J.C.Guild: Southern African Institute of Welding, Johannesburg, South Africa



Key Words: Failure Investigation, Boiler, Chiller Tube, Boiler Tube, High Temperature Failure, Steam Blanketing.





The investigation into the failure of a boiler chiller tube is described. The failure was by way of longitudinal bursting of the tube. It was ascribed to excessive temperatures being reached, and it is recommended that the boiler be modified in such a way that these tubes are no longer required.





Two chiller tubes run horizontally along each side of the boiler and their function is to chill slag and prevent it running into the grate side seals. The failure had occurred as a burst in the middle of the top tube on the west side of the boiler. At the time of failure the boiler was idling until the next steam blow down. Although, the boiler normally operates at 60 – 80 bar during this period, the pressure had dropped to below 60 bar and had then been raised to about 90 bar at the time of failure. This is still 12 bar below the normal working pressure.


A previous failure had occurred in the same tube in a similar position about a year earlier. At the time of the previous failure a repair had been made which involved the insertion of a replacement section of tubing approximately 1 metre in length. The failure being investigated in this paper, occurred in this section of replacement tubing.





The tube had failed by longitudinal bursting along virtually the whole length of the inserted piece of tubing. This can be seen in Figure 1.

Text Box:  
Figure 1: Illustrating the full length of the burst. Also visible is the corrosion track.


There was considerable deformation of the tube wall along the length of the fracture and substantial thinning of the fracture edges in the center portion of the split. This can be seen in Figure 2.


Text Box:  
Figure 2: Centre of split, displaying thinning of the wall and a corrosion track 40mm wide along top of tube.
The fracture had initiated approximately at the 2 o’clock position on the tube circumference. This position corresponded to the grate side of a light corrosion track, about 40mm wide, evident along the top of the tube. This is also visible in Figure 2. The bust had followed this corrosion track except at either end where it had slightly deviated from it. It was also evident that the corrosion track, which had been noted at the time of the previous failure, was considerably heavier in the original tubing on either side of the replacement section. The circumferential tearing of the tube, which had occurred at one of the butt welds of the inserted piece, is considered to be of a secondary nature to the longitudinal burst. It is thought that the weld metal was probably stronger than the tube metal and its resistance to fracture transferred the remaining bursting energy into the plane of the heat affected zone of the weld, resulting in the circumferential tearing. The fracture edge, itself, had a slightly laminated or stepped appearance. This can be seen in Figure 3.


Text Box:  
Figure 3: Note laminated appearance of fracture edge.
The general appearance of the inside of the tube was smooth and black although the black oxide coating had been lost in the center portion where the fracture was thought to have initiated from. The outside surface was generally clean and it was noticed that there was an indentation corresponding to the short cracks evident on the inside surface in the middle of the corrosion track at the position of initiation. This is visible in Figure 2. When the burst occurred it is thought that the tube hit part of the anchoring system causing this indentation.


As far as we have been able to ascertain, these chill tubes are listed in the manufacturer’s manual as being 76.2 mm Outside Diameter (OD) tubes in 7 swg. i.e. 4.470 mm wall thickness. The material is indicated elsewhere, as being carbon manganese steel to BS 3059 Grade 45S2 quality. The original and replacement sections of tubing were both 76.2 mm OD but it was evident that the original tubing was thicker than that used for the previous repair. Measurements of the thickness of the replacement tube gave results of 4.3 – 4.4 mm away from the fracture suggesting that the tube was, in fact, 7 swg. Although measurements in the corrosion track of the original tube gave wall thicknesses as low as 4.2 mm, elsewhere the thickness was in the range 5.5 – 6.3 mm. This indicates the original tubing was almost certainly a heavier gauge. Possibly 5 swg, (5.385 mm) or even thicker.


Samples of both the original and replacement tubes were submitted for chemical analysis. The results obtained are given in Table 1.



BS 3059 (1968) Gr 45

Original Tube




Result 1

Result 2



0.12 – 0.18





0.10 – 0.35





0.90 – 1.20





0.035 max





0.040 max







Not Detected







Table 1: Chemical composition of tube materials.


It can be seen from the above that although the replacement tube meets the chemical requirements of BS 3059, the original material does not. The results do however not obviously indicate that the original material was meant to conform to another specification.


Hardness test were carried out with the following results:


Original tube:               140 / 158 Hv (30 kg)

Replacement Tube:    151 / 155 Hv (30 kg)


These figures indicate both materials to have a tensile strength of the order of 50kg/mm2 which would meet the room temperature test requirements of BS 3059.


Microsections were taken from various positions on the fracture edge and also through the tube thickness at the end of the fracture.


A transverse section taken at the end of the fracture indicated no significant abnormalities. The outside surface was oxidized to a general depth of about 41m and there were oxide inroads up to a depth of 85m. The inside surface showed a thin oxide layer on the surface with some evidence of copper deposition in the oxide. Oxide inroads were also present to a depth of 92m on the inside surface.


A longitudinal section taken in line with the burst indicated that the steel cleanliness of the replacement section was relatively poor with numerous fine oxide stringers and elongated sulphides. Although the cleanliness of the steel is not considered a primary reason for failure it does explain the laminated appearance of the fracture edge and probably, also explains the deviation of the ends of the fracture from the corrosion track.


Sections remote from the fracture indicated both the original and replacement tubing to have very similar microstructures of fine grained pearlite and ferrite, typical of normalized carbon manganese steels. Sections from the thicker part of the fracture edge indicated no abnormalities with no significant difference in microstructure. However, sections from the thinnest portion of the fracture edges showed substantial elongation of the grains, decarburisation, and spheroidisation indicating that this had been the origin of failure and that the highest temperatures had been achieved in this position. There was no substantial evidence of creep deterioration, indicating that the failure had been a straightforward short term, high temperature, tensile rupture.


In view of the findings of the microstructural investigation, calculations were performed to indicate the temperatures that would have to be achieved to result in this type of failure. Since the original tube had not failed and thicknesses as low as 4.2mm had been measured in the corrosion track it was felt that 4mm was a reasonable figure to take as the minimum figure for the thickness of the replacement section at the time of failure. There is no evidence to suggest that the wall thickness could have been lower than this as even on the fracture edge, readings of 4.3mm were obtained and the depth of the corrosion track measured only 0.3mm. It was calculated that for a 4mm thick tube at 90 bar pressure a temperature in the range of 650 – 750°C must have been achieved prior to a tensile failure of the type envisaged.





Failure occurred by a short term, high temperature, tensile mechanism. Both the visual evidence and the calculations indicate that this was the case and that temperatures far in excess of the design temperature were achieved.


Although a thinner gauge of tubing was used for the replacement section in the previous repair this is not considered to be the cause of the failure. There is ample evidence to indicate that steam has been generated in the top of the tube (with the consequent danger of overheating) and this is considered the primary cause of the burst. Steam formation followed by subsequent wetting of the oxide results in repetitive spalling and growth of the protective magnetite layer on the inside of the tube surface thus leading to a loss of metal thickness. The corrosion track in the top of the tube indicates that this has, in fact, occurred. However, it is fairly evident that at the time of the incident the amount of metal thickness lost was not sufficient to have resulted in a failure of the type which occurred, even under full operating conditions. It is therefore fairly certain that the failure occurred due to an excessive rise in the tube wall temperature, just prior to the burst, and was caused by a steam blanketing effect. As the corrosion track appears to be present along the full length of the tube, this suggests that poor circulation is the probable cause of the steam blanketing rather than “hogging” of the tube. It also seems highly likely that the poor circulation is associated with abnormal operating conditions, such as running the boiler at reduced rates.


Since all these horizontal tubes appear to be at risk to failure by the same mechanism, we agree that the best solution would be to modify the boilers, such that they can be done away with. In fact, it is understood that subsequent boilers from the same manufacturer have already had this design change incorporated.