One Eighty Materials Testing was appointed by a leading wine company in South Africa to investigate the failure of a wine storage tank one of their cellars.
The tank failed catastrophically and imploded as seen in Figure 1 below. The consequences of the damage can be seen in Figure 2 below, where the surrounding wall collapsed.
One Eighty has twenty years of root cause failure investigations for a wide range of industry sectors. Our customers have come to know and rely on our thorough and intelligent approach to these investigations that always elucidate the cause for failure.
Our approach in this case was to conduct a site visit to assess the situation and determine the most appropriate methodology in order to identify the root cause failure of the wine tank.
We selected and recovered samples from site and delivered and delivered them to our labs for further metallurgical testing.
The grade of material was determined by way of spectrographic analyses to calculate the chemical composition of the tank material and supporting columns, and whether it complies to any normal steel grade for wine tanks. Samples recovered were analysed by way of visual, macro-examination, microstructural and hardness testing through the cross-section near the location of failure in order to determine whether any deleterious features in the material could have contributed to the failure. Inspection of current cleaning procedures, including but not limited to the chemical agents used, was inspected in order to determine whether the correct procedures were applicable.
We had an option to carry out Scanning electron microscopy (SEM) and electron dispersive x-ray spectroscopy (EDS) inspection of the fracture surface was conducted to determine the possible mechanism of failure – should the root cause not have been fully identified by way of light microscopy.
The steps we took in finding the root cause of the wine tank failure can be summarised as follows:
- Visual inspection
- Wine tank floorplan
- Failed seams/edges
- Structural support beams
- Proximity vessels and strictures
- Chemical analysis
- Spectrographic analysis to identify alloy composition
- Weld near fracture opening
- Tank wall welding
- Support column flange plate
- Microstructural analysis
- Tank welds
- Support column flange plate
- Hardness results
- Weld on the tank
- Support column flange plate
- Examining cleaning method for the tank
- MSDS (Material Safety Data Sheet) for all chemicals used
Visual examination of all relevant failed sections was used to determine the method of failure. Various fractured sections contained different modes of failure, such as ductile overloading and brittle failure. Sections of steel near the crevice of the tank, located behind the interior support column member, revealed incomplete welded sections of steel on the exterior side of the tank wall. The incomplete welded sections contained only tack welds, which would have been used for the initial constructions of the tank. Although the crevice with incomplete exterior welding was observed, the interior was reinforced with a backing strip, lap welded to the tank wall. The exterior surface of the tank showed clear evidence of surface corrosion behind the support column member, localised around the incomplete welds, where a crevice was left as a result of incomplete welding.
Material- and chemical analysis/observations
Spectrochemical analysis showed that the material of the tank and support column flange plate to match with stainless steel 304, designation AISI S30400/S30409, and low carbon steel, designation AISI 1108/S355, respectively. The flange plate steel was noted to be slightly out of specification to be an exact match.
The procedure and chemicals used in the cleaning procedure did not indicate any contribution to the failure as was evident of the pristine condition of the stainless-steel interior surfaces.
Figure 1: Collapsed tank
Figure 2: Structural beam through wall
Macro-examination revealed that the exterior welding near the open edge was incomplete. Tack welds and open crevices were left without a complete weld. The location of the incomplete welds was behind the support column, providing little to no way of have visually observing the incomplete seam welds.
Cross-section macro-examination revealed the extent of the incomplete welds, with partial weld penetration on the base metal. The insufficient welding or lack thereof would allow for moisture from the surrounding air to seep in between the open seems and slowly corrode away at the backing strip and surrounding welds. Welded sections of the tank wall near the open seam showed no signs of bevel grooves indicating a lack of weld preparation/procedure.
Microstructure analysis on the weld sections revealed that the welds near the open edges were subjected to corrosion attack. The incomplete welding near the open seams would allow for small amounts of moisture to slowly deteriorate the weld over a prolonged period of time until the remainder material can no longer withstand the stressed induced by the weight of the contained fluid. The lap welding on the backing strip on the interior side of the tank showed no signs of corrosion but did show signs of insufficient weld penetration to the base metal.
Hardness analysis of the weld section taken from a section of steel away from the open edge showed that the tank wall weld to not vary significantly and within specification to stainless-steel 304. This would indicate the overall welding across tank, with welding on both sides, to be sufficient in strength. In order to fully determine the quality of the welding, analysis of a valid welding procedure specifications (WPS) and accompany welding procedure qualification records (WPQR) are required to ensure that the tank conforms to any relevant standard.
|Vickers hardness results – Stainless steel welding|
|Parent material 1||Heat affected zone 1||Weld metal||Heat affected zone 2||Parent material 2|
We concluded that the failure of the wine tank was due incomplete welding on the vertical section. This enabled the corrosion of the tack welds in the presence of atmospheric moisture. The incomplete welds appeared to have been localised behind a support column that was obscured visually from the exterior, allowing for corrosion to slowly deteriorate the open seams until failure occurred. The probability of overfilling the tank would not have been the primary cause behind the failure of the tank as typical vessel design would allow for a sufficient safety factor.
Due to the composition and age of the tank, the effect of corrosion required a significant amount of time for the vessel to fail. This is why the tank did last 20 years.
Due to early identification of the root cause of failure, SEM/EDS (Scanning Electron Microscopy/ Energy Dispersive X-ray) analysis was not utilised.
One Eighty recommends that other wine tanks with similar age and construction should be further examined by way of non-destructive testing to ensure that no other failures occur.
Additionally, technical schematics, structural analysis by way of finite element analysis (FEA) and qualified welding procedures should be generated for each wine tank in order to certify all vessels with the relevant codes and standards. The use of a certified and qualified welding institute is recommended to conduct any repairs on similar wine tanks currently in operation.