Failures in boilers and other equipment taking place in power plants that use steam as the working fluid are discussed in this article. The discussion is mainly concerned with failures in Rankine cycle systems that use fossil fuels as the primary heat source. The general procedure and techniques followed in failure investigation of boilers and related equipment are discussed. The article is framed with an objective to provide systematic information on various damage mechanisms leading to the failure of boiler tubes, headers, and drums, supplemented by representative case studies for a greater understanding of the respective damage mechanism.

The phenomenon of on-load corrosion is directly associated with the production of magnetite on the water-side surface of boiler tubes. On-load corrosion may first be manifested by the sudden, violent rupture of a boiler tube, such failures being found to occur predominantly on the fire-side surface of tubes situated in zones exposed to radiant heat where high rates of heat transfer pertain. In most instances, a large number of adjacent tubes are found to have suffered, the affected zone frequently extending in a horizontal band across the boiler. In some instances, pronounced local attack has taken place at butt welds in water-wall tubes, particularly those situated in zones of high heat flux. To prevent on-load corrosion an adequate flow of water must occur within the tubes in the susceptible regions of a boiler. Corrosion products and suspended matter from the pre-boiler equipment should be prevented from entering the boiler itself. Also, it is good practice to reduce as far as possible the intrusion of weld flash and other impedances to smooth flow within the boiler tubes.

During microscopic examination of a number of cases of caustic cracking, a certain feature has been recognized that appeared to be associated only with caustic cracking. This was a preferential attack on the carbide envelopes and lamellae of the pearlite grains. Evidence suggests that the intergranular path of caustic cracks in steam boilers may be due largely to the presence, probably on a submicroscopic scale, of carbides at the grain boundaries, thus rendering these regions susceptible to preferential attack. It is known that some steels are more liable to develop caustic cracks than others, although their microstructures may not show any significant differences, and it seems probable that this behavior may be related to the amount and continuity of the grain-boundary carbides.

This article explains the main types and characteristic causes of failures in boilers and other equipment in stationary and marine power plants that use steam as the working fluid with examples. It focuses on the distinctive features of each type that enable the failure analyst to determine the cause and suggest corrective action. The causes of failures include tube rupture, corrosion or scaling, fatigue, erosion, and stress-corrosion cracking. The article also describes the procedures for conducting a failure analysis.

Rivets from the longitudinal seam of the terminal shell ring of a 12 year old Lancashire boiler broke off easily during examination. Cleavage fractures indicated a brittle material. Microstructure of a sectioned rivet head was typical of a normal rimming steel except the ferrite crystals contained numerous nitride needles. Their existence indicated an abnormally high nitrogen content. If such a steel is heated for a lengthy period to a temperature of that prevailing in a boiler, precipitation of the nitrides may be expected, with consequent embrittlement. In this case, embrittlement of this type was the primary cause of the breaking off of the type rivet heads. Nothing was observed in the course of the examination that suggested caustic cracking.

One tube in a watertube boiler developed leakage from a perforation. The external surface was covered with a dark deposit indicative of local fusion. Perforation resulted from the development of a crack from the internal surface. Microscopic examination revealed extensive intergranular penetration by molten copper. Particles of copper were seen in scale deposits on the bore of the tube. The tube in general showed a ferritic structure with partially spheroidized carbide. The fact that fusion of the copper had occurred indicated temperatures of 1100 deg C (2012 deg F) had been experienced locally, and the structural condition suggested that the tube in general had been heated at a lower temperature of the order of 600 deg C (1112 deg F) for some appreciable time. In this instance, overheating of the tube in the absence of the copper deposits may not have led to failure.

A backwell tube situated in the combustion chamber of a 100 atm boiler, which had been in service for many years, failed. The temperature of the saturated steam was about 300 deg C. Two pipe sections with attacked areas in the circumferential welding joint were examined for cause of failure. First section showed strong pit or trench-like attack in the welding seam on the inner surface. A bluish-black corrosion product adhered to the pits. The second section showed small blisters at the welding seam. The metallographic examination of the first section showed welding seam was strongly reduced in bulk from the inside and covered with a thick crumbling layer of magnetic iron oxide (Fe3-O4). This was a corrosion product resulting from the operation of the boiler. In addition, it was decarburized from the inside, and interspersed with grain boundary cracks. This form of attack is typical for the decarburization of steel by high-pressure hydrogen. Hence, the defects in the pipe sections were the result of scaling during the operation of the steam boiler. It was recommended to avoid unnecessary overheating during the welding of materials for high-pressure steam boiler operations.

Several ruptures took place in the front wall tubes of a water tube boiler. Some rupture samples showed ductile failure while others showed brittle failure. Specimens taken from the rupture where a thick edge had been produced, i.e., with little evidence of prior plastic deformation, showed a coarse microstructure indicative of gross overheating. The examination indicated that failure in the main resulted from gross overheating arising from water starvation as could have been due to a number of causes. The ruptures in some tubes were of the type commonly found in overheated tubes, the material being drawn out to a feather edge at the time of rupture. Other ruptures in the same and other tubes were of a more brittle type, this being associated with penetration of material by molten copper derived from scale.

A seamless hot-drawn boiler tube NW 300 of 318 mm OD and 9 mm wall thickness made of steel 15Mo3 was bent with sand filling after preheating allegedly to 1000 deg C. In the process it had cracked repeatedly in the drawn fiber. The composition corresponded to specifications, but exceptionally high copper content was noticeable. Microstructural examination showed the damage was due to overheating and burning during preheating and bending. Furthermore, crack formation was promoted by precipitation of metallic copper that had penetrated into the austenitic grain boundaries under the influence of tensile stresses that arose during bending. This phenomenon is known as “solder brittleness.”

A water tube boiler with two headers and 15.5 atm working pressure became leaky in the lower part due to the formation of cracks in the rivet-hole edges. The boiler plate of 20 mm thickness was a rimming steel with 0.05% C, traces of Si, 0.38% Mn, 0.027% P, 0.035% S, and 0.08% Cu. The mean value of the yield point was 24 (24) kg/sq mm, the tensile strength 39 (38) kg/sq mm, the elongation at fracture, d10, 26 (24)%, the necking at fracture 71 (66)% and notch impact value 11.5 (9.4) kgm/sq cm (the values in brackets are for the transverse direction). The specimen from inside surface of the boiler was polished and etched with Fry-solution, which revealed parallel striations formed due to the cold bending of the plate. The zones of slip were concentrated around the rivet holes. The cracks were formed here. The structure examination proved that the cracks had taken an exactly intercrystalline path, which is characteristic for caustic corrosion cracks. It was recommended that the internal stresses be removed through annealing or alternatively lye-resistant steel should be used.

During the inspection of a boiler containing cracks at the superheater header connection, cracking also was detected within the main steam drum. This was fabricated from a Mn-Mo-V low-alloy steel. It operated with water and saturated steam at approximately 335 deg C. Cracking was detected at the nozzles connecting the tubes for the entry of steam and hot water to the drum, at the downcomers, and at the connection to the safety valve. All cracks had a similar morphology, running in a longitudinal direction along the drum from the cutouts in the shell. All the cracks had developed under the influence of the hoop stress and were associated with the locally increased stress levels relating to the cutouts at nozzle and pipe connections. At their ends the cracks were filled with corrosion products, and their surfaces were seen to be very irregular. The process of crack growth was not due to fatigue only but can most probably be attributed to corrosion fatigue. The boiler steam drum design should be reviewed to reduce the local level of stress at the shell-nozzle connections.

The failure of a boiler operating at 540 °C and 9.4 MPa was investigated by examining material samples from the near-failure region and by thermodynamic analysis. A scanning Auger microprobe, SEM, and commercial thermodynamic software codes were used in the investigation. Results indicated that the boiler failure was caused by grain-boundary segregation of phosphorous, tin, and nitrogen and the in-service formation of carbide films and granules on the grain boundaries.

A 150 cm ID boiler drum made form ASTM A515, grade 70, steel failed during final hydrotesting at a pressure of approximately 26 MPa. Brittle fractures were revealed in between two SA-106C nozzles and remainder was found to involve tearing. Short, flat segments of fracture area, indicative of pre-existing cracks, were revealed by examination of the fracture surface at the drain grooves arc gouged at the nozzle sites. A thin layer of material with a dendritic structure was observed at the groove surface. The dendritic layer was revealed by qualitative microprobe analysis to contain over 1% C, higher than the carbon content of the base metal. The cracks in the drain groove surface could have occurred after arc gouging, during subsequent stress-relieving, or during the hydrostatic test. Flame cutting is not recommended for the type of steel used in the boiler drum because it can lead to local embrittlement and stress raisers, potentially initiating major failures.

This paper reviews several fatigue failures from the waterwall, superheater, and economizer portions of the boiler, their causes and how they were mitigated and monitored. Some cases required simple field modifications by cutting or welding, repair of existing controls, and/or changes in maintenance. Nondestructive inspections by visual, magnetic particle, ultrasonic, and radiographic methods for detecting and monitoring damage are discussed. These failures are presented to provide hindsight that will help others in increasing the success rate for anticipating and analyzing the remaining life of other units.

A stud that was part of a man hole cover on an old steam boiler broke when it was dropped. The boiler was more than 100 years old, but still performed satisfactorily when fired up at regular intervals. Chemical analysis showed the steel to be a low-carbon EN2 type. Microstructural examination indicated brittle fracture caused by strain aging. Replacement of all studs on the man hole cover and examination of the boiler steel in other places were recommended.

A type 304 stainless steel tube that failed in a boiler stack economizer was analyzed to determine the cause. The investigation consisted of visual, SEM/EDS, and metallographic analysis. Several degradation mechanisms appeared to be at work, including pitting corrosion, chloride stress corrosion cracking, and fatigue fracture. Investigators concluded that the primary failure mechanism was fatigue fracture, although either of the other mechanisms may have eventually caused the tube to fail in the absence of fatigue.

A desuperheater diffuser nozzle in the steam supply line failed within nine months of service in an 8.25 MN/sq m (1200 psig) steam line. The nozzle was an austenitic stainless steel casting in conformance to material. The nozzle had numerous cracks on the inside and outside surfaces, and the cracks had penetrated through the wall thickness in several areas. The fracture surfaces had distinct beach markings delineating the crack front, representative of crack propagation stages. The cracks were transgranular and, unlike classical corrosion-fatigue cracks, exhibited branching, characteristic of chloride-induced SCC in austenitic stainless steels. The failure resulted from chloride-induced SCC, possibly assisted by cyclic stress. The recommendation for alternate material for the desuperheater nozzle included nickel base alloys per ASTM B 564, Grades 600 or 800 titanium alloy per ASTM B 367, Grades C3/C4, or ferritic stainless steel alloy per ASTM 182, Grade FXM27.

On attempting to manipulate or bend a boiler tube some 22 ft. long, sudden failure occurred at what appeared to be a butt weld in the tube. Externally, the weld reinforcement had been ground flush and the entire tube surface painted. Internally, the appearance and width of the heated band suggested that the weld had been made by the oxy-gas process. A lack of root fusion over most of its length was evident. Examination of the fracture faces, which were of crystalline appearance indicative of brittle behavior, indicated incomplete fusion of the weld root. Microscopic examination showed the deposit to possess a large grain size with a low carbon content disposed as carbides along the grain boundaries, a feature which would provide an explanation of the brittle behavior. Subsequent inspection showed that this tube was one of several of the batch ordered for retubing of a boiler and which had a 2 ft. length welded to one end to make up the length.

Following leakage which developed within the furnace of a horizontal multi-tubular type boiler, examination revealed a series of cracks adjacent to the stiffening rings in the first plain furnace ring. The fire-side surface of the sample was coated with a layer of oxide scale. Microscopical examination of sections through the cracks showed them to be filled with oxide and to be of the multi-branched type, having blunt terminations. The general nature of the cracks was characteristic of cracking from thermal or corrosion fatigue, as results from the operation of varying stresses in an oxidizing or corrosive environment. The cracking in this particular case was due principally to the inordinately large gap between the components. Additionally, several of the sealing welds of the tubes to the back tube plate were cracked in a radial manner, and it would appear that in addition, abnormal thermal conditions may well have been experienced intermittently in service.

Carbon steel tubes from a boiler feedwater heater feeding a deaerator were treated to control scale formation, but the treatment instead produced more iron oxide. The additional iron oxide reduced the tubing to a totally corroded condition. Investigation showed that the chelate injected to control the scaling was added ahead of the preheater, where the boiler water still contained oxygen. As the chelate removed iron oxide, the O2 in the water continued to form more. Recommendations included moving the chelate addition to a point after the deaerator to stop the corrosion.