THERMOGRAPHIC ANALYSIS OF COMPOSITE COBONDS ON THE X-33
Samuel S. Russell
Nondestructive Evaluation Team
NASA Marshall Space Flight Center
James L. Walker
Nondestructive Evaluation Team
NASA Marshall Space Flight Center
Matthew D. Lansing
Nonmetallic Materials Group
NASA Marshall Space Flight Center
matt. lansing@msfc .nasa. go V
During the manufacture of the X-33 liquid hydrogen (LH2) Tank 2, a total of thirty-six reinforcing caps were
inspected thermographically . The cured reinforcing sheets of graphite/epoxy were bonded to the tank using a wet
cobond process with vacuum bagging and low temperature curing. A foam filler material wedge separated the
reinforcing caps from the outer skin of the tank. Manufacturing difficulties caused by a combination of the size of
the reinforcing caps and their complex geometry lead to a potential for trapping air in the bond line. An inspection
process was desired to ensure that the bond line was free of voids before it had cured so that measures could be
taken to rub out the entrapped air or remove the cap and perform additional surface matching.
Infrared thermography was used to perform the precure "wet bond" inspection as well as to document the final
"cured" condition of the caps. The thermal map of the bond line was acquired by heating the cap with either a flash
lamp or a set of high intensity quartz lamps and then viewing it during cool down. The inspections were performed
through the vacuum bag and voids were characterized by localized hot spots. In order to ensure that the cap had
bonded to the tank properly, a post cure "flash heating" thermographic investigation was performed with the vacuum
bag removed. Any regions that had opened up after the preliminary inspection or that were hidden during the
bagging operation were marked and filled by drilling small holes in the cap and injecting resin. This process was
repeated until all critical sized voids were filled.
The X-33 reusable launch vehicle technology demonstrator was designed to store liquid hydrogen fuel in a pair of
composite fuel tanks, shown in Figure 1. Each multi-lobed tank is an integral part of the vehicle's load bearing
primary structure. The tanks are attached to the rest of the airframe at points around their forward and aft ends.
Adjacent to each of these attachment points, a cast structural foam filler and bonded reinforcing cap are used to
distribute loads to the tank skins. The general locations of these caps are shown in Figure 2, viewing one of the
tanks from the aft perspective. The typical cap geometry is evident from the photographs of bonded caps shown in
Figure 1. X-33 internal structure.
Figure 2. Locations of reinforcing caps on tank.
The method of thermographic inspection employed to inspect the bond lines consists of exposing the reinforcing cap
surface to a rapid heat flux by pulsing a pair of high intensity flash lamps or slowly saturating the surface with heat
using lower intensity hand-held quartz lamps. After the cap has been heated, an infrared camera is used to record
the surface temperature as the heat transfers into the structure and the cap begins to cool. A void between the cap
and underlying foam acts as an insulator, and that region of the cap will retain higher temperatures over the void
than the surrounding area. Bond line voids are thus indicated as bright "hot" regions in the thermographic images.
Thermographic inspections of the wet bond line prior to curing required a novel bagging method. The
thermographic imager was capable of viewing the temperature of the cap surface through the bagging and release
ply material, but could not image through a traditional breather cloth, used to ensure thorough evacuation of the
vacuum bag. It was also determined that when air was trapped in the bond line it was very difficult to get it to move
to the edge of the cap so that it could be evacuated. The solution to this problem involved drilling a grid of vent
holes over the acreage of the cap and placing a series of split, cut in half along their length, plastic tubes over the cap
vent holes to facilitate evacuation. This arrangement required that minimal breather cloth only be used around the
cap edges and over the tubing, allowing the camera to image most of the cap area. The region covered by the tubes,
however, remained obstructed from view.
A digital thermography system was used to inspect most of the caps during bonding and all of the caps after the
bonding process was complete and the vacuum bagging materials removed. An Amber Radiance IT infrared
camera, with a 13 mm lens and a 12" x 12" field of view, was used to image the caps after flash heating. A Thermal
Wave Imaging EchoTherm data acquisition system was used to record the digital images from the camera to files on
a PC hard drive. A Thermal Wave Imaging flash hood set to deliver 6.4 kJ of energy was used to rapidly heat the
caps prior to image acquisition.
An analog thermography system was also used for some of the cap inspections. The analog system, although lower
in sensitivity than the digital system gave real-time images of the uncured bond line and as such gave the fabrication
team more time to work out any trapped air. Here, an Inframetrics SC-IOOO infrared camera was used to image the
caps after slow heating by way of a pair of 500W hand-held quartz shop lamps. Thermal images from the camera
were recorded to VHS tape.
Reinforcing Cap Defects
Inspection of the wet bond line may have also detected defects in the reinforcing caps above the bond line. Defect
standards were fabricated to simulate defects in the reinforcing caps themselves. Inserts were fabricated into sample
panels between plies at various depths to simulate internal unbonds. The images shown in Figure 4 were acquired
with the digital system previously described, with the exception that a 25 mm lens (6"x6") field of view was
employed. In the images the depth of the insert is given as a fraction of the panel thickness (T). Three insert
diameters were used including 0.50, 0.25 and 0.125 inch.
Figure 4. Thermograms of defect standard simulating disbonds in reinforcing caps.
Bond line Defects
As a method to simulate unbonds during cure a defect panel was fabricated with inserts in the wet bond line. The
simulated void shown thermographically in Figure 5 was produced by inserting a 0.5 inch diameter rubber o-ring
between the simulated cap and underlying material. The cap was covered with a layer of release ply and vacuum
bagged. Narrow strips of breather cloth were used in this instance rather than the split plastic breather tubes. The
part was then inspected by heating the surface with a pair of shop lamps for about 10 seconds. Note how the void
trapped within the o-ring remains hotter than the surrounding bond line demonstrating the presence of a void.
As previously described, the initial inspections were performed when the caps were still under the vacuum bag. In
this manner, voids or air pockets, trapped under the cap could be eliminated before the cure was complete or if the
void was less than a previously determined critical size it could be accepted. Figure 6 demonstrates a thermogram
of a cap section with two small, but acceptable, voids. When the part was cured and the bag removed a second
inspection was performed (Figure 6b) revealing the first two known defects (B and C) as well as on that was hidden
(A) beneath the breather ply strip.
Figure 6. Thermogram showing initially hidden void.
On average it took six image sequences to fully cover each cap. The results of thermography inspection of a
reinforcing cap which did not reveal any unacceptable unbonds are shown in Figure 7. The images shown here were
obtained after the bond line had cured and the vacuum bag had been removed. Minor temperature variations exist
due to differences in the thickness of a foam filler material beneath the cap. A grid of hot points is visible where
vent ports were drilled in the reinforcing cap to prevent the entrapment of air between the foam and the cap.
The results of thermography inspection of a reinforcing cap which revealed an unacceptable unbond are shown in
Figure 8. Again, these images were obtained after the vacuum bag had been removed. A branching unbond is
visible in the upper left sector with a maximum length of approximately 4.4 inches. The unbond was just barely
visible in the thermograms taken before the vacuum bag was removed being all but totally obscured by breather
tubes and cloth that was placed around the outer edge of the panel.
Since this defect was unacceptably large an action was taken to try to fill the void with epoxy. To aid in the repair
process lead foil tape markers were placed on the cap identifying the exact location of each leg of the void and
verified by re-inspecting the region. Holes were then drilled in the reinforcing cap over the void and injected with
epoxy to fill it. Thermograms were taken at various stages in this process are shown in Figure 9. After the third
injection of epoxy, the size of the void had been diminished sufficiently that it was deemed acceptable.
The same process of thermographically inspecting each cap during and after cure was performed for the rest of the
tank. Defects found after the post cure inspections were measured and if deemed necessary filled with resin. A final
post repair check was then performed to complete the inspection process.
Figure 8. Sector inspections of a reinforcing cap with an unacceptable 4.4" long unbond (upper left).
H&SA MSFC ED32
Second attempt Post Repair (Third attempt)
Figure 9. Repair of a void in the wet bond line by epoxy injection.
Thermographic inspection proved largely successful in the detection of voids in the wet bond line between
reinforcing caps and underlying foam filler on the X-33 LH2 Tank 2. Voids that were detected precure were often
eliminated by rubbing them out or removing and reshaping the cap. There were some undetected voids that had
either opened up after the initial thermographic inspection or were obscured by the breather tubes used in vacuum
bagging. Epoxy was injected to fill these voids, resulting in an acceptable repair.
The authors would like to acknowledge the Lockheed Martin and Alliant Techsystems composite fabrication crew
and Don Bryan (NASA MSFC Thermodynamics and Heat Transfer Group) for their assistance in this effort.