Aluminium Demineralized Water Storage Tank Leak

[Bambie]

thread238-153684
A 50 ft dia x 34 ft high aluminum demineralized water storage tank located outdoors began leaking from the floor at a single location on the perimeter in March.

After draining, internal inspection revealed a 1/2"dia hole in the floor plate approximately 6 inches inboard from the shell/floor corner weld.

The tank shell sits on a 6 foot deep cylindrical concrete slab that has a 6 inch wide and 4 inch high curb around the perimeter that was coated with tar.

The floor plates are supported by a 4 inch deep layer of sand that is retained by the curb.

The tank has a single 17AWG copper ground cable bolted to a welded lug on the exterior shell surface and lag bolted to the concrete curb.

The tank is heated in winter by a steam coil and is connected to the Turbine Building via a 20"dia buried pipe.

The hole is located diametrically across from the ground cable.

My question is whether the ground continuity was lost and that resulted in a single hole from either galvanic corrosion or a lightning strike.

Will measuring ground resistance then disassembling, cleaning and re-assembling the ground cable connections followed by ground resistance measurement point to the cause?

 

[Bambie]

TugboatEng, LittleInch and itsmoked,

The 4"dia aluminium steam heating coil is u-bolted to 2 foot high pedestal supports welded to the floor plates. 4 inches of sand is what lies beneath 96% of the floor plates - except in the area near the perimeter where it was washed out during the leak.

Borescope, replica and UT examination are all planned however the constant threat of thunderstorms has prevented entry for the past week.

Lack of temperature control during winter has caused the 8 ft x 4 ft floor plates to bubble-up a few inches, measured at their centers. The floor plate edges are butt welded with fillet welded backing plate strips and do not bend when compressed under constrained thermal expansion.

Besides bubbling, I suspect seasonal thermal cycling has also caused radial movement of the concrete supported floor perimeter resulting in damage to the tar coating, which would provide a site for galvanics if the grounding wire has also lost continuity.

I would be interested in suggestions on how to re-support the floor plates to prevent reverse bending stresses at seam welds during re-fill especially in the area surrounding the leak site.

 

[TugboatEng]

The bubbling up may also be related to pitting corrosion on the underside of the tank as the aluminum oxide takes up substantially more volume than the parents aluminum did.

 

[Bambie]

TugboatEng,

Unlike most metals, aluminum oxide density is larger (3.96 g/cm^3) than the parent (2.7 g/cm^3), so that should mean that the corroded mass would occupy less volume than the parent.

 

[Bambie]

Ultrasonic floor thickness measurements found 20 sites of galvanic corrosion around the perimeter of the tank.

A typical corrosion site comprises several square inches where thickness dropped abruptly from 0.5" to less than 0.25" with localized pitting thicknesses of approximately 0.125".

The four grounding cable continuity checks were all acceptable, indicating that active or passive cathodic protection should have been provided.

The corroded areas are inboard of the concrete curb approximately 2", indicating that neither floor movement nor tar coating breakdown was a causal factor.

Although corrosion sites are present all around the perimeter, the 1/2"dia hole and a significant cluster is next to a Turbine Building door and walkway where snow removal and road salt is dispensed in winter.

It appears that individual galvanic cells were formed when road salt contaminated water moved from the curb to the 4" sand base around the perimeter during spring rains.

With no previous wall thickness measurements available, the rate of corrosion is not known, therefore cathodic protection is highly recommended.

 

[TugboatEng]

I don't think galvanic corrosion is the correct assumption here. If the tank is sitting on a sand bed and there is salt water in the sand, you have created ideal conditions for crevice corrosion. It may be more productive to investigate methods to keep the sand bed dry and free of chlorides.

 

[Bambie]

TugboatEng

Are you suggesting that the gap between the sand bed and aluminum plate is the crevice?

None of the corrosion cavities is located at a floor seam weld or the fillet welded backing bar where you might find a crevice.

Wikipedia states the following:

"For a given crevice type, two factors are important in the initiation of crevice corrosion: the chemical composition of the electrolyte in the crevice and the potential drop into the crevice."

Reducing the potential drop in the "sand gap" with buried magnesium anodes is a simple and inexpensive mitigating technique.

Cutting out and replacing 1,000 square feet of perimeter floor plate, replacing the sand and sealing the perimeter from moisture ingress would need a solid cost/benefit argument.

This tank is adjacent to station transformers that should have a protective copper grounding grid, which raises the possibility that it may be the source of a high electrical/chemical potential that could be driving galvanic corrosion.

 

[TugboatEng]

Anywhere the sand is in contact with the aluminum is the crevice. The trouble with using anodes is that they require good conductivity of the sand to be effective. That means any wet areas, think of puddles, that are isolated from the anodes by dry sand are not protected and will become sites of pitting.

 

[Bambie]

TugboatEng,

I was intending to bury anodes in the wet soil surrounding the concrete foundation and run copper conduits to the tank shell.

Stray currents resulting from electrical potential would be routed from the aluminum tank to the magnesium to the soil rather than from the aluminum tank to the wet sand to the concrete to the soil.

 

[TugboatEng]

You're hyper-focused on the stray current problem. Is the entire underside of your tank wet? You have described localized pitting corrosion which excludes galvanic protection as a viable option.

https://smartech.gatech.edu/bitstream/handle/1853/11108/green_leon_c_197608_ms_261144.pdf

 

[Bambie]

The attached link is very similar.

https://www.google.com/url?sa=t&rct...7229A355.pdf&usg=AOvVaw2yZChi0UW1oDeLOGjpGUup

 

[TugboatEng]

In the link the problem is specifically that the bottom of the tank had come in contact with the copper ground grid and it was suggested that the entire bottom be replaced with a fiberglass liner.

The root of the problem is the moisture. Aluminum in contact with copper won't corrode without moisture. In your case, the copper isn't present but you have suggested chlorides are. I suggest that if you can prevent the sand under the tank from getting wet you can stop further corrosion of your tank bottom. Or, saturate the sand with water and use anodes/impressed current.

 

[Bambie]

TugboatEng,

The problem with burying anodes and/or providing impressed current in the soil adjacent to the 6 foot deep, 60 foot diameter, massive concrete foundation, is the unknown conductive characteristics of the concrete.

Burying anodes in the wet sand under the floor plates would require cutting, welding, wiring etc. - more work than could be justified if there was a simpler solution.

As the tank was being drained for inspection and the floor plates rebounded into their bubbled configuration, water egressed from multiple locations around the perimeter of the tank, which suggested that the entire 420 cubic feet of sand was probably saturated with demineralized water.

To ensure this is the case, a temporary berm could be erected on the concreate curb around the perimeter, filled with demineralized water and left to soak into any pockets of dry sand and dilute the chlorides.

Sand saturation and chloride dilution could be performed every spring to mitigate galvanic corrosion.