZ microstructures that the soft zone observed in the grade 91 steels during welding is located in the ICHAZ. Other creep rupture studies on cross-weld specimens show that the type IV cracking occurs in the FGHAZ, which is not the same region as the soft zone. Finite element analysis of stress in a cross-weld specimen during creep testing predicts a triaxial stress state in the FGHAZ, which will accelerate creep void formation that could lead to type IV cracking.85 The stress state in the specimen could be altered by changing the width of the HAZ or weld groove angle.94 Doing so can alter the creep behaviour of the weld. An additional finite element analysis model of the cross-weld specimen also places the origin of type IV cracking in the FGHAZ by allowing for grain boundary sliding.95 Such grain boundary sliding relaxes the constraint from the surrounding heterogeneous microstructure, thus promoting failure in the FGHAZ region.
Although the above analysis places the type IV cracking in the FGHAZ, Laha et al. 86, 87 came to a different conclusion and place the type IV failure in the ICHAZ. Within a Cr–Mo steel weld HAZ, each zone is surrounded by regions that have different creep flow behaviour, and because of this inhomogeneity, a triaxial stress distribution develops across the weldment during creep testing. Using finite element modelling analysis, stress accumulation has been predicted in both the FGHAZ and the ICHAZ.85, 96, 97 The triaxial state of stress that develops in the specimen is the main reason for the development of creep cavitation in the ICHAZ, which leads to type IV cracking. This is further confirmed by the metallurgical observation of more frequently formed creep cavitation in the ICHAZ.86 In addition, metallurgical features such as coarse carbide precipitates in the ICHAZ may increase the possibility of creep void nucleation.
Creep cavities have always been found to be associated with large carbide particles. This has been attributed to the higher stress concentration at the interface between the carbide particles and the soft matrix in the ICHAZ region,96 leading to nucleation of creep cavities. Such an event and coalescence of cavities into cracks during deformation aided by the triaxial stress in the ICHAZ leads to type IV cracking.
Abe et al. 98 have investigated creep rupture properties of Gr 92, Gr 92N, and 9Cr–B (9 wt-%Cr–3 wt-%W–3 wt-%Co–0·02 wt-%V–0·05 wt-%Nb) steels using simulated HAZ specimens. Grade 92 exhibited a significant decrease in time to rupture after a thermal cycle to a peak temperature close to A3; the other two steels showed properties that were the same as those of the base metal and no type IV cracking in welds. Characterisation of the microstructure revealed that the Gr 92 steel after thermal cycle had a fine grain structure with a very little precipitation of M23C6 carbide along the PAGBs and sub-boundaries, whereas the other two steels had a grain size similar to that of the base metal and extensive precipitation of carbides along the PAGBs and also along the lath and block boundaries. The authors attributed the degradation in creep rupture properties in Gr 92 after thermal cycle to A3 temperature to reduced precipitation of carbides and reduction in grain boundary and sub-boundary hardening, not to the fine grain structure.
The assessment of type IV cracking in different regions of the HAZ has been investigated through the use of small punch creep testing, whereby discs measuring 3 mm in diameter and 0·35 mm in thickness were removed from specific regions of the electron beam weld HAZ in a 8Cr–2W–VTa experimental CSEF steel.92 In that study, discs taken from the tempered HAZ, FGHAZ, CGHAZ, fusion line and base metal were compared. The creep life of the FGHAZ was markedly lower than those of the other regions. A disc taken from the ICHAZ was not specifically identified, and it is not clear whether the authors used the tempered HAZ as a synonymous term with the FGHAZ. In addition, the specimens were extracted from a very narrow HAZ of an electron beam weld, and it is unclear by how much the specimens were representative of the various zones of the HAZ. However, the small punch creep testing method has great potential for testing and understanding creep behaviour of HAZs with microstructural gradients. A similar study93 compared the HAZ of the base metal and the HAZ of the weld metal with specimens extracted from the weld metal and base material. The HAZ of the base metal showed a pronounced decrease in life as compared to other examined constituents. These results offer an intriguing possibility in the examination of the comparative life between HAZ regions by testing small specimens extracted from real weldments.
Mitigation of type IV cracking
Creep strength enhanced ferritic steel welds have a very complex HAZ that results in a decrease in creep rupture strength and failure of the weldment during service. This is a major concern for the utilities. Premature failure is due to the influence of the heterogeneity in the HAZ microstructure, a triaxial stress state in service and the accumulation of residual stress during welding. In particular, there is a concentration of stress and coarsening of carbide in the ICHAZ and FGHAZ, leading to nucleation and growth of voids that lead to type IV cracking and premature failure. It is yet to be determined whether the crack initiates in the FGHAZ or ICHAZ. An example of a type IV failure in a 9 wt-%Cr alloy (in this instance E911) is shown in Fig. 18.
An approach that could eliminate type IV cracking in these alloys and make them very useful would be to subject the whole weldment after welding to a normalisation and tempering treatment and to eliminate the complex HAZ microstructures and stresses. However, that approach would be impractical and costly. Therefore, a number of investigators have focused on increasing the cracking resistance of the alloys.94, 97, 99– 101 The methodologies to address resistance to cracking have been summarised by Laha et al. 87 The cracking resistance of the steel has been found to increase by alloying and modifying the steel with C and B and by altering the carbide precipitation and stability. In addition, efforts have been made to alter the stress state of the weldment by joint design, weld process and varying HAZ width by changing the heat input.94 Several investigators102– 104 have studied the effect of welding parameters on type IV cracking. Both joint geometry and preheat have been found to improve creep life significantly.
Boron has been found to be an effective agent in increasing the cracking resistance of the alloy. Kondo et al. 99 have investigated Cr steels (9Cr–3W–3Co–VNbB) with boron levels of 47 to 180 ppm and with low N levels. The N level should be controlled carefully so that BN does not form and reduce the amount of B available for stabilising the carbides. A creep rupture test revealed that a weld in high Cr steel containing B had superior properties compared with a weld in the same steel but without B. In addition, there was no evidence of type IV cracking in the steel. This improvement in properties and increase in cracking resistance have been found to be due to the absence of FGHAZ in weldments of B containing steel, unlike weldments in high Cr steel that does not contain B. In addition, B effectively suppresses the M23C6 carbide coarsening by replacing some of the C with B in the carbide.105, 106 Figure 22 shows a transmission electron microscopy (TEM) micrograph of normalised and tempered P(T)91 steel with and without the addition of B.107 The steel without B has a much coarser M23C6 carbide precipitate than the steel added B. The C content of the alloy has been found to influence its creep properties and its susceptibility to type IV cracking.
Images (TEM) of normalised and tempered P(T)91 steel a without B and b with B107 (with permission of Springer Science)
Hirata et al. 108 have found a correlation between C level and the resistance of the steel to cracking. They investigated 9Cr–3Co–3W–VNb alloy with C ranging from 0·005 wt-% to 0·1 wt-%. As the C content decreased, the creep strength of both base metal and the simulated HAZ increased. The normal decrease in HAZ creep strength was eliminated. The increase in resistance to cracking has been attributed to the decrease in coarsening rate of carbide and the volume fraction of carbide as the carbon level decreases; longer term tests are still needed to definitively prove this claim.
An approach not explored actively to eliminate type IV cracking is to alter the heat treatment schedule rather than change composition. Abe et al. 98 have found that Gr 92N subjected to normalising treatment and no tempering had the same creep rupture life as the base metal. However it is not very conclusive if this, and absence of type IV cracking is due to a change in the heat treatment procedure or composition since Gr 92N contained nitrogen.
Dissimilar metal welds
Austenitic to ferritic and ferritic to ferritic steel transition joints are often used in fossil fired power plants. In conventional boiler designs, austenitic steels are used in the final stages of superheaters and reheater pendants, where the temperature is high and where conventional ferritic and CSEF materials would not last the anticipated design life. Hence, there is a need to have dissimilar metal weld (DMW) joints between austenitic and ferritic steels. Some practical insight into such joints is available in the literature.48, 70 Failures in the power plants involving dissimilar metal joints brought about an increase in the research activities in the 1950s.109 The failure of the joints was attributed to differences in the coefficient of thermal expansion between the two steels, namely, 2¼Cr–1Mo ferritic steel and 316 austenitic stainless steel, and accelerated creep in a carbon depleted zone due to carbon migration in the ferritic matrix adjacent to the interface. Exhaustive studies have been conducted to understand and provide better engineering guidance to increase the performance of austenitic to ferritic DMWs.110– 113 Recent studies of DMWs between P(T)91 and alloy 800 using INCONEL 182 filler metal have shown P(T)91 to be susceptible to a similar failure mechanism observed in P(T)22 weldments using nickel base filler materials.114
The earliest elastic stress analysis in the joint geometry was performed jointly by General Electric Company and Oak Ridge National Laboratory in the mid-1970s.115 Their analysis showed that introducing a piece of transition metal (alloy 800H), whose coefficient of thermal expansion is between those of 2¼Cr–1Mo steel and 316 austenitic steel, reduced the stresses significantly. Further significant work in this area is described in the literature.116, 117
When dissimilar joints are made between P(T)91, P(T)11 and P(T)22 steels (ferritic to ferritic), stresses developed at the joint interface are low because their coefficients of thermal expansion are low and their thermal conductivities are high. When exposed to PWHT or service temperatures, such joints undergo significant redistribution of interstitial C and N and modification of the microstructure.118, 119 Carbon migrates from the low Cr Cr–Mo steel to the high Cr Cr–Mo steel. Such a migration of C leads to a C depleted zone, a soft zone characterised by low volume fraction of carbide precipitates in the low Cr steel. Similarly, on the high Cr steel side, a C enriched zone develops, which is a hard zone characterised by a higher volume fraction of carbide precipitates. The migration of C is driven by a gradient in chemical potential of C between the two alloys.120, 121 The low carbon zone tends to dissolve carbides and therefore experiences no grain boundary pinning effects. This leads to recrystallisation of grains in the region, the development and concentration of a triaxial stress state and accumulation of deformation, resulting in premature failure of the weldment.122, 123 The evaluation of cross-weld CSEF DMWs made between P(T)91 and P(T)23 in creep has demonstrated a similar propensity for premature failure due to carbon migration.124, 125
Recent efforts in DMW research has focused on mitigating carbon diffusion and the formation of hard and soft zones in dissimilar ferritic/ferritic joints. One approach is to have a nickel based diffusion barrier between 2¼Cr–1Mo and 9Cr–1Mo steels.119 The presence of such a diffusion barrier has been found to reduce the propensity to form hard and soft zones. A numerical analysis to model and simulate the formation of soft and hard zones in the DMW joint with a diffusion barrier has been developed and has been used to optimise the thickness of the diffusion barrier interlayer.
Another approach has led to the development of HFS6, a new nickel based filler metal with similar thermal expansion properties to the low alloy ferritic steel and lower Cr content to avoid carbon migration.126 The alloy was never used because of microfissuring problems. After further development and extensive characterisation, a new filler metal electrode was developed and has been commercialised as filler metal EPRI P87(Ni–0·01C–0·3Si–1·5Mn–9·0Cr–2·0Mo–1·0Nb–38Fe–0·008S–0·008P, in wt-%). EPRI P87 has demonstrated excellent elevated temperature properties and promises an improvement in cross-weld rupture life between CSEF steels and austenitic stainless steels. No tendency for C migration or detrimental carbide formation has been documented when EPRI P87 was welded against a 9 wt-%Cr CSEF steel and austenitic stainless steel.127, 128
Recently, a novel concept of graded transition material (a functionally graded steel whose composition and coefficient of thermal expansion change gradually from the ferritic to austenitic) has been proposed.129 It could replace the traditional transition piece of alloy 800. Current DMW practice mostly involves buttering one end of the base metal with a suitable filler metal, PWHT, and joining to the other.
Weld repair
To meet power demands and to keep costs low, utilities would like to keep the time required for the maintenance and forced outages due to repair to a minimum. To address this need, utilities require a sound and well engineered repair strategy.130 Such an approach could extend the operating life of a power plant and could reduce cost and the replacement of damaged components. In the USA, most of the power plants are >25 years old, and a survey of utilities by EPRI found that the utilities have no standard guidelines for repair welding procedures for extending the life of aging power plants.131 Electric Power Research Institute's survey of power plant equipment manufacturers, power producers and users revealed that utilities were experiencing repeat cracking at locations where repairs had been made, revealing an urgent need for improved repair welding practices. Based on these surveys, EPRI is developing procedures and guidelines for performing reliable weld repair of fossil fired power plant components fabricated from CSEF steels.
P(T)91 is the steel of choice for heavy sections such as pipes and headers in the advanced fossil fired power plants today. It has a martensitic microstructure with high hardness and requires tempering or PWHT, during which the hardness of the steel decreases gradually.131 The PWHT operation is expensive and time consuming, and repeated PWHT thermal cycles (such as in weld repair) may have a negative effect on the life of a given component. Utilities recognise the potential benefit in a developed, suitable repair technology that would eliminate PWHT after repair in CSEF materials such as P(T)91. It is known that when ferritic steels used in fossil fired power plants are exposed to the weld thermal cycle, the steel generates a range of microstructures in the HAZ (Fig. 1). The coarse grained region may be susceptible to cracking (e.g. hydrogen cracking, reheat cracking) soon after welding; some CSEF materials are more susceptible to these mechanisms. The coarse grained region also has high hardenability and will likely form martensite upon cooling. Any weld repair process that can temper and/or eliminate the coarse grained region is highly desirable to reduce risk to the previously mentioned cracking mechanisms.
The repair technique developed and most commonly used is the temper bead welding technique, also known as cold welding. The use of temper bead welding in fossil fired power plants has its roots in materials highly susceptible to reheat cracking, such as CrMoV steels and in low alloy Cr–Mo steels.132, 133 The process has been added for qualification to section IX of the ASME Boiler and Pressure Vessel Code. In the temper bead process, weld metal is placed in a location to alter the properties of the previous weld bead (see Fig. 23a).134 Once the top layer has been deposited, a second layer is deposited, and the heat from the second layer penetrates into the layer below it, where it tempers the weld bead in the first layer and the HAZ (Fig. 23b). Sometimes tempering of the first layer is improved by grinding off part of the first layer bead and its HAZ before the second layer is applied. This process is called the half bead tempering technique.
a remelting of weld metal and HAZ of first bead due to thermal effects of second bead; b effect of second layer of weld metal on first layer134
Bhaduri et al. 135 have investigated two variations of the temper bead process using artificially aged 2·25Cr–1Mo and 9Cr–1Mo steels. The process variation involves the half bead temper bead and butter bead temper bead techniques. In butter bead temper bead technique, the heat input to the weld is altered by suitably changing the size of the electrode. Repair welds were made on artificially aged specimens; the SMAW process and matching filler metal were used according to the ASME Boiler and Pressure Vessel Code. Of the two processes, the butter bead temper bead process was found to be more suitable for both 2·25Cr–1Mo and 9Cr–1Mo steels.
Significant advances, such as variations to the temper bead procedures, narrow groove welding and high deposition rate processes such as FCAW,136 are being made to develop other, more effective welding and weld repair technologies. In the past decade, a number of investigators have examined the potential of using a temper bead weld repair for grade 91 components.137– 140
To eliminate PWHT after welding grade T91 steel, Siefert et al. 141 have investigated two temper bead repair welding processes, the consistent layer process and the controlled deposition process, in which EPRI P87 nickel base filler metal was used in tubing applications. In the consistent layer process, the heat energy dissipation is controlled to develop a tempered martensitic structure in the first few millimetres of the HAZ. In the controlled deposition process, the heat input is varied by controlling the electrode size. Of the two, the consistent layer process was found to be more effective in tempering the repair welds. P87 was found to be beneficial in a number of ways over conventional nickel base materials (ERNiFe-2 and ERNr Cr-3), namely in that it prevents the formation of detrimental type I carbides in the CSEF steel at the fusion line of the weld. The application of such welding techniques was specifically examined and reported for tubing applications; because the potential for premature failure of the repair is internal to the furnace, it has far less risk in causing injuries to plant personnel.
Continuing research at EPRI has been initiated in examining the potential of applying temper bead welding techniques to thick section grade 91 components using a variety of filler materials and techniques (Table 8).32 The application of a temper bead weld repair may pose fewer issues in martensitic steels than with the implementation of similar techniques in legacy bainitic Cr–Mo steels or CrMoV steels because martensitic steels have been experimentally shown to accumulate less residual stress during welding. This was observed in experiments employing the Satoh test (see Fig. 24).142, 143. The Satoh test is a simple constrained bar test, whereby the material of interest is heated to various peak temperatures or profiles consistent with manufacturing processes (such as welding, normalising, or tempering). Upon cooling, the thermal contraction leads to an accumulation of stress, which can be plotted as a function of temperature. As can be seen in Fig. 24, transformation plasticity during the cooling of a uniaxially constrained sample from the austenite phase field acts to relieve the build-up of thermal stress as the sample cools. Residual stresses were observed to accumulate only after transformation was completed and the specimens approached ambient temperature.144– 146
Key aspects of variables considered in evaluation of repair techniques for grade 91 piping and components32
Thermanit is a registered trademark of Bohler Thyssen Schweisstechnik Deutschland.
INCO-WELD is a registered trademark of the Special Metals Corporation family of companies.
The letter after each temper bead procedure signifies a specific welding procedure, detailed below it.
Service exposed grade 91 (‘A’).
Service exposed grade 91 (‘A’) renormalised and tempered Grade 91 (‘b’).
By contrast, the non-transforming austenitic steel exhibited a continuous increase in residual stress with decreasing temperature, as might be expected from the thermal contraction of a constrained sample. This observation is profound in that it may be entirely possible to utilise a temper bead procedure for grade 91, and it may be the case that martensitic materials are more accepting of such procedures than their bainitic counterparts. This observation may account for some of the issues reported in bainitic CSEF materials T/P23 and T/P24, which were not of concern during the development and widespread introduction of T/P91 and T/P92.
Outlook
Coal will continue to be a major source of energy in the world. Coal fired power plant steam conditions have successfully evolved from subcritical to USC and continue to extend into the AUSC regime. The prognosis of continued improvement in plant efficiency and a reduced carbon footprint looks promising as significant advances are being made in a number of technical disciplines, such as alloy design, microstructure–property relationships, fabrication techniques (welding in particular) and welding power supplies, sensors, and controls. New alloy design methodologies can be used to engineer, design and develop advanced structural alloys and welding consumables for boilers and boiler components. Ideally, these alloys should be easily weldable and consistently reproducible. As alloys with better high temperature creep properties, corrosion resistance and other thermophysical properties are developed, welding of such advanced alloys inherently may become more challenging. Therefore, attention should be paid to the weldability issues early in the alloy design stage of future materials. Such attention to weldability is well placed because a boiler requires tens of thousands of welds as it progresses from construction in a fabrication facility to field erection and commissioning.
In the fabrication of boilers and other components, greater understanding is needed of the metallurgy of materials subjected to the weld thermal cycle. An immediate need is to better characterise existing complex CSEF steels gaining traction in industry (specifically, grades 23, 24 and 92). Such characterisation, as it relates to material behaviour, the evolution of microstructure and the microstructure–property relationship that develops when the alloys are welded, will directly enable more precise life assessment and more applicable weld repair procedures. This undertaking must be enhanced and further refined using applicable computational modelling and realistic weld simulation studies with proper material property input and understanding of the residual stress states induced by welding. In this regard, future studies will involve an integration of materials, welding processes, microstructure, the microstructure–property relationship, sensors and controls and their models to develop reliable materials and welds (Fig. 25).147
Integration of process, microstructure, properties, process control and automation and their models147
Current high deposition rate welding processes used in the construction of boilers are SAW and SMAW. Other available processes that show promise encompass advanced arc welding processes such as FCAW, narrow groove arc welding, hot wire GTAW to high energy density welding processes such as reduced pressure electron beam welding and laser welding. The feasibility and benefits of these processes need to be evaluated on both a material performance basis and a cost basis. Other advanced processes such as FSW and hybrid (laser gas metal arc) welding have the potential to improve both productivity and quality of the welds. In addition, they are low heat input processes, which have the advantages such as the reduction in residual stresses and distortion and the potential elimination of problems such as type IV cracking.
Welding transition joints between dissimilar alloys pose an unavoidable challenge in the construction of current and future coal fired power plant boilers. The design and implementation of new materials that will be subjected to DMWs must be thoroughly evaluated and understood so that costly outages from unanticipated issues in material compatibility can be avoided. For this issue to be addressed, existing filler materials used for similar applications in other industries will need to be evaluated, new filler metals may need to be designed, and new joining processes may have to be investigated to ensure acceptable performance of the DMWs. The new concept of a graded composition transition joint has the potential to replace the current technology and to mitigate problems such as thermal fatigue due to carbon migration and mismatches in coefficients of thermal expansion.
With the worldwide power grid transforming at a rapid pace, it is inevitable that current fossil fired power plants will have to move from a baseload operation to a cyclic operation to accommodate the addition of renewable energy sources. Thus, higher strength materials will be required for thinner sections. The emphasis will be on components and welds being able to withstand thermal fatigue and other damage mechanisms specific to cycling conditions. This realisation may alter the way in which materials are designed (currently strength dominated) to a focus on other high temperature material requirements, such as fatigue and ductility.
Regardless of the impetus for new material development, weldability assessment or weld process development, future research in CSEF steels must provide adequate understanding to develop well engineered solutions. It is inevitable that as the complexity of the base material, filler material or welding process increases, the amount of data needed to adequately assess a given material or process multiplies dramatically. This observation was not fully appreciated before the implementation of longitudinally seam welded piping, application of T23 and T24 in waterwall panels, the realisation of improper PWHT cycles and the lack of detailed education for fabricators, erectors and end users regarding these complex materials. Well engineered solutions demand a thorough understanding of past failures and limitations as well as an intimate interaction between researchers and industry to understand and ultimately solve today's most pressing challenges. The unfortunate reality is that in the forum of CSEF steels, there are countless examples where this has not been the case. As such, the opportunity for research surrounding these materials remains fluid, extremely intriguing and ripe with opportunity.
Footnotes
Acknowledgements
The authors wish to acknowledge Dr J. Chen (Oak Ridge National Laboratory) for his valuable time and assistance in preparing this manuscript. We thank Dr W. F. Newell, Euroweld, Newell associates and K. Coleman (EPRI) for valuable comments. We also would like to thank Professors H. K. D. H. Bhadeshia (Cambridge University), T. DebRoy (Pennsylvania State University) and Dr J. Shingledecker (EPRI) for their review and valuable comments. Research is sponsored by the Office of Fossil Energy Advanced Research Materials Program, US Department of Energy, under contract DE-AC05-00OR22725 with UT-Battelle, LLC.