An abrupt, catastrophic explosion and rupture of a sump tank occurred. The sump tank was actually not a normal sump tank, but was a two compartment tank. A solid, full section, plate of steel separated both tank sections and there was no passageway for gases or fluids to pass from one compartment to the other. The subject sump tank was partially buried. The two tanks were welded together, giving a visual appearance of being a single tank. Operating personnel would have had absolutely no indication that the subject tank was, in fact, two completely independent tanks. The tank has been described as a short tank (which contained sump liquid) and a long tank (which contained a non-functional pressure gauge). The two tanks were separated by a solid baffle which prevented gaseous or liquid communication between the long and short tank sections.
Evidently, the workmen were attempting to pressurize the larger tank section in order to blowout or displace sump tank fluids from the smaller tank section into a nearby waste storage vessel.
It was reported that the pressure source being used was gas from the high pressure gas sales line. It was further reported that the gas pressure being used by the workmen at the time of the rupture was probably about 800 psi. The failed long tank, sitting on a two wheel dolly, is shown in Photograph A.
Photograph A The failed circumferential weld is located at the right end of the tank.
The circumferential weld which failed is shown to the right in Photograph A (see white arrows). A rear quarter view, showing an overall view of the failed end of the long tank is shown in Photograph B.
Photograph B Long tank, showing the failed circumferential weld.
A side profile view of the short tank is shown in Photograph C.
Photograph C View of the short tank. The circumferential weld which failed is shown.
The circumferential weld which failed is at the left end of the short tank in Photograph C (see red arrows). Photograph D shows the baffle plate which completely separated the short tank from the long tank.
Photograph D Failed weld and baffle plate on the short tank.
It is noted that the baffle shown in Photograph D is slightly dished, i.e., concaved. It is evident that this baffle did not allow liquid or gaseous communication between the two tanks.
The long tank fracture surface and the mating short tank fracture surface were photographed in detail. Visual examination of the girth weld fracture surface revealed three significant fracture features. An overall view of the short tank fracture surface is shown in Photograph E.
Photograph E Failed circumferential weld on the short tank.
The root of the circumferential weld contains what appears to be welding slag and evidence of incomplete weld root penetration. The welding slag is the darker, gray material at the weld root located at the junction between the weld metal and the outside diameter of the baffle (see red arrows in photo E). Additionally, above the incomplete root penetration area (toward the outside weld surface), a darker discolored thickness of weld metal is observed. This appears to be a region where the weld failed progressively at some prior time, under the influence of cyclic pressure, by metal fatigue. The fact that this area of the fracture is darker colored (more corroded), is evidence of crack growth occurring over a period of time, as a result of periodic or cyclic pressure.
A third area of fracture, created during the tank rupture is located near the outside edge of the failed weld (see white arrows). A close-up view of the three distinct areas within the short tank fracture is shown in Photograph F. The area exhibiting lack of root penetration is denoted with red arrows.
Photograph F View of details of circumferential weld fracture.
The area of the fracture exhibiting old fatigue fracture is denoted with a white bracket in Photograph F. The area of newer, fresher fracture is denoted with white arrows.
When critical pressure levels occurred, the crack increased in size. As a result, the thickness of weld metal available to resist these pressurization stresses diminished.
On the day of the incident, the pressure which workmen put into the long tank was of sufficient magnitude that the effective wall thickness that remained in the girth weld was insufficient to resist the resulting longitudinal stress and the vessel rupture occurred.
As a result of the studies conducted in this matter, it was determined that:
- Failure of the subject sump tank was the result of pressurization of the long tank.
- The girth or circumferential weld connecting the long and short tanks had been weakened by fatigue cracking and/or cyclic tearing resulting from prior pressurization during the vessel's lifetime.
- The subject long/short tank girth weld exhibited lack of weld root penetration. This lack of weld root penetration provided a possible source for metal fatigue and/or cyclic tearing crack initiation.
- The vessel, as configured at the time of the accident, did not have a pressure relief valve. Such a relief valve would have limited the user's ability to apply pressures sufficient to result in crack growth within the subject girth weld. Such a relief valve would have reduced or eliminated the possibility of catastrophic vessel rupture.
- The pressure gauge on the subject tank was visually inoperable and thus could not be relied upon to indicate the tank's internal pressure.
- The tank owner had not verified that the subject tank was suitable for use as a sump tank, i.e., by pressurizing one side of the tank, sump fluids should be forced from the other side. That is, the tank owner had not verified that pressure communication existed between the long and short tanks.
- The tank owner had not verified the integrity of the sump tank for its intended purpose. The tank owner had apparently not determined the integrity of the circumferential girth weld.