HEAT TREATMENT OF WELDED JOINTS Heat treatment is an operation that is both time consuming and costly. It can affect the strength and toughness of a welded joint, its corrosion resistance and the level of residual stress but is also a mandatory operation specified in many application codes and standards. In addition it is an essential variable in welding procedure qualification specifications. Before discussing the range of heat treatments that a metal may be subjected to, there is a need to clearly define what is meant by the various terms used to describe the range of heat treatments that may be applied to a welded joint. Such terms are often used incorrectly, particularly by non-specialists; for a metallurgist they have very precise meanings. Solution treatment Carried out at a high temperature and designed to take into a solution elements and compounds which are then retained in solution by cooling rapidly from the solution treatment temperature. This may be done to reduce the strength of the joint or to improve its corrosion resistance. With certain alloys it may be followed by a lower temperature heat treatment to reform the precipitates in a controlled manner (age or precipitation hardening). Annealing This consists of heating a metal to a high temperature, where recrystallisation and/or a phase transformation take place, and then cooling slowly, often in the heat treatment furnace. This is often carried out to soften the metal after it has been hardened, for example by cold working; a full anneal giving the very softest of microstructures. It also results in a reduction in both the yield and the tensile strength and, in the case of ferritic steels, usually a reduction in toughness. Normalising This is a heat treatment that is carried out only on ferritic steels. It comprises heating the steel to some 30-50°C above the upper transformation temperature (for a 0.20% carbon steel this would be around 910°C) and cooling in still air. This results in a reduction in grain size and improvements in both strength and toughness. Quenching This comprises a rapid cool from a high temperature. A ferritic steel would be heated to above the upper transformation temperature and quenched in water, oil or air blast to produce a very high strength, fine grained martensite. Steels are never used in the quenched condition, they are always tempered following the quenching operation. Tempering A heat treatment carried out on ferritic steels at a relatively low temperature, below the lower transformation temperature; in a conventional structural carbon steel this would be in the region of 600-650°C. It reduces hardness, lowers the tensile strength and improves ductility and toughness. Most normalised steels are tempered before welding, all quenched steels are used in the quenched and tempered condition. Ageing or Precipitation hardening A low temperature heat treatment designed to produce the correct size and distribution of precipitates, thereby increasing the yield and tensile strength. It is generally preceded by a solution heat treatment. For steel, the temperature may be somewhere between 450-740 degree C, an aluminium alloy would be aged at between 100-200°C. Longer times and/or higher temperatures result in an increase in size of the precipitate and a reduction in both hardness and strength. Stress relief As the name suggests, this is a heat treatment designed to reduce the residual stresses produced by weld shrinkage. It relies upon the fact that, as the temperature of the metal is raised, the yield strength decreases, allowing the residual stresses to be redistributed by creep of the weld and parent metal. Cooling from the stress relief temperature is controlled in order that no harmful thermal gradients can occur. Post heat A low temperature heat treatment carried out immediately on completion of welding by increasing the preheat by some 100°C and maintaining this temperature for 3 or 4 hours. This assists the diffusion of any hydrogen in the weld or heat affected zones out of the joint and reduces the risk of hydrogen induced cold cracking. It is used only on ferritic steels, where hydrogen cold cracking is a major concern i.e. very crack sensitive steels, very thick joints etc. Post Weld Heat Treatment (PWHT) So what does the term 'post weld heat treatment' mean? To some engineers it is a rather vague term that is used to describe any heat treatment that is carried out when welding is complete. To others however, particularly those working in accordance with the pressure vessel codes such as BS PD 5500, EN 13445 or ASME VIII, it has a very precise meaning. When an engineer talks of post weld heat treatment, annealing, tempering or stress relief it is therefore advisable. Heat treatment following welding may be carried out for one or more of three fundamental reasons: to achieve dimensional stability in order to maintain tolerances during machining operations or during shake-down in service to produce specific metallurgical structures in order to achieve the required mechanical properties to reduce the risk of in-service problems such as stress corrosion or brittle fracture by reducing the residual stress in the welded component The range of heat treatments to achieve one or more of these three objectives in the range of ferrous and non-ferrous metals and alloys that may be welded is obviously far too extensive to cover in great detail within these brief Job Knowledge articles. The emphasis in the following section will be on the PWHT of carbon and low alloy steels as required by the application standards although brief mention will be made of other forms of heat treatment that the welding engineer may encounter in the ferrous alloys. There are two basic mechanisms that are involved, firstly stress relief and secondly microstructural modifications or tempering. Stress Relief Why is it necessary to perform stress relief? It is an expensive operation requiring part or all of the welded item to be heated to a high temperature and it may cause undesirable metallurgical changes in some alloys. As mentioned above there may be one or more reasons. The high residual stresses locked into a welded joint may cause deformation outside acceptable dimensions to occur when the item is machined or when it enters service. High residual stresses in carbon and low alloy steels can increase the risk of brittle fracture by providing a driving force for crack propagation. Residual stresses will cause stress corrosion cracking to occur in the correct environment eg carbon and low alloy steels in caustic service or stainless steel exposed to chlorides. What causes these high residual stresses? Welding involves the deposition of molten metal between two essentially cold parent metal faces. As the joint cools the weld metal contracts but is restrained by the cold metal on either side; the residual stress in the joint therefore increases as the temperature falls. When the stress has reached a sufficiently high value (the yield point or proof strength at that temperature) the metal plastically deforms by means of a creep mechanism so that the stress in the joint matches the yield strength. As the temperature continues to fall the yield strength increases, impeding deformation, so that at ambient temperature the residual stress is often equal to the proof strength (Fig 1). To reduce this high level of residual stress, the component is reheated to a sufficiently high temperature. As the temperature is increased the proof strength falls, allowing deformation to occur and residual stress to decrease until an acceptable level is reached. The component would be held at this temperature (soaked) for a period of time until a stable condition is reached and then cooled back to room temperature. The residual stress remaining in the joint is equal to the proof strength at the soak temperature. Figure 1 shows that residual stress in a carbon manganese steel falls reasonably steadily from ambient to around 600 degree C but that the high strength creep resistant steels need to be above 400 degree C before the residual stress begins to fall. Stainless steel is hardly affected until the temperature exceeds 500 degree C. There is therefore a range of soak temperatures for the various alloys to achieve an acceptable reduction in residual stress without adversely affecting the mechanical properties of the joint. In carbon manganese steels this temperature will be between 550-620 degree C, in creep resistant steels somewhere between 650-750 degree C and for stainless steels between 800-850 degree C. Tempering Tempering is a heat treatment that is only relevant to steels and is carried out to soften any hard micro-structures that may have formed during previous heat treatments, improving ductility and toughness. Tempering also enables precipitates to form and for the size of these to be controlled to provide the required mechanical properties. This is particularly important for the creep resistant chromium-molybdenum steels. Tempering comprises heating the steel to a temperature below the lower critical temperature; this temperature being affected by any alloying elements that have been added to the steel so that for a carbon-manganese steel, the temperature is around 650°C, for a 2¼CrMo steel, 760°C . Quenched steels are always tempered. Normalised steels are also usually supplied in the tempered condition although occasionally low carbon carbon-manganese steel may be welded in the normalised condition only, the tempering being achieved during PWHT. Annealed steels are not supplied in the tempered condition. Tempering of tool steels may be performed at temperatures as low as 150 degrees C, but with the constructional steels that are the concern of the welding engineer the tempering temperature is generally somewhere between 550- 760°C, depending on the composition of the steel. Post Weld Heat Treatment (PWHT) As mentioned, PWHT is a specific term that encompasses both stress relief and tempering and is not to be confused with heat treatments after welding. Such treatments may comprise ageing of aluminium alloys, solution treatment of austenitic stainless steel, hydrogen release etc. PWHT is a mandatory requirement in many codes and specifications when certain criteria are met. It reduces the risk of brittle fracture by reducing the residual stress and improving toughness and reduces the risk of stress corrosion cracking. It has, however, little beneficial effect on fatigue performance unless the stresses are mostly compressive. It is an essential variable in all of the welding procedure qualification specifications such as ISO 15614 Part 1 and ASME IX. Addition or deletion of PWHT or heat treatment outside the qualified time and/or temperature ranges require a requalification of the welding procedures. PWHT temperatures for welds made in accordance with the requirements of EN 13445, ASME VIII and BS PD 5500 are given below in Table 1. Table 1: PWHT Temperatures from Pressure Vessel Specifications Steel Grade C Steel BS EN 13445 ASME VIII Temp range °C Normal holding temp °C Temp range °C 550600 593 580-620 BS PD 5500 Steel Grade BS EN 13445 ASME VIII BS PD 5500 C 1/2 Mo 550620 593 630-670 1Cr 1/2 Mo 630680 593 630-700 2 1/4 Cr/Mo 670720 677 630-750 5CrMo 700750 677 710-750 3 1/2 Ni 530580 593 580-620 Note from Table 1 that ASME VIII specifies a minimum holding temperature and not a temperature range as in the BS and EN specifications. As mentioned above, PWHT is a mandatory requirement when certain criteria are met, the main one being the thickness. BS EN 13445 and BSPD 5500 require that joints over 35mm thick are PWHT’d, ASME VII above 19mm. If, however, the vessel is to enter service where stress corrosion is a possibility, PWHT is mandatory, irrespective of thickness. The soak time is also dependant on thickness. As a very general rule this is one hour per 25mm of thickness; for accuracy, reference must be made to the relevant specification. These different requirements within the specifications mean that great care needs to be taken if a procedure qualification test is to be carried out that is intended to comply with more than one specification. A further important point is that the PWHT temperature should not be above that of the original tempering temperature as there is a risk of reducing the strength below the specified minimum for the steel. It is possible to PWHT above the tempering temperature only if mechanical testing is carried out to show that the steel has adequate mechanical properties. The testing should, obviously, be on the actual material in the new heat treatment condition. Maximum and minimum heating and cooling rates above 350-400°C are also specified in the application codes. Too fast a heating or cooling rate can result in unacceptable distortion due to unequal heating or cooling and, in very highly restrained components, may cause stress cracks to form during heating. Application of PWHT The method of PWHT depends on a number of factors; what equipment is available, what is the size and configuration of the component, what soaking temperature needs to be achieved, can the equipment provide uniform heating at the required heating rate? The best method is by using a furnace. This could be a permanent fixed furnace or a temporary furnace erected around the component, this latter being particularly useful for large unwieldy structures or to PWHT a large component on site. Permanent furnaces may be bogie loaded with a wheeled furnace bed on to which the component is placed or a top hat furnace that uses a fixed hearth and a removable cover. Typically, a furnace capable of heat treating a 150tonne pressure vessel would have dimensions of around 20m long, a door 5x5m and would consume around 900cu/metres of gas per hour. Furnaces can be heated using electricity, either resistance or induction heating, natural gas or oil. If using fossil fuels care should be taken to ensure that the fuel does not contain elements such as sulphur that may cause cracking problems with some alloys, particularly if these are austenitic steels or are nickel based – corrosion resistant cladding for example. Whichever fuel is used the furnace atmosphere should be closely controlled such that there is not excessive oxidation and scaling or carburisation due to unburnt carbon in the furnace atmosphere. If the furnace is gas or oil fired the flame must not be allowed to touch the component or the temperature monitoring thermocouples; this will result in either local overheating or a failure to reach PWHT temperature. Monitoring the temperature of the component during PWHT is essential. Most modern furnaces use zone control with thermocouples measuring and controlling the temperature of regions within the furnace, control being exercised automatically via computer software. Zone control is particularly useful to control the heating rates when PWHT’ing a component with different thicknesses of steel. It is not, however, recommended to use monitoring of the furnace temperature as proving the correct temperatures have been achieved in the component. Thermocouples are therefore generally attached to the surface of the component at specified intervals and it is these that are used to control the heating and cooling rates and the soak temperature automatically so that a uniform temperature is reached. There are no hard and fast rules concerning the number and disposition of thermocouples, each item needs to be separately assessed. As mentioned earlier, the yield strength reduces as the temperature rises and the component may be unable to support its own weight at the PWHT temperature. Excessive distortion is therefore a real possibility. It is essential that the component is adequately supported during heat treatment and trestles shaped to fit the component should be placed at regular intervals. The spacing of these will depend on the shape, diameter and thickness of the item. Internal supports may be required inside a cylinder such as a pressure vessel; if so, the supports should be of a similar material so that the coefficients of thermal expansion are matched. Whilst heat treating a pressure vessel in one operation in a furnace large enough to accommodate the entire vessel is the preferred method this is not always possible. In this case the pressure vessel application codes permit a completed vessel to be heat treated in sections in the furnace. It is necessary to overlap the heated regions – the width of the overlap is generally related to the vessel thickness. BS EN 13445 for instance specifies an overlap of 5√Re where R = inside diameter and e = thickness; ASME VIII specifies an overlap of 1.5 metres. It should be remembered that if this is done there will be a region in the vessel (which may contain welds) that will have experienced two cycles of PWHT and this needs to be taken into account in welding procedure qualification testing. There is also an area of concern, this being the region between the heated area within the furnace and the cold section outside the furnace. The temperature gradient must be controlled by adequately lagging the vessel with thermally insulating blankets and the requirements are given in the application codes. It is, of course, possible to assemble and PWHT a vessel in sections and then to carry out a local PWHT on the final closure seam. Local PWHT will be discussed in the next part of this series on heat treatment. When it is not possible to place the entire component in a furnace for heat treatment (because of the size of the fabrication, circumferential welds in a pipework system or when installing equipment on site, for example), then a local PWHT may be the only option. Local PWHT needs careful planning to ensure that heating and cooling rates are controlled and that an even and correct temperature is achieved. Uneven and/or rapid heating can give rise to harmful temperature gradients producing thermally induced stresses that exceed the yield stress. This may result in the development of new residual stresses when the component is cooled. Local PWHT may be carried out using high velocity gas burners, infra red burners, induction heating and high or low resistance heating elements. Electrical equipment is more easily installed and controlled than heating using natural gas or propane, particularly on site. High voltage resistance heating is rarely used on site due to the need for the radiant heaters to be positioned a set distance from the surface and, more significantly perhaps, the health and safety risks involved with the use of high voltage current. Low voltage electrical resistance heating and induction heating are the two most commonly used methods. High velocity gas burners are more advantageous when large areas need to be heat treated, particularly if, for example, firing can take place within a pressure vessel which then becomes its own furnace. For local PWHT of vessel circumferential seams internal insulating barriers can be used to localise the heat source. Motorised valves and micro-processor control of the combustion conditions enabled precise management of the heating cycle to be achieved. Low voltage electrical resistance heating uses flexible ceramic heating elements, colloquially known as corsets, an appropriate number being assembled to cover the area to be heat treated. Induction heating uses insulated cables that can be wrapped around the joint or shaped to fit the area to be heated or specially designed fitting for repetitive PWHT operations as illustrated in Fig 1. To perform the PWHT, temperature control thermocouples are firstly attached, often by capacitor discharge welding, the elements placed in position and the area then lagged with thermal insulating blankets to reduce heat loss and to maintain an acceptable temperature gradient. There are no standard terms used to describe the various regions within the locally PWHT'd area. In this article the terms 'soak band', 'heated band', 'gradient control band', 'temperature gradient', which may be axial and through thickness, and 'control zone' as suggested by the ASME will be used (see Fig 2). The soak band is the area that is heated to, within the specified PWHT temperature and time range. It comprises the weld, the two HAZs and part of the surrounding parent metal. The heated band is the area covered by the heating elements, the temperature at the edge of the heated band generally being required to be at least half that of the soak temperature. The temperature gradient control band is the region where thermal insulation, perhaps supplemented by additional heating elements, is applied to ensure that an acceptable axial temperature gradient is achieved from PWHT temperature to ambient. A control zone is the region where a number of heating elements are grouped together and controlled by a single thermocouple, enabling different regions to be heated independently; particularly useful with large diameter items or where there are variations in thickness. Temperature gradients may be axial (along the length of a pipe or vessel) and through thickness. The through thickness temperature gradient is caused by heat losses from the internal surface and is a function of both thickness and internal diameter, the larger the diameter, the greater the effect of radiation and convection losses. Both the width of the soak band and the temperature achieved can be substantially less than that on the outside of the pipe or tube. Insulation on the inner surface will reduce the temperature/width differential but may not be possible on small diameter tubes or pipework systems. This through thickness gradient is one of the reasons that specifications and codes require the soak or heated band to be a minimum width, generally related in some way to the thickness of the component. As mentioned above, there are rules in the application codes concerning the size of the heated area, normally related to the thickness. In a circular component such as a pipe butt weld or a pressure vessel circumferential seam the width of the band is easy to calculate. ASME VIII for instance requires the soak band width to be twice the thickness of the weld or 50.8mm either side of the weld, whichever is the lesser. ASME B31.3 requires the soak band width to be the weld width plus 25.4mm either side of the weld. BS EN 13445 does not specify a soak band width but instead specifies a heated band width of 5√Rt centred on the weld and where R = component inside radius and t = component thickness. There are no requirements in the ASME codes regarding heated band width. A very approximate rule of thumb for flat plate is that the heated band should be a minimum of twice the length of the weld although practical considerations may prevent achieving this ideal. There are no requirements, in any code or specification, on the width of the thermally insulated band although BS EN 13445 recommends 10√Re. It is essential that the relevant specification is referred to for specific guidance on what is required and it is worth remembering that the specification requirements on soak or heated band widths are minima and very little is lost by ensuring the specified dimensions are comfortably exceeded. What is an acceptable axial temperature gradient? Again, there is little advice in the codes and specifications. It is generally assumed that if the temperature at the edge of the heated band is above half that of the soak temperature then the temperature gradient will not be harmful. During heating and cooling BS EN 13445 specifies a maximum temperature difference of 150°C in 4500mm below 450°C (1°C in 3mm) and 1000C in 4500mm above 4500C (1°C in 4.5mm). To ensure that gradients and temperatures are controlled within acceptable limits sufficient thermocouples need to be attached to provide both temperature control and recording. For small diameter tubes, eg less than 100mm diameter, one control zone and one recording thermocouple are regarded as sufficient; between 100-200mm one control zone and one recording thermocouple at each of the 12 o’clock and 6 o’clock positions; above 250mm diameter one control zone and one recording thermocouple at each 900 quadrant, 12, 3, 6 and 9 o’clock, are suggested. These thermocouples should be placed on the centre line of the weld. Thermocouples will also be needed at the edge of the soak band and the edge of the heated band. Ideally, thermocouples should also be placed on the opposite surface to the heating elements to ensure that the correct through thickness temperature has been achieved although this is rarely possible on pipe systems. It is advisable to double up on the thermocouples to cope with the possibility of a thermocouple failure. Thermocouples use a hot and a cold junction to measure the temperature, the hot junction being attached to the component, the cold junction within the temperature recorder. For accurate temperature measurement the hot junction must obviously be at the temperature of the component. Errors can be introduced if the junction is not firmly attached, either by capacitor discharge (CD) welding, by mechanically fixing the wires to the component or by overheating of the thermocouple junction. CD welding of the thermocouple wires gives the most accurate results, particularly if the two wires are separated by 3-4mm. Mechanically attached wires will probably need to be insulated by covering the junction with heat resistant putty to prevent overheating of the thermocouple by the overlying heater. If the wire covering is stripped back then the bare wires also need to be insulated. It is advisable to specify the positions of the thermocouples on a drawing and to include these within a formal written heat treatment procedure document that covers both the specification and best practice requirements. Fig 1. Induction PWHT of Pipework Fig 2 Schematic of Temperature bands within a local PWHT (Reproduced with permission of the American Welding Society (AWS), Miami, Florida, USA) ABOUT POSTWELD HEAT TREATMENT/STRESS RELIEF HEAT TREATMENT Post Weld Heat Treatment (PWHT), or stress relief as it is sometimes known, is a method for reducing and redistributing the residual stresses in the material that have been introduced by welding. The extent of relaxation of the residual stresses depends on the material type and composition, the temperature of PWHT and the soaking time at that temperature. A commonly used guideline for PWHT is that the joint should be soaked at peak temperature for 1 hour for each 25mm (1 inch) of thickness, although for certain cases a minimum soak time will be specified. In addition to reduction and redistribution of residual stresses during the welding process, PWHT at higher temperatures permits some tempering, precipitation or ageing effects to occur. These metallurgical changes can reduce the hardness of the as-welded structure, improving ductility and reducing the risks of brittle fracture. In some steels, however, ageing/precipitation processes can cause deterioration in the mechanical properties of the steel, in which case, specialist advice should be taken on the appropriate times and temperatures to use. The necessity for PWHT depends on the material and the service requirements. Other factors that influence the need for PWHT are the welding parameters and the likely mechanism of failure. In some standards, PWHT is mandatory for certain grades or thicknesses, but where there is an option, cost and potential adverse effects need to be balanced against possible benefits. The energy costs are generally significant due to the high temperatures and long times involved, but costs associated with time delays may be more important. Detrimental effects include distortion, temper embrittlement, over-softening and reheat cracking, which means that control of heating and cooling rates, holding temperature tolerances and the times at temperature are extremely important, and must be carefully controlled in order to realise the full benefit of the process. Quenched and tempered (Q&T) steels have the PWHT temperature limited to below the original tempering temperature of the steel, as higher temperatures can change the microstructure of the base material from what was expected or required. AVOIDING PWHT - CAN IT BE JUSTIFIED? (MAY 2005) Current BSI and ASME codes for the construction of pressure vessels, boilers and piping specify that post-weld heat treatment is required if the thickness of the components being welded exceeds a specified value. This value depends on the type of material being used, and varies from code to code. An alternative procedure is available for deciding whether or not PWHT is necessary to avoid the risk of failure by fracture. This involves conducting a fracture mechanics assessment using procedures such as those in BSI 7910 (or API 579). The use of these procedures is permitted in the British pressure vessel standard BS PD 5500:2003. Alternative to code rules Welding thick walled components generates residual stresses that can be the cause of failure mechanisms such as brittle fracture and stress corrosion cracking. A criterion for PWHT based on a fracture mechanics assessment is more complicated than the code criterion of thickness alone. It may at first seem unlikely that designers, owners or certifying authorities would abandon the thickness-based criteria in favour of a more complicated approach. However, there are cases when PWHT is a code requirement but it may be considered unnecessary, excessively expensive, or practically impossible. In these cases, a fracture mechanics assessment may be used, subject to the agreement of the concerned parties. A fracture mechanics approach is based entirely on avoidance of failure by fracture or plastic collapse, Fig.1. Inspection engineers should also give consideration to the influence of heat treatment on avoiding other mechanisms such as fatigue and stress corrosion cracking, before adopting this approach. Fig. 1. Fracture mechanics assesses the combined effects of flaw size, material properties and stresses acting on the flaw Another Option: 'Patch' PWHT When a component is too large to be furnace heat treated, local heat treatment of a circumferential band is allowed. Both API 570 and AWS D10.10 allow for the possibility of local heat treatment band on components, subject to various precautions. (British codes specify the size of local 'heated band' and American codes specify the size of the 'soak band'). The ASME B&PV Code will allow the width of the patch to vary provided the resulting temperature gradients are not harmful. The severity of the resulting residual stresses due to temperature gradients from varying patches can be investigated using finite element analysis (FEA). The acceptability of any proposed variation in a soak band (eg. variable width, band shape, etc), can then be easily determined by evaluating the magnitude of the calculated residual stresses. That is (for a particular joint geometry), the magnitude of residual stress calculated by FEA, after modelling the code rules, versus the calculated stresses after modelling the proposed variation, Fig.2. Fig. 2. Residual stress after PWHT: full circumferential band (left) versus triangular patch (right) An FEA approach can be used at the time of original fabrication or in-service repair. The FEA should model the material properties in an elastic-plastic manner and allow stress relaxation during PWHT, as well as account for the initial stresses caused by welding. Experimental measurement of the residual stresses after patch PWHT will also be required to validate the results obtained from 'patch' PWHT. A fracture mechanics case study Fracture mechanics analysis is based on a consideration of the stresses acting at critical locations in a structure, the local geometry, the mechanical properties, the size of flaws (which may have escaped detection or been detected but left unrepaired after NDE), and the fracture toughness of the parent metal, weld metal and HAZ. A boiler operator wished to waive PWHT of a stub-to-header weld repair. The parent material was 2¼Cr½Mo steel. The header collects steam from the boiler superheater via several hundred separate boiler tubes, Fig.3. The tubes emerge vertically from the furnace and each tube is welded to a shorter length of tubing ('the antler'), which is welded to a short stub tube, which is welded directly to the header. The stub tubes are welded to the header in the fabrication works. Fig. 3. Boiler header and tube assembly PWHT of the stub-to-header welds (along with the header welds) in the fabrication works is not excessively difficult or expensive. The stub-to-antler welds can be made on site (or in the fabrication works) and do not normally require PWHT (even for 2¼Cr½Mo) because they are thin sections. However, PWHT of a single stub to header replacement (repair) during service is cumbersome. Such PWHT requires uniform heating of a cylindrical band around the header circumference. This is complicated by the many tubes protruding from the header, and the operation may cause an expensive delay to a repair outage. Assumed flaws To justify the boiler repair without PWHT, hypothetical flaws were assessed, Fig.4: Surface and embedded longitudinal flaws in the header at the weld toe; Surface and embedded longitudinal flaws in the stub at the weld toe; and Embedded transverse flaws around the stub-to-header weld. For each flaw location, the primary stresses (from pressure loading) were estimated in the stub at the weld toe, in the weld itself and in the header at the weld toe. The header and stub were 52mm (2in) and 10mm (0.4in) thick, respectively. Fig. 4. Locations of assumed flaws in boiler stub-to-header welds More than pressure stress Thermal stresses resulting from thermal expansion were also estimated. Thermal stresses arise during shutdown and a worst case temperature change of 100°C in one minute was assumed at the inside surface. The thermal stresses acting on the postulated flaws were actually found to be compressive. Their presence reduces the magnitude of the total stresses driving failure by fracture (results in a lower total crack driving force than that associated with applied and welding residual stresses alone). Since these stresses occur only as a result of a thermal shock, the assessment was performed assuming no thermal stresses. In accordance with the recommendations of BS7910, residual stresses in the aswelded condition (ie. after repair, without PWHT) were assumed to be uniform across the thickness as follows: For longitudinal flaws at the weld toe in the header or stub, the welding residual stress was assumed to be the lesser of the room temperature yield strengths of the weld or parent metal, ie. 275MPa (399ksi). For transverse flaws in the weld, the welding residual stress was assumed to be equal to the room temperature yield strength of the weld, ie. 370MPa (537ksi). NDE rationale and toughness assumptions Based on a review the boiler manufacturer's available data relevant to the parent metal, weld metal and HAZ, the fracture toughness expressed in terms of K Ic was assumed to be 3162N/mm3/2 (92ksi/in½ ) at the operating temperature. A full volumetric inspection of the stub-to-header weld repair was considered practically difficult in-situ, by both ultrasonic testing and radiography. The repair weld was only subject to visual inspection and magnetic particle examination which would readily discover surface-breaking flaws but not embedded flaws, Table 1. Table 1. Summary of NDE detection capabilities Flaw Type Versus NDE Method Surface Embedded Type Diagnosis Length Measurement Height Measurement Visual ✔ x Good ✔ x Radiography ✔ ✔ Good ✔ x Ultrasonics ✔ ✔ Poor ✔ ✔ Magnetic Particle ✔ x Good ✔ x Dye Penetrant ✔ x Good ✔ x Eddy Current ✔ ✔ Poor ✔ ✔ Potential Drop ✔ x Poor ✔ ✔ The boiler operator argued that a plausible embedded flaw height might be up to 12.5mm (½in). (This would be the height of a root flaw which could have extended as a hydrogen crack in the weld metal and HAZ, mainly below the header outer surface). Such a flaw may be either longitudinal or transverse with regard to the welding direction. Longitudinal flaws were considered to have a maximum length equal to the weld circumference, and transverse flaws would be limited to the width of the weld. Assessment results The results of the fracture mechanics assessment demonstrated that the assumed embedded flaws in the as-welded condition were acceptable, i.e. are non-critical in terms of fracture and plastic collapse. The 12mm (~½in) deep surface flaws assumed to exist in the header were also acceptable. The critical height of a surface flaw in the stub, with a length equal to the weld toe circumference, was found to be only 5.3mm (0.2in). If the minimum surface flaw height that can be reliably detected using visual or magnetic particle NDE is less than the tolerable height, say 3mm (~1/8in), then larger unacceptable flaws (height >5.3mm) can be detected by NDE and dealt with. Therefore, it maybe concluded that non-detectable surface flaws do not threaten the integrity of the stub repair in the as-welded condition. Based on the above and assuming that no other mechanisms (eg. creep-fatigue) may lead to extension of the original flaws, it was concluded that the weld repair was fit-forservice under operating loading in the as-welded condition. Financial justification It was shown that avoiding PWHT was technically justified. The cost of this fracture mechanics analysis was negligible in comparison with the total cost associated with carrying out PWHT on site which was cumbersome and expensive. The main advantage of the codes' thickness criterion is its simplicity. Unfortunately, it is not possible to use fracture mechanics to justify a general relaxation or rationalisation of the thickness criteria in the codes. The chances of making a successful case for avoidance of PWHT are best with a good knowledge of the fracture mechanics input parameters. Assumptions regarding fracture toughness, flaw sizes and applied stresses can be crucial to the outcome of the analysis.