The refinery fuel tank repair

Jul 1, 2008 by

The case of the month involves a fuel tank located in a petroleum distribution terminal in a very remote area.  The tank was 48 feet high and 65 feet in diameter, with a gross capacity of over 28 thousand barrels.  Normally it was filled with premium unleaded gasoline.  It was constructed by welding 5/16 inch thick, 8 feet wide by 24 feet long steel plates together, in 6 horizontal courses. The tank was over thirty years old and had seen its share of bad weather.  Corrosion and pitting were proportionate to the tank’s age.  The last typhoon it experienced damaged the tank, which required repairs.

As a result a consultant was engaged to evaluate the tank.  He did a life expectancy analysis and recommended the tank be strengthened. His proposed solution was to install stiffening rings, angle iron curved into sections that matched the curve of the tank.  These sections would then be joined to each other at their ends and welded around the tank to form rings around the tank (roughly in the upper third). Three encircling bands would strengthen the tank against future damage.  Unfortunately, the owners of the tank decided it was an expense they didn’t want to make at that time.

Since the tank was to remain empty for a while, they decided to fill it with water to act as ballast against the approaching winds of the next typhoon.  Their solution worked against the storm winds-the tank sat solid as a stone.

However that region was well known for being seismically active, as well as being subject to typhoons.  A while after the last typhoon hit, an earthquake rumbled through the terminal and the sloshing water within the tank buckled the tank walls.  The water was drained out and the tank was surveyed.  The tank had bulges in and out of its sides. It was similar to gently squeezing an empty coke can- some portions of the wall bulge inward, and some outward.  For the fuel tank, maximum measured deflection was more that two inches in some sections, and about 25% of the circumference was affected.  At this point, our expert was brought in to recommend repairs for the damage caused by the earthquake.

The first thing he noted was the use of water as ballast for the tank.  Typically in the construction of new tanks, when they are completed, they are filled with water for a short time to test for leaks and then they are pumped dry.  This is a form of stress testing-and if there are leaks, you want them to be water, not 98 octane gasoline. The specific gravity of water is 1.0, by definition.  The specific gravity of gasoline is .74.  This means that for an equal volume, gasoline weighs 26% less than water.  The design of the walls of the tank was to support the long term loads imposed by the lighter fluid (gasoline), not the heavier fluid (water).  Keeping water in the tank for an extended period of time exceeded the design strength of the wall thickness for the lower two courses.  Throw a seismic event at an over stressed wall and buckling occurs.

Seasoned operators of a distribution facility should have known better than to overstress a 30 plus year old tank.  He also noted that had the recommended ring stiffening procedure been completed, and if the tank had not been filled with an overloading fluid, the tank would have been fine in both the storms and in the earthquake.

That being said, our expert recommended cutting out the damaged sections, and welding in new sections, one 4 by 8 plate at a time.  The total strip replacement area would be approximately 4 feet by 56 feet. As might have been expected, the operators of the plant wanted a new tank, and not just a repair (but they wanted someone else to pay for said new tank). They then came back with a list of criticisms of our expert’s recommendations, saying he greatly underestimated the cost of the repair.  Of course, there is a vast disparity between the cost of a repair and the cost of a new tank.

First, they said that his repair recommendations ignored the damage to the roof and floor of the tank.  The response was that while there was damage to the floor and the roof structures, the damage was due to weather and water corrosion, not the earthquake.

Then they said the estimate did not provide for scaffolding costs, nor for removal and replacement of the internal pontoon roof during the repair period.  As it happens, there was no need to remove the internal pontoon roof as the scaffolding would be supported from the shell, and the work could be done from the outside.

The next criticism was it would be prohibitively expensive to fly in X-ray equipment to X-ray the welds to inspect and verify weld quality.  The response was that there are other, equally reliable methods of verifying weld quality.  These alternative methods include ultrasonic, magnetic particle testing, and dye checking.  Dye checking would entail grinding smooth sections of weld, applying the dye, inspecting, and then applying a cover pass weld over those inspected areas. About one foot in twenty feet of weld would provide a sufficient statistical sample to verify weld quality for the repair.

The final criticism was the estimate did not address the failure of the coatings inside the tank.    As our expert observed, the interior tank coating was an after-market product that had been poorly applied. Now that the tank was empty, it was easy to see portions of the coatings sliding down the interior wall, leaving whole sections unprotected and thusly corroding.  The earthquake had nothing to do with the interior coating.

Related Posts

Share This