A Hydrotest Goes Awry

The destructive hydraulic forces of mild-mannered water were displayed in a classic style on a summer evening in 1996. A Gulf Coast chemical plant had just rejuvenated a recovery tank and started a hydrotest when things went wrong. The poorly executed test added 35,000 to the repair costs. Other than embarrassment, no one was injured. In the previous three weeks this tank had received repairs consisting of a new floor and new walls in the lower section. 

While preparing for the test, two mechanics blinded a six-inch (15 cm) pressure relief flanged top-mounted nozzle. The craftsman also capped two smaller nozzles on the tank roof with flanges. These smaller nozzles normally were equipped with a pressure gauge and 

When the maintenance crew returned to the job site a short time later, they observed a water stream forcefully squirting from the two 1/2-inch vents. They noticed this normally vertical tank leaning at a slight angle. The bottom bulged upward at the bottom-to-shell joint as much 8 inches (20.3 cm) above the foundation. Anchor nuts ripped through the anchor chairs, and catwalks attached to the tank roof had torn away from welded clips. 

At this section of the plant in the past, chemical process operators (not maintenance mechanics) handled the tank-filling step in the hydrotest process. Operators would typically drape a fire hose into an oversize roof nozzle and fill the tank from the nearest hydrant. In this case, it was reported the two mechanics asked their maintenance supervisor if they could connect the hose to a flanged lower-valved nozzle on the tank. The supervisor remembers requesting the mechanics to roll a blind flange at the top for venting purposes if they used the lower nozzle for filling. There must have been a miscommunication. Obviously the mechanics failed to understand the dynamics of the filling operation when they chose to open only the two 1/2-inch top vents. 

Two 1/2-inch-diameter pipe openings provided only 0.46 square inches of relieving area. These openings were sufficient for the air being expelled, but not for the water that would 

Total restoration costs approached 35,(X)0. Foundation repairs including reselling the anchor bolls and other support work cost about 5,000. An additional 30,000 was spent to lilt the lank, fabricate and install a new tank bottom, and to reinstall and reconnect the pipe to ihe vessel. 

---