ANALYSIS OF THE FAILURE OF A T JOINT BETWEEN A PROCESS PIPELINE AND A RELIEF VALVE PIPELINE

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

Key Words: Failure Investigation, Pipeline, T joint, Compensation, Reformer effluent, Relief valves

ABSTRACT

The investigation into the failure of a T joint where a relief valve line was joined into a process line is described. Due to the service of the line, a number of different degradation mechanisms were possible. These had to be investigated before the final conclusion as to the failure mode could be reached. Recommendations to prevent future failures are given.

INTRODUCTION

A failure had occurred at the T joint between a 14 inch process line, containing reformer effluent, and a 6 inch relief valve line. The failure occurred in the early hours of the morning, when the plant was apparently running under normal operating conditions. A considerable fire ensued and a plant shutdown, of approximately eight days, was necessitated to carry out repairs resulting from the failure. A previous failure had occurred in the 6 inch line, some three months before the current failure being investigated. That failure had been attributed to corrosion by condensed steam, resulting in severe thinning of the pipe wall thickness. At that time the 6 inch line was replaced, and an inspection carried out on the 14 inch line. This inspection had indicated that no significant degradation of the 14 inch line had occurred.

INVESTIGATION

The 14 inch line carries reformer effluent (steam / hydrocarbon mixture) at approximately 33 bar and 370°C from a waste heat boiler. Just below the top outlet vertical bend, the horizontal 6 inch line is welded into the 14 inch line to make a T joint. Two relief valves are situated on this line approximately 750mm away from the T joint. The discharge lines from these relief valves run to the main vent. Originally there was only one relief valve in the 6 inch line, but a second was added as part of a general up-rating of the plant.

Examination of the failed sections of pipes indicated that fracture had initiated at the toe of the weld between the two different sized pipes. Direction of propagation had been at 45° into the 14 inch pipe material, and not along the heat affected zone contour. The 6 inch line had completely pulled out of the T junction. These features are shown in Figures 1 and 2.

The fracture faces indicated that failure had occurred by a ductile shear mechanism. Virtually the whole circumference had a similar 100% shear fracture face with the shear having occurred at 45° to the axis of the 6 inch line. At the bottom position of failure there were small areas of ‘woody’ fracture. At the top some splitting of the 14 inch line had occurred. This is shown in Figure 3, and it is believed to be the result of failure having occurred such that the top section tore away last resulting in a hinge effect there. The 14 inch pipe material was bulged outwards in the vicinity of failure, again indicating reasonable ductility in the fracture. There was no evidence of fatigue on any of the fracture faces.

The 14 inch line is believed to have been in service since the plant was commissioned and the appropriate line diagram indicated that it was constructed from ¼ inch plate. Wall thickness measurements indicated no significant loss of thickness in the vicinity of the failure where results of 6mm were obtained at fracture edges. The 6 inch line was believed to have been replaced some three months earlier, with six inch schedule 40 material and again no significant loss of wall thickness was detected. The line diagram indicated that this T joint should have a reinforcement pad on the 14 inch line. However, there was no evidence that this reinforcement had ever been adopted.

Examination of the inner surfaces of the pipes revealed a “tide” mark deposit in the bottom of the 6 inch line with a similar deposit in a corresponding position in the 14 inch line. This can be seen in Figure 4. Chemical analysis of these deposits indicated significant quantities of calcium and magnesium. There seems little doubt that water has been formed by condensation in the 6 inch line on the inlet side of the relief valves and that at some stage it has spilled or been drawn down into the 14 inch line. It should be noted, however, that the water was probably present in a film form rather than as a substantial slug. This is indicated by the wide angle between the deposits on either side of the bottom of the 6 inch pipe.

Materials of construction for the pipes and weld metal are specified as 1 ¼ CrMo steel. Metal analysis of samples gave the results in Table 1.

Sample Position	C	Ni	Cr	Mo	V
14 Inch Parent Metal	0.12	0.06	1.24	0.49	0.010
6 Inch Parent Metal	0.10	0.06	1.15	0.50	0.006
Weld Metal	0.07	0.05	1.26	0.54	0.010

Table 1: Table showing the chemical analysis of the materials.

The results given in table 1 are considered typical and satisfactory for these materials and indicate no abnormalities.

Sections were prepared for microscopic examination from several positions in the fracture region. These revealed no significant abnormalities but served to confirm the general assessment of the nature of the fracture that had been based on the visual and low power microscope examination of the fracture surfaces. No evidence was detected of hydrogen damage, creep or abnormal microstructures. The cleanness of the 14 inch line steel was satisfactory and although that of the 6 inch line steel was moderate / poor it is not considered to have been significant in terms of the failure. Examination of the fracture edges showed grain deformation confirming the ductile nature of the failure. This can be seen in Figure 5.

Hardness testing results are shown in Figure 6 and Table 2. No significant abnormalities were detected.

Pos’n	Hardness	Pos’n	Hardness
1	172	10	232
2	181	11	241
3	191	12	226
4	206	13	219
5	210	14	226
6	214	15	255
7	226	16	232
8	247	17	219
9	219

Table 2: Hardness test results.

When the 6 inch line had been replaced, the following “welding procedure” had apparently been adopted:

1) Preheat with oxy-acetylene torch to 100 – 150°C.

2) Weld using KV5 electrodes.

3) Postheat with oxy-acetylene torch.

The use of oxy-acetylene torch for post heating CrMo steels, or other steel for that matter, should be avoided since little or no control can be achieved in terms of heating rate, soaking temperature and cooling rate. In this case it was evident that no significant harm had resulted from the post heating operation but it should be noted that serious consequences can occur.

CONCLUSIONS AND RECOMMENDATIONS

The 45° slanted fracture face is typical of a ductile shear failure occurring as a result of the resolved stress of a tensile load applied predominantly along the longitudinal axis of the 6 inch pipeline.

The original design specification specified the need for reinforcement of the T joint and a study of the relevant piping codes confirm its requirement. This would seem to indicate that this junction has been utilized close to its ultimate limit for a considerable time. Apparently there was a front end pressure increase recorded as initiating approximately half an hour before the failure but this increase was only of the order of a few bar and could be considered to be within normal limits of operation. It should be noted, however, that until approximately two weeks before the failure, one of the relief valves was known to be passing. It was tightened down at the time and consequently the pressure required to activate either, or both of the valves, was substantially unknown. Possibly this was the first pressure increase that could have activated a valve. It is known that relief valve operation imposes a shock load to local pipe work and because of the “under design” of the joint this may have resulted in an almost instantaneous rupture.

It was suggested that water could have been present in the inlet side of the relief valve and that shock-loading conditions could have resulted when the pressure surge forced the water through the valves. Visual evidence does not support this theory. The tide mark deposits detected on the surface of the 6 inch line suggests that the quantity of the water present was too small to impose significant shock conditions.

A further suggestion by plant personnel was that water could have been present on the discharge side of the relief valves. This is considered feasible as water could condense from the gas passing the valve. It has also been suggested that water from the main vent could somehow drain back down these lines. If significant quantities of water were present then it would seem possible that the pressure surge could have activated a relief valve. This would then have resulted in a sizeable shock load, both as a result of the relief valve operation and also because of the extra forces imposed when any water was pushed through the system.

When repairs were made to this section of pipe work after the fire, a thicker 14 inch bend and straight section were installed such that reinforcement was no longer required at the T joint. Also the 6 inch line was canted at a slight angle so that water formed by condensation on the inlet side of the relief valves would run down into the 14 inch line and be re-converted to steam. Plans are also in hand to provide a drain line on the outlet side of the valves.

The steps mentioned should ensure that a further occurrence of this nature is unlikely in this position. It would also seem wise to review other lines to ensure similar situations do not exist elsewhere.