Grounding the Space Shuttle—NASA’s Foam Insulation Problem

Gordon H. Hart

Gordon H. Hart, P.E., is a consulting engineer for Artek Engineering, LLC. He has over 35 years of experience working in the thermal insulation industry. He is an active member of ASTM committees, including Committees C16 on thermal insulation and F25 on marine technology, ASHRAE's Technical Committee on Insulation for Mechanical Systems, and the National Insulation Association's Technical Information Committee. He received his BSE degree from Princeton University. and his MSE degree from Purdue University, both in mechanical engineering. He is a registered professional engineer. He can be reached at gordon.hart®

December 1, 2005

It is now more than 2 years since the space shuttle Columbia disintegrated during re-entry to the earth’s atmosphere on February 1, 2003, killing all seven astronauts aboard and, in the process, putting a spotlight on foam insulation.

In the years since the tragedy, the U.S. National Aeronautics and Space Administration (NASA) has sought a solution to a well-publicized problem with the foam insulation on the space shuttle’s external fuel tank. While the shuttle’s insulation problem is no longer headline news, it has not been resolved, nor has it disappeared. If NASA does not find and implement a solution, the space shuttle fleet could be grounded. How did this exciting, highly technological space exploration program come to be threatened by something as seemingly mundane as foam insulation material?


Lockheed Martin is NASA’s fuel tank contractor. The company also spray-applies the polyurethane foam insulation over the shuttle’s external fuel tank. Within the external tank is liquid hydrogen fuel in one cavity (at -427°F) and, in another cavity, liquid oxygen (at -297°F). Both of these temperatures are considered cryogenic. (For comparison, natural gas is liquefied at about -310°F and typically transported and received as liquefied natural gas [LNG] at -260°F.) The foam insulation has a minimum thickness of 1 inch and a design density of 2.4 pcf. In some areas, such as around protruding structural members, it is applied even thicker than 1 inch. Lockheed Martin applies a coating over the insulation to help protect it and to act as a vapor barrier, but a separate vapor barrier sheet has not been applied over the foam insulation—a step usually considered necessary by those in the insulation industry.

A June 2003 article in the publication Florida Today reported that a study of NASA records showed that all 113 shuttle missions flown before the Columbia disaster were damaged by launch debris. It was noted that connectors to the external fuel tank, known as the intertank, shed the most foam. This area also is most susceptible to harboring ice. The article explained that when workers apply foam to the intertank and bipods (which protect the connector between the fuel tank and the orbiter’s nose), they shave the protective coating and poke tiny holes into the foam. The shaving and the holes can provide paths for both gas and moisture to penetrate the insulation. The absorbed moisture can subsequently freeze and turn to ice, which can then cause the foam to pop off during flight. Clearly, foam with ice can be much heavier than pure 2.4-pcf foam and can cause more serious damage to the shuttle orbiter’s thermal shields. In addition, the fuel tank is often exposed to rain and humidity while the shuttle sits on the launch pad waiting for appropriate launch conditions. Before its last flight, Columbia sat on the pad for 39 days during heavy rains.

Accident Investigation

In March 2003, NASA convened a “Shuttle External Tank Technical Forum” consisting of 25 specially selected scientists and engineers from academia, government research centers and industry to study the foam insulation problem and make recommendations.

Then, in June of that year, NASA investigators reported that they found the smoking gun—proof that a piece of foam insulation did, in fact, damage a heat shield. At the Southwest Research Institute in San Antonia, Texas, NASA tested a 1 2/3-pound piece of foam by firing it with an air cannon at a panel taken from another shuttle’s wing. At speeds up to 530 mph, the foam blew a 16-inch square hole into the wing panel. A hole of only 10 inches across would have been enough to lead to the disintegration of the space shuttle upon re-entry.

A year and a half later, NASA officials reported that super-cooled chemicals inside the tank caused ice to form outside as the shuttle prepared to launch and confirmed that the use of 1-inch foam insulation merely reduced the amount of ice formed (rather than prevent ice formation).

The View From One Engineer’s Perspective

Review of the brief history since the Columbia disaster and the prominent role of the spray-applied polyurethane foam insulation raises several questions. First, how can 1 inch of polyurethane insulation adequately insulate a surface at -427°F and prevent ice formation, even assuming that no cracks or voids form?

This author does not have access to the specification of NASA’s spray-applied polyurethane insulation but offers for consideration the American Society for Testing and Materials (ASTM) specification C1029-02 for spray-applied rigid cellular polyurethane thermal insulation, which covers material with a temperature range only from -22°F to 225°F.

ASTM C1029 does not provide thermal conductivity values but does give an R-value of 6.2°F – ft2 °F / Btu for a 1-inch thick sample at a mean temperature of 75°F. Taking the inverse, 1/R-value, gives a thermal conductivity, or K-value, of 0.161 Btu/h ft2 -°F. At a mean temperature of -200°F (approximately what the foam insulation would experience on a -427°F tank in a 75°F environment), the thermal conductivity should be considerably lower.

A broad thermal evaluation yields one possible alternative—a rigid foam insulation that can be used to -200°F, ASTM C578-04, Type XIII polystyrene insulation. While this material probably has a different thermal curve than spray polyurethane, its thermal behavior is also likely to be similar enough to that of spray polyurethane to use as a surrogate. Inputting into the computer program 3E Plus® Version 4.0 (available free from the North American Insulation Manufacturers Association at using the default thermal conductivity values for the Type XIII polystyrene in the program, this author estimated the insulation thickness required to prevent ice formation in a 75°F, 75 percent RH environment with a 5-mph wind (likely an unrealistically mild scenario). This thickness is predicted to be somewhere between 1 and 1.5 inches. Based on these mild weather conditions, it would appear that another 0.5 inch or so of foam insulation would need to be added to the 1 inch NASA currently uses to prevent ice formation under very mild launch conditions.

Re-run the 3E Plus evaluation with more realistic ambient conditions—a cool February morning in Florida, a 50 F air temperature, 80 percent RH, and 0 mph wind—and the scenario changes dramatically. In that case, one would need between 3.5 to 4 inches of foam insulation to prevent ice formation on the skin of the insulation. While these scenarios certainly do not provide scientific proof, they do offer an argument that the foam insulation used on the space shuttle’s external fuel tank may be under designed—at least from a thickness perspective. In particular, the insulation could be underdesigned on the bipod and intertank areas, where pieces of foam insulation have become dislodged during past takeoffs.

But why does ASTM C1029 spray-applied rigid cellular polyurethane insulation have a lower temperature limit of only -22°F, 400 F higher than the -423°F temperature of liquid hydrogen? While I don’t know for certain, I believe that it is likely due to dimensional instability (i.e., it shrinks significantly at cryogenic temperatures). Furthermore, at some cryogenic temperatures, the blowing agent gas would condense and freeze. It is not clear how this has been accounted for. If, as has been reported, Lockheed Martin uses an HFC blowing agent, is it a special HFC blowing agent?

Another question is how NASA has accounted for an increase in thermal conductivity with time, characteristic of organic foam insulations over their first 6 months of life. While Lockheed Martin tests the booster rocket with the cryogenic liquid fuels, it likely does so with foam insulation that is fresh, maybe only a week or 2 old. In contrast, the external fuel tank is typically insulated 6 months or so prior to an actual launch.

The next question, then, is what about the formation of cracks and gaps? Any crack that forms in foam insulation in a cryogenic application can create a direct opening to the cold surface. One such crack, even on a pipe or tank that does not move, can be catastrophic due to the subsequent formation of ice. An ice ball will grow rapidly in a humid environment until the insulation is split and separated from the insulated surface. This is a major reason why LNG pipe insulation—even when it is cellular and has a very low vapor permeance—is covered with a thick, rubber vapor barrier. (A vapor retarder is insufficient for cryogenic pipe insulation applications). Surveying all of the known information, NASA’s fuel tank foam insulation has a coating, but not a separate vapor barrier, applied over it. Can a coating effectively prevent moisture from intruding into foam insulation and prevent cracks from forming in the insulation itself?

The ASTM C1029 value for water vapor permeability is a maximum of 3 perm-inches, which is not a particularly low value (compare this to ASTM C552 cellular glass insulation, which has a water vapor permeability of 0.005 perm-inches, a value that is 99.8 percent lower than that of polyurethane insulation). Further, the water absorption value in C1029 is a maximum of 5 percent, again not a particularly low value (again compare to C552 cellular glass, which has a water absorption value of 0.5 percent maximum, 95 percent lower than that of polyurethane foam insulation).

Lessons Learned

As a result of the investigation into the Columbia tragedy, NASA implemented several changes. As early as June 2003 it was reported (Florida Today) that NASA was redesigning a bipod ramp as a solution to the problem of insulation breaking away from the shuttle’s 15-story fuel tank. NASA itself reported in December 2004 that redesign of the external fuel tank included the addition of heaters at key points to prevent ice formation before launch.

Earlier this year, prior to the launch of the shuttle Discovery, NASA reported that while it was likely that some foam insulation would be dislodged during that shuttle’s liftoff, the pieces would be of insufficient size to damage the shuttle’s thermal shield tiles. Operating under the motto “Return to Flight” since the Columbia disaster, NASA engineers focused on minimizing foam insulation loss during launch and eliminating foam insulation from the bipod connecting the fuel tank to the forward part of the orbiter. Engineers modified several of the tank’s external fixtures so that the foam insulation could be sprayed on more uniformly, without voids, and could expand during launch to keep from dislodging. NASA engineers reportedly ran millions of computer simulations of various sizes of foam insulation pieces being fired at the reinforced carbon heat shields.

In July 2005, NASA reported that they changed the foam insulation a decade earlier, switching from a foam-blowing agent that used an environmentally damaging chlorofluorocarbon (CFC) to one using a more benign hydrofluorocarbon (HFC) blowing agent. The newer HFC-blown foam insulation is a significant change since it is reported to be more brittle than the originally specified insulation material.

The Return to Space

After spending $1.5 billion to fix the fuel tank and implement other safety upgrades, NASA successfully launched the space shuttle Discovery July 26, 2005. However, 2 minutes after liftoff, photos showed a large piece of foam insulation coming off the external fuel tank (although it did not appear to hit the leading edge of the shuttle). NASA estimated the piece to be 24 to 33 inches on one side, 10 to 14 inches on the other, and 2 to 8 inches thick—approximately the same size as the one that critically damaged Columbia 2 years earlier. Although the Discovery landed safely at Edwards Air Force Base in California at the end of a 14-day mission that included in-flight repairs on damaged heat shields conducted by one of the astronauts, NASA suspended future shuttle flights because of risks posed by falling foam insulation debris.

Additional Modifications

Now, NASA is removing the 37-foot “prototuberance air load ramp,” a long foam insulation protrusion designed to smooth airflow over the tank at high speeds and ease vibration to nearby piping and cables. It is believed to be the source of the 1-pound piece of foam insulation that broke off during Discovery’s launch, nearly hitting the orbiter’s right wing. It will be replaced by a new type of foam insulation that will be applied with more exacting techniques, designed to prevent shedding.

Looking Forward

At a news conference on October 14, 2005, NASA officials target a May 3 to May 23, 2006 window for the next Discovery launch. They reported that shuttle workers will likely replace and modify areas of insulation on the external tank where foam broke loose during the July 2005 launch. Program manager Wayne Hale says that a series of tests over the next several weeks will help further clarify the tank issues. “I think we’re beginning to have our hands well around the technical problems that we have and to find the fixes that are going to be necessary to fly again,” he said.

NASA clearly has studied the foam insulation problem extensively and determined that the causes leading to foam insulation breakage include an imperfect surface and overcooling around structural protrusions, such as the bipod area. The imperfection problem is being addressed by greater attention to the application of the insulation and coating, and the local overcooling problem is being addressed by adding electrical heaters to those areas where ice formation has apparently occurred.

Let’s hope we’ve seen the last of the foam insulation problem.