The industrial gas industry produces the elemental components of “air” (nitrogen, oxygen, and argon) that are used in other industries as part of their processes. Once separated from air, each element exists as an extremely cold cryogenic temperature liquid. Typical liquid temperatures are −185° C. Large, field-erected tanks are used to store these liquids, and these tanks range from 100,000 gallons to over 2,000,000 gallons in size. API-620/Appendix Q standards, or their equivalent, are the routine guiding design documents. These tanks are built as flat-bottom, domed-top tanks, typically with stainless steel inner containers, carbon-steel outer containers, and the top/side insulation space between these tanks filled with three to five feet (radial dimension) of granular perlite insulation. The inner tank sits on a sandwich structure of Foamglas® insulation and concrete. During cooldown and commissioning, these tanks need to be dried, inerted, and brought down to very cold temperatures to allow product to be introduced. If done incorrectly during cooldown, these tanks risk vacuum collapse. This article discusses the details surrounding one such occurrence. 

Case Study Specifics

A client with a 450,000-gallon liquid oxygen tank attempted cooldown using their historical procedures, which had been used successfully before. These tanks are often built to a design pressure of one to five psig, yet are very weak in any sort of vacuum condition. The tank must first be prepped for cooldown until it is sufficiently dry and inert. The typical process for cooldown is to introduce end-product into the tank, typically through a top connection, to lessen top-to-bottom temperature differential and metal stress, compared to bottom entry. Various methods can be employed, either cold gas entry or momentary ‘spurts’ of liquid, with defined interruption between applications to allow product vaporization and temperature normalization. In almost all cases, the supply line (and control valves) used for this step is very small (one to three inches in diameter), enabling good volumetric control of the incoming cold gas/liquid. In this case, however, they were forced to use a 12-inch diameter line. Externally, the client still used a small (two-inch diameter) line from the liquid source to the tank’s 12-inch diameter line connection, believing this would provide the fine degree of control they needed for the incoming cold fluid. At some point during the cooldown process, the dome collapsed into itself and dropped nearly six feet. With inner and outer containers connected at the top/center through a manway, this connection caused both the outer and inner tank top dome to collapse, as the inner tank suffered the actual vacuum condition.

Probable Causes

There are a number of probable causes for the collapse. Differences of opinion exist between the client company and the external consulting service employed to investigate, but certainly any or all of the following were contributors:

 • Lack of operational discipline—Tank pressures and temperatures were monitored by the control room throughout the process. Screen capture recordings of the process indicated that the tank suffered multiple dips into a vacuum condition prior to a large, final vacuum condition, which then caused the inner container to collapse. Staff did not mention any unusual conditions, but interviews with personnel indicated a major management push to get the work done, which was severely behind schedule. 

• An abnormally large internal supply line within the tank (12-inch diameter line)—This more than likely accumulated a large volume (800 gallons) of cold liquid, rendering well-controlled smaller input volumes via the external two-inch diameter supply line ineffective. Small incremental spurts of liquid that were introduced could have vaporized and lifted a much larger quantity of stored liquid within this 12-inch line to spill into the main tank. It was found later that the lower tank penetration of this 12-inch line through the outer tank sidewall also showed excessive external ice accumulation, indicating insufficient insulation in this area and subsequent heat leak. Either heat leak or ineffective input liquid control contributed to unusually large amounts of liquid entering the inner tank during cooldown. 

• Cold fluid applied to a warm tank causes the fluid to vaporize and create pressure—This continues but not throughout the process. At some point, there is a cold-gas atmosphere that exists within the tank, and any (excessive) subsequent addition of liquid causes the gas contents to condense, which would vacuum-collapse the tank instantaneously. Although the tank was protected by vacuum relief valves, the instantaneous nature of the condensation/vacuum condition creates an enormous flow requirement that no vacuum relief valve (regardless of size) can process. To illustrate the violence and uncontrolled nature of the collapse, there is a YouTube film clip showing a similar (purposeful) tank collapse of a railway tanker:  https://youtu.be/2WJVHtF8GwI 

Application

Although this case study was for an industrial gas product application, the same failure can occur in other industries and products, such as LNG (a cryogenic fluid at − 162° C) and various oil/gas scenarios where large tanks are often steam-cleaned internally before poor weather affects the plant site.

Image: Example of a tank collapse. Credit: European Industrial Gases Association