Microbiologically Influenced Corrosion Attacks Welds Before Component Is Put Into Service

About 200 feet of 2-inch, schedule 10, Type 304 stainless steel transfer piping was installed to replace a bulk handling operation.  The product, which could produce nitrous oxide as a breakdown product, was one that had been routinely handled at the plant for many years.  Within two weeks of when the pipeline began carrying the product, the line was observed to be leaking.  The insulation was stripped from 20 feet of the horizontal run on either side of the suspected leaking area and the pipe was visually inspected.  When no readily apparent leak sources could be identified, the welds attaching the 90º elbows to the straight lengths were dye penetrant inspected.  The dye penetrant inspection found two pinhole sized indications in two circumferential welds.

To characterize the failure, the leaking elbow, a suspect elbow about four feet downstream of the leaking elbow and a flanged end from one straight length were cut from the line.

The elbow that contained the leaking welds was longitudinally split to reveal the ID (inner surface) of the pipe, Figure 1.  In the area of the weld perforations there were reddish orange to black, circular discolorations and/or mounds of corrosion product on the ID surface of the welds.  Away from the perforations, the circumferential welds were discolored by a “bleeding rust” deposit.  A blue/black band was evident on both sides of these welds, about ¼-inch from the weld bead.  This band was typical of oxidation that takes place when stainless steel is welded without a protective atmosphere.  The weld area in the pipe ID showed a lack of penetration in addition to the bleeding rust colored deposits. 

Both the elbows and the straight pipe section contained longitudinal weld seams.  The straight pipe section contained a longitudinal weld seam with an unusual “feature.”  A length of wire-like metal, which extended out of the weld bead, apparently had remained attached to the weld seam throughout the construction and operation of the transfer line, Figure 1.  The wire was rectangular in cross-section and appeared to have been a feed wire for an MIG type welding technique.

The elbows were cold formed from straight lengths of pipe and not cast.  This type of construction, if executed with consideration for good craftsmanship standards, was allowed by the plant pipe code and considered acceptable for the intended service.

There was a pinhole on the OD (outside surface) of one of the leaking welds.  The pinhole was located on the edge of the weld, close to the pipe base metal.  A pinhole on the ID of the same weld was not directly opposite the OD pinhole.  Instead, it was slightly away from the OD pinhole and on the edge of the weld.  The locations of the two pinholes suggested that the leak path was slanted through the weld.

A cross section of the weld in the area of the leak was prepared for metallographic examination.  There was severe corrosion under the ID surface, Figure 2.  The corrosion was predominantly concentrated in the weld metal.  The corrodent dissolved the austenitic matrix of the weld and left behind a network of ferrite, which became embedded in the corrosion product.

Because of the relatively short time between start up and the detection of the leak (two weeks), the plant was concerned that the materials of construction were not to specification.  To confirm that the pipeline complied with ASTM A 312 for Type 304 stainless steel pipe, a sample of the pipe was analyzed for its chemical composition, another was subjected to tests to determine its mechanical properties.  The material was found to be in compliance to the standard.

How Did The Pipe Fail?

There were several issues to sort through for this failure analysis.  A review of the plant pipe code found that the pipeline was not fabricated to the welding standard.  The oxidations around the welds and the lack of penetration provided sufficient evidence that the ID of the pipe was not protected with an inert atmosphere during welding.

However, this appeared to have been the only deviation (though a significant one) from the pipe code.  The pipe contained the unusual feature of weld wire protruding from the ID side of the longitudinal weld.  However, this longitudinal weld was not attacked in any fashion in the areas available for inspections.  In addition, the mechanical and chemical tests showed that the pipe met the property requirements of the ASTM standard A 132 referenced in the pipe code.

The pipeline failed as a result of selective pitting attack of two circumferential welds.  There was evidence of corrosion by ferric chloride, which selectively attacks the austenitic phase of welds in the 300 series stainless steels.  Ferric chloride is an aggressive medium which, under the correct conditions, can produce the corrosion rates sufficient to have penetrated the welds within a month or less.  However, ferric chloride at the concentrations sufficient to cause the observed damage, was not part of the product process stream.  To have caused this attack, a concentrator mechanism was necessary.

The Case For Microbiologically Influenced Corrosion (MIC)

There are literally hundreds of different microbe types that have been found to be capable of thriving in what are considered to be extraordinarily hostile environments – but they have to become established first.  The microbes, some of which can be found in well water, tend to establish themselves at weld sites, especially at rough surfaces and crevices.  They begin by producing a protective mound, a biomass, that resists common biocides.  Their metabolic activity effectively pumps selected chemicals from the surrounding fluid to the surface of the metal below the biomass.

Ferric chloride is sometimes a by-product of this activity.  Ferric chloride aggressively attacks stainless steel, particularly when the steel is covered with a biomass that excludes the oxygen that stainless steel normally uses to form a protective oxide film.  In this case the steel becomes no more corrosion resistant than carbon steel.  Furthermore, considerable amounts of ferric chloride are released into the surrounding water.  This produces the bleeding rust often associated with MIC, which was present on this piping.

Finally, MIC was suspected because the speed of the attack was too high to have been caused by the plant process, even by the possible formation of nitric acid. The two weeks that the product was in the line prior to the discovery of the leak was not a sufficient exposure time.  MIC, on the other hand, is known to be capable of penetrating stainless steel within a period of weeks or months.

Discussion with plant personnel revealed that the line had been hydrotested with well water nine months prior to process startup.  [It has been well documented that microbes, which have been linked with MIC, exist in this type of water.]  Residual water, at low points in the pipeline, would have provided an adequate supply of nutrients for the microbes.  Since efforts were not made to completely dry the lines, the lines probably sat with water in the low points for nine months.  This was enough time for the MIC to penetrate the welds prior to the introduction of the product.

Path Forward

The nature of MIC and the evidence at hand led us to believe that the pipeline was in good condition overall – except at the circumferential welds.  These welds were suspect and subsequently examined by radiography.  The welds that showed signs of pitting attack were cut out and repaired. The line was placed back in service with no further leaks for over two years.

Lesson

Because all welds could have pitted, the plant had initially planned to replace them all to prevent further “weld corrosion.”  The radiographic inspection determined which had been attacked and which had not.  Only the pitted welds were replaced and the line was returned to service.

The plant has since changed hydrotest procedures – they use steam condensate rather than well water for the testing and they either delay the hydrotest until shortly before the equipment is placed into service or dry the line thoroughly immediately after testing.

web@corrosionlab.com