NASA pinpoints thermography as one NDT technique to help get the shuttle flying.

NASA has tested the thermograph system as a way to in situ test the wings of a space shuttle between flights. Photo: Thermal Wave Imaging

An infrared imager captures data in a test of the thermographic system. Photo: Thermal Wave Imaging

While President Bush lays out lofty goals for the future of the National Aeronautics and Space Administra-tion, scientists and engineers are working at more down to earth goals, as the agency tests and analyzes nondestructive test (NDT) and other inspection technologies as part of the effort to return the shuttle to space.

The agency has identified several nondestructive technologies to look for potential damage or degradation that are not visible to the naked eye: ultrasound to look for delaminations and voids, eddy current to look for localized oxidation, X-ray to look for cracks, and thermography to look for delaminations or cracks.

Flash thermography is expected to play a large part in inspecting the leading edge of the reinforced carbon-carbon (RCC) panels. The RCC panels that comprise the wings were damaged when a piece of insulating foam broke loose from the shuttle Columbia during launch last January. The foam struck the leading edge of the wing causing a breach, which allowed superheated gases to enter the wing during reentry and led to the destruction of the craft.

Thermographic techniques will be used prior to a shuttle launch. According to NASA's Return-to-flight plan, the individual shuttle components are to be nondestructively evaluated by the suppliers before shipping to the Kennedy Space Center (KSC, FL). Thermography techniques will be used on site to establish a baseline and compare and correlate this data to the vendor nondestructive evaluation (NDE) data. Between launches, thermography will play a big role in inspection of shuttle components in situ.

While still going through testing stages, NASA expects to use both ultrasound and thermography on the shuttle Atlantis, the next shuttle to go into space, tentatively scheduled for launch in the fall.

NASA scientist Dr. William Winfree is an NDT expert who works out of the Langley Research Center (Hampton, VA) and has worked with thermographic techniques for more than 18 years. Thermography, Winfree says, is "particularly good for finding flaws close to the surface, such as a void behind the RCC silicon carbide outer layer."

The silicon carbon outer layer protects the carbon-carbon material underneath it from the extremes of space, but it also prevents inspectors from seeing any significant oxidation that might have occurred. "One of the concerns is that the carbon carbon becomes oxidized and leaves a large cavity," Winfree says. "You don't know where it is going to happen."

When it becomes oxidized, the silicon carbide flakes in chunks of about 1/4 inch by 1/4 inch square. Previously, the part was inspected by running a magnetic glove across the surface.

Thermographic testing was conducted on RCC specimens last fall by Langley researchers. At least 10 specimens were used, which had voids underneath the surface as a result of being placed in an arc jet, and a blind specimen in which the researchers did not know what defect it had. "We had a series of specimens that we made that had flat bottom holes, which the thermographic technique was to detect," Winfree says. "We looked at pretty much anything we could get our hands on, including some panels that had been used at the impact test at San Antonio."

In the impact tests in San Antonio, TX, foam insulation was shot at RCC samples causing breaches. These tests were used by the Columbia Accident Investigation Board to declare the insulation strike as the physical cause of the accident.

A thermographic analysis of an RCC sample after an impact test reveals a crack. Photo: NASA


Thermography, or in this case flash or pulse thermography, is a process whereby a thermal stimulation is given to a part at the surface and data is collected as the part cools. Flash lamps direct a thermal pulse at the component striking it for about 2 milliseconds and delivering a small amount of heat, about 20 degrees above ambient, to the specimen. "As the heat moves from the surface into the part, we can infer what is going on inside the part, both structurally and in terms of material properties," says Dr. Steven Shepard, a physicist and president of Thermal Wave Imaging (TWI, Ferndale, MI).

TWI was awarded the contract to assist NASA in developing procedures and best practices for thermographic inspection of the thermal protection system (TPS) on the space shuttle. NASA scientists have been testing the TWI EchoTherm system, which was developed as part of a Phase II Small Business Innovative Research Contract (SBIR) with NASA and the U.S. Navy. The EchoTherm system, working in tandem with the company's Thermal Signal Reconstruction (TSR) software, measures the thermal energy on the surface as it is conducted into the cooler interior of the sample. The surface temperature drops over time, usually in a uniform manner and subsurface defects affect the cooling rate as compared to the areas that do not have the defects.

Every object, those that are not at absolute zero temperature, gives off infrared radiation, and infrared thermography uses a special camera and computer to detect the minute variations in heat given off the structure. The pulse thermographic system uses two flash lamps, enclosed in a hood, with combined pulse energy of 12 kilojoules to heat the surface of the component being tested. An infrared camera, with a 320 by 256 focal plane array detector is used to image the component. The camera operates at a 120 Hz frame rate and the TWI data acquisition system stores images as 16 bit, 320 by 256 data files.

Why thermography?

A shuttle wing presents a number of problems including its size and shape, the material characteristics and the uncertainty of when and where oxidation might take place. "The RCC is basically a three-layer structure and the situation with the shuttle wing is that one doesn't know exactly which layer or interface is the important one to look at, so you have to look everywhere."

The RCC panels are made from non-homogenous materials and Winfree says that the differences in the thermal coefficients between the materials make thermography particularly useful.

"The different coefficients of expansion affect the thermal response of the material. That is one of the things we are looking at: ‘how do you treat that variability that you have from spot to spot, and how does that limit your detectability?"

The real advantage of thermography is that it is fast, Winfree says, which is important because of the size of the panels. "We can do square feet in seconds, while most other techniques tend to be a lot slower."

On simpler parts, adds Shepard, it might be a matter of simply heating the surface and looking at the image data or applying conventional image processing techniques. "With the shuttle components, we are looking at the entire cooling signature and analyzing the entire data sequence. We are not looking at a series of images; we are looking at a set of data points in time. Each pixel is an entity unto itself."

The thermographic system plots the pixels as a curve which is then analyzed. This allows researchers to circumvent noise and contrast problems that could occur because of the complexity of the material.

Through analysis of reams of data, scientists know the thermal signature of a good piece of RCC and they understand what the time evolution of the pixels should be as it occurs during the 12 seconds between the flash of the lamps to the end of the data collection period.

"We do not compare one image to another," Shepard says, "These are not homogenous materials and there is lots of variation in them, but we understand what the significant interfaces and sources of variance are, and what the resulting cooling curve should look like. As a result, we can identify those curves that fall outside of the normal variation of the material properties."

According to Winfree, the silicon material impedes the inspection because it tends to dominate the signal and act as a kind of mask. Data acquisition systems help to unmask the data. "The most significant change to thermography has been in the data acquisition systems. When we first got into it, we were video taping the images that we got off the infrared imagers," he says.

The IR images were a single detector that Winfree and the other users would sweep back and forth over the region of interest. "Now, we have focal point arrays, whole arrays of charge coupled device (CCD) cameras. The signal to noise is a lot better than it used to be. The data is acquired digitally. The computer also controls the flash heating, which wasn't possible to do before. It is a much more controlled inspection than 20 years ago and we can get significantly more data."

Evaluating the voluminous amount of data used to be a laborious challenge, if not a prohibitive one. Mathematical equations were developed that improved this process. TWI's solution is Thermographic Signal Reconstruction (TSR).

"In the case of the leading edge, each shot we take generates between 120 and 130 megabytes of data," Shepard says. "Our TSR process converts each pixel to an equation, which takes that 130 megabytes of data and turns it into 4.5 megabytes of data. As a result, we can look at many images at once and process them simultaneously. We can create a single image that might actually represent a gigabytes worth of data, but look at it all at once. Instead of looking at one square foot of the wing, we can look at one square meter, manipulate it and process it because it is not just a set of pictures, it is real data."


  • NASA plans to use a combination of NDT technologies to inspect the shuttle in situ. Flash thermography is one of the main technologies expected to be used.
  • Flash thermography directs a pulse of heat at the specimen. The heat travels into the part and how it cools can be analyzed to detect flaws.
  • Thermography is good for finding defect close to the surface.
  • During the 12 seconds during which data is collected, about 130 megabytes of data are collected. Software converts this to a more manageable 4.5 megabytes.