Thermographic scan of a Boeing 737 fuselage shows waffle-doubler disbonding. Source: Thermal Wave Imaging Inc.

Thermography is a well-established technique for predictive and preventative maintenance (P/PM), where the inspector uses the infrared (IR) camera in a passive mode to measure the steady state surface temperature of a component during its normal operation. Malfunctions that affect performance, such as missing or wet insulation in a building, faulty bearings in a rotating machine or a short circuit in an electrical switch box, will cause anomalous temperature indications to appear in the IR image. However, for nondestructive testing (NDT) applications, where detailed information about the internal structure is required, thermography must be applied in an active mode, where a thermal stimulus is applied to the test piece and the IR camera monitors the change in temperature as the part returns to its original temperature.

Until recently, active thermography was considered to be a qualitative adjunct to more familiar NDT methods such as ultrasound or radiography. Despite several attractive attributes-the process is noncontact, requires only single-side access to the test piece and is capable of inspecting curved as well as flat surfaces-deployment of thermography for NDT has been infrequent, limited to noncritical inspections performed by specialists. However, because of a new generation of inspection systems, thermography is becoming increasingly popular as a primary NDT method. In applications ranging from routine inspection of the leading edge of the NASA Space Shuttle to quality assurance in the manufacture of aircraft engine components, helicopter rotor blades and land-based turbine blades, thermography has been used as a stand-alone inspection method, often replacing ultrasound or radiography.

 



Leading edge delaminations are shown in this reinforced carbon-carbon (RCC) wing panel. Source: Thermal Wave Imaging Inc.

Flash Thermography

The most widely used, active mode NDT method is flash thermography, in which the surface of the test piece is heated by a light pulse lasting only a few milliseconds. Under normal conditions, the part cools after flash heating, as the heat deposited at the surface flows toward the cooler interior. However, internal anomalies in the test piece, such as voids, inclusions, delamination, moisture, or changes in thickness or density, cause changes in the cooling rate at the surface. These changes are transient and, in many cases, too small or subtle to be observed directly in the IR camera image. <p>

With early thermographic NDT systems, the inspector would view IR images of the sample surface as a continuous movie, or at various times during the cooling sequence, in order to identify areas of contrast between an indication of a subsurface flaw and the flaw-free background. In practice, sensitivity of this approach was limited, and it was only useful for finding large, severe flaws that were smaller than the field of view. Image processing techniques and signal averaging could be applied to the data to improve the appearance of the images somewhat, but the overall performance of these systems was unremarkable. <p>



Modern flash thermography systems are derived from thermographic signal reconstruction (TSR), a signal processing technique based on the physics of thermal diffusion. Source: Thermal Wave Imaging Inc.

Signal Processing Makes the Difference

Modern flash thermography systems are derived from thermographic signal reconstruction (TSR), a signal processing technique based on the physics of thermal diffusion. Each pixel in the IR camera image is treated as a separate time history of the cooling behavior of a point on the sample. A pixel is converted to an equation that reveals specific local characteristics of the sample, in a manner analogous to an ultrasonic A-scan. As a result, it is possible to extract information from a single pixel, without referring to an image or adjacent pixels. The significant pixel characteristics are based on the rate of change of cooling, and not absolute temperature, which may vary with ambient conditions or inspection variables; as a result, evaluation of pixel data is based on numerical values that remain constant, independent of flash energy, camera calibration or other variables that are likely to change between inspections.

Results obtained from TSR allow detection of features that are not detectable in the unprocessed data sequence. Because data is evaluated based on the characteristics of each pixel, rather than an image, it also is possible to identify indications that are larger than the field of view, for example, in cases where the test piece is entirely good or bad, so that no contrast is apparent.

Figure 2: Thermal response cure (2nd derivative) of a two pixels – one showing heat damage (blue) in an aircraft composite component and the other no damage (red). For most NDT applications, inspection of a single square-foot area requires a 5-second data acquisition period after flash heating, which generates a digital IR image file that is approximately 50 MB. However, the data can be regenerated by using the coefficients of the TSR equations, which only require 5 MB; as a result, it is not necessary to store the original once the TSR conversion is complete.

Aside from the reduced storage requirement, the smaller file size makes it possible to examine several data sets simultaneously, so that inspection results from large structures such as aircraft control surfaces, which may involve several hundred individual sequences, can be viewed as a single result. The same capability has proven useful in manufacturing quality assurance, where an entire production run of nominally identical parts can be evaluated simultaneously.



Figure 2: Thermal response cure (2nd derivative) of a two pixels – one showing heat damage (blue) in an aircraft composite component and the other no damage (red). For most NDT applications, inspection of a single square-foot area requires a 5-second data acquisition period after flash heating, which generates a digital IR image file that is approximately 50 MB. However, the data can be regenerated by using the coefficients of the TSR equations, which only require 5 MB; as a result, it is not necessary to store the original once the TSR conversion is complete.

Aside from the reduced storage requirement, the smaller file size makes it possible to examine several data sets simultaneously, so that inspection results from large structures such as aircraft control surfaces, which may involve several hundred individual sequences, can be viewed as a single result. The same capability has proven useful in manufacturing quality assurance, where an entire production run of nominally identical parts can be evaluated simultaneously.



Defining the Limits

Like every NDT method, thermography is ideal for some applications, while other applications may be better served by different techniques. Although enormous improvements have been made in flash thermography technology, there are fundamental limits that define the boundaries of how thermography may be applied. Unlike radiography or ultrasound, where energy injected into the sample travels in a straight line and reflects off surfaces, thermography interrogates the interior of a sample by heat conduction, a diffusion process that requires a more complex, statistical description. The implication of the diffusion process on thermographic NDT is that the size of the smallest detectable feature increases as a function of its depth beneath the surface. A guideline for determining minimum detectable defect size is shown in Equation 1.



Equation 1.



This rule of thumb is an approximation that does not consider signal processing, which may improve results, or the nature of the flaw, sensitivity of the camera or surface condition of the sample-all of which may degrade results. However, it does convey a general sense of the range over which thermography applies. For example, it is apparent from Equation 1 that detecting a flaw with diameter = 1 centimeter at a depth of 2.5 millimeters is reasonable because the diameter-to-depth ratio is 4, in this case. However, the same 1-centimeter-wide flaw at a depth of 2.5 centimeters does not meet the criterion of Equation 1, so that detection is not likely to be feasible.

In situations where inspection time is critical, the time required to inspect a sample with particular thickness L does not depend on the instrumentation, but rather, on the thermal diffusivity (a) and thickness of the test piece, according to the relationship shown in Equation 2.



Equation 2



For many metals, which have relatively high thermal diffusivities, a 0.25-inch thick plate can be inspected in less than 1 or 2 seconds. However, composites and polymer materials have lower diffusivities, and may require 10 or 20 seconds to interrogate that same thickness. Equation 2 also indicates that as thickness increases, the inspection time increases as the square of the thickness; so, if a 1-centimeter part requires 5-second data acquisition, a 2-centimeter part will require 25 seconds.



Seamless mapping of thermographic image sequences is performed using TSR software. Source: Thermal Wave Imaging Inc.

Thermography in Action

In a typical NDT inspection of a large structure, the area to be inspected is mapped out in advance, using chalk or reflective foil tabs to mark locations where the flash thermography hood is to be placed. The same mapping is replicated in the TSR software program, so, as the inspector moves the hood and acquires data from each location, the new data is automatically added to the TSR image. The software can be set to identify indications based on anomalous cooling behavior, or it can be calibrated to measure wall thickness or thermal diffusivity.

As the popularity of thermography for NDT continues to grow, standards for inspection and operator training are evolving, so that thermography itself is evolving from an exotic technique to the NDT mainstream. Advances in thermographic signal processing have significantly improved the sensitivity and accuracy of the NDT process and removed dependence on subjective interpretation of images by the operator. The shift toward quantitative analysis of inspection data has led to the recent introduction of thermographic measurement systems, which measure defect depth, wall thickness or thermal diffusivity with accuracy comparable to traditional methods. As progress continues, systems capable of automated quality control and defect detection are being introduced. ndt

Steven Shepard is president of Thermal Wave Imaging Inc. (Ferndale, MI). For more information, call (248) 414-3730 ext. 310, e-mail [email protected] or visit www.thermalwave.com.