Since its introduction in the 1980s, the field of Flash Thermography for Nondestructive Testing (FTNDT) has benefited from enormous advances in the underlying IR camera and computer technologies that enable it, while the flash excitation component has remained essentially unchanged. A typical FTNDT setup comprises a pair of helical or U-shaped xenon flashlamps designed for use in commercial photography, each powered by a bank of capacitors under computer control. A single flashlamp-power supply unit is often described as 4-6 kJ of energy with duration 2-3 msec, the Full Width Half Maximum (FWHM) duration of the flash. However, saturation of the IR camera detector may persist considerably longer due to the extended tail of the flash pulse. In high thermal diffusivity materials, the presence of saturation may mask features of interest, and limit access to early onset signals normally used for depth measurement of subsurface features. Saturation may be mitigated to some degree, but not entirely, by reducing flash energy (power supply voltage). However, more effective elimination of saturation is accomplished using a dedicated hardware device to truncate the duration of the flash pulse. In this paper, we compare the effect of varying flash energy by adjustment of power supply voltage and flash duration on detection of near surface features in an aluminum plate.
Although active thermography has traditionally been regarded as a qualitative NDT method, its potential for quantitative
measurement of thermophysical properties including wall thickness, flaw size and depth, thermal diffusivity or effusivity
has been the subject of numerous investigations. Enabled by improvements in IR camera technology and fast, abundant
and inexpensive computing power for advanced signal processing, measurement results have been reported using a
variety of excitation and signal processing schemes. Results are often presented as a correlation between thermography
data and nominal properties or independent measurements by another "validated" method. However, given the diffusion
mechanism that underlies thermography, and the quantization and sampling conditions implicit in using an IR camera as
a temperature sensor, there are definite limits to what can be achieved in a thermography measurement. While many
benefits can be achieved with improved instrumentation, efficient energy insertion or optimized signal processing,
ultimately, the limits imposed by diffusion and instrumentation take precedence, and cannot be circumvented. In this
paper, the effects of camera frame rate and sensitivity on measurement of the thickness of a slab are examined, using
basic 1-dimensional diffusion approximations.
The use of thermography as a nondestructive testing method has increased significantly in recent years. However, implementation is usually based on visual interpretation of image results by a trained inspector. Although this is accepted practice in the aerospace industry, where thermography is widely used, it is not appropriate for automotive manufacturing, where higher production rates demand higher inspection throughput that can only be accomplished with automated detection. The logarithmic derivatives of flash thermography time histories provide an excellent basis for automated detection. A unique model, based on the logarithmic derivatives of a series solution for the surface temperature of a flash heated plate with adiabatic boundary conditions, provides a template for identifying pixels that deviate from uninterrupted diffusion that is characteristic of a defect-free sample. We demonstrate automated defect detection using this model on a composite sample.
To reduce costs and facilitate automation in the automotive industry, adhesive bonding has gained popularity as a replacement for conventional mechanical fasteners such as bolts, screws, rivets, and welding. Adhesive bonding is particularly useful for bonding parts made of plastics and polymer composites, which are playing an increasing role in reducing vehicle weight. However, the adhesive bonding process is more susceptible to quality variations during manufacturing than traditional joining and fastening methods. Shearography and pulsed thermography are full-field, noncontact, nondestructive testing methods that are widely used in the aerospace industry, offering significant potential as practical tools for in-process inspection of adhesive bond quality. The two techniques are often used to address a common set of aerospace applications, e.g., delaminations or skin-to-core disbonds in composite structures. However, they are fundamentally different, based on different flaw detection mechanisms: Shearography measures the sample's mechanical response to mechanical stresses, while pulsed thermography measures the sample's thermal response to an instantaneous thermal excitation. For the convenience of potential users and readers, the authors review shearography and pulsed thermography. The potential of these techniques for inspecting adhesive bonding is demonstrated and compared.
Since its introduction 5 years ago, the Thermographic Signal Reconstruction (TSR) method has become a widely accepted approach to processing and interpreting active thermography data. Although the technique generates a noise free replica of an experimental data sequence, it is the time derivatives of the replica that have proved to be most useful. In particular, the second derivative of the TSR signal offers increased sensitivity, noise and artifact suppression and reference-free evaluation and quantification of results. Unlike contrast methods, where a point is evaluated by comparison to a defect-free reference, the TSR second derivative of a single pixel indicates the presence of a subsurface defect, or in the absence of a defect, it can be used to measure the local thermal diffusivity.
Flash thermography is widely used to inspect Thermal Barrier Coatings (TBC) during manufacturing and maintenance for defects such as delamination or contamination. However, attempts to use thermography to quantify TBC thickness have been less successful. In conventional thermographic NDT applications, the sample surface is opaque to an incident light pulse, and highly emissive in the infrared. The situation is more complex in TBC's, as the coatings are translucent to visible light and near-IR radiation (including the IR component of the flash). Furthermore, TBC's are translucent to the mid-IR wavelengths at which many IR cameras operate. Thus, in the absolute worst case, the flash pulse does not heat the coating, and the camera does not see the coating. Although the latter problem can be mitigated by judicious choice of camera wavelength, it must also be recognized that both the heating and cooling mechanisms in a flash-heated TBC are different from the usual thermography model, where transit time of a heat pulse from the sample surface to a layer interface is an indicator of coating thickness. The resulting time sequence is processed using the Thermographic Signal Reconstruction to generate thickness maps which are accurate to an accuracy of a few percent of the actual coating thickness.
In conventional flash thermography a brief pulse of light with a full width half maximum duration of 2-4 ms is applied to the surface of a solid sample and the surface temperature response is recorded with an infrared camera. In practice, the flash duration is typically fixed, and the amplitude is the only adjustable flash parameter. Flash amplitude is normally adjusted to provide maximum illumination of the sample surface without causing saturation of the camera detector array. However, more precise interrogation of the subsurface structure is obtained if the timing parameters of
the flash excitation and the detector are carefully determined and controlled. In particular, limiting the pulse duration and the offset between the pulse and the detector integration time significantly increases correlation between modeled and experimental results during the early post-flash frames. Additionally, a new method is described to precisely detect the initiation and duration of the pulse for common high performance infrared cameras.
In various studies, thermographic methods have been used to measure thermophysical properties of materials. The most widely used such method is the Parker flash technique for diffusivity measurement, in which the transit time of a heat pulse applied to the front face of a sample of known thickness is measured by observing the temperature at the rear surface. In recent investigations, there has been considerable emphasis on single-sided techniques for materials
characterization. Typically, quantitative analysis using a single-sided thermographic approach requires some a priori knowledge about the sample, such as thickness, thermal diffusivity, or perhaps a calibration standard with back drilled holes with known diameter and depth. In fact, in certain cases it is possible to use single side pulsed thermographic data to measure, or at least estimate, properties such as thickness, thermal diffusivity and subsurface feature depth with no a priori information about the sample.
Sonic, or thermosonic nondestructive testing, which is based on the vibrothermography method introduced in the late 1970’s, has attracted a great deal of recent interest as a means for detection of cracks that were previously considered to be undetectable using thermographic inspection methods. Excitation of a solid sample with bursts of high-energy (500 - 3000 Joule), low-frequency (10 - 50 kHz) acoustic energy has been demonstrated to be effective in generating transient localized heating at crack sites, making them detectable by an infrared camera. Despite the apparent simplicity of the scheme, there are a number of experimental considerations that can complicate, or in some cases even prevent, the implementation of vibrothermography-based inspection. Factors including acoustic horn location, horn-crack proximity, horn-sample coupling, and effective detection range all significantly affect the degree of excitation (or whether any excitation occurs at all) that occurs at a crack site for a given energy input. In cases where the experimental objective is precise measurement of crack length, the method used to visualize the data from the IR camera and its optic must also be taken into consideration.
KEYWORDS: Infrared cameras, Thermography, Signal to noise ratio, Cameras, Signal processing, Image enhancement, Image processing, Diffusion, Infrared imaging, Data acquisition
KEYWORDS: Thermography, Visualization, Infrared cameras, Image processing, Cameras, Signal processing, Signal to noise ratio, Nondestructive evaluation, Temperature metrology, Infrared imaging
Analysis and processing of thermographic NDT data has been primarily based on either visual analysis of single images acquired during the cooling sequence, or examination of a contrast curve based on comparison to a defect free reference region. Analysis of the sequence is complicated by the fact that the surface temperature of the sample decreases monotonically over a large dynamic range, yet the temperature differences generated by subsurface features may be small by comparison. A simple linearization operation can be performed on the entire sequence so that the time history of each pixel behaves in a manner similar to that of the contrast curve. The linearized data allows visualization of the entire sequence in a fixed dynamic range that is small enough to allow visualization of weak features, as well as presentation of the data in a line slice manner similar to an ultrasonic b-scan.
KEYWORDS: Thermography, Infrared cameras, Infrared imaging, Cameras, Nondestructive evaluation, Diffusion, Defect detection, Neodymium, Data acquisition, Signal to noise ratio
Conventional methods for analysis of pulsed thermographic NDE sequence data are highly susceptible to noise, nonlinearity of the IR camera response, and the presence of surface features on the sample. Furthermore, the ability of conventional methods to significantly improve the ability to retrieve deep or weak subsurface features beyond the original unmodified image is limited. We have developed a Thermographic Signal Reconstruction (TSR) technique that enhances defect to background contrast, increases the depth range over which pulsed thermography can be applied, and reduces the amount of blurring due to lateral diffusion that is typical of thermographic imaging. The TSR approach also reduces the amount of data that must be stored by an order of magnitude. The reduction in size of the data structure allows simultaneous manipulation of data from numerous locations on a sample, so that fast parallel processing of large structure data is possible. The results of the parallel processed TSR data consistently offer higher spatial resolution, less blurring and more precise depth and size measurement than the original data. Examples on aircraft and power generation components will be presented.
Visualization and analysis of pulsed thermographic data for NDT has generally been based on simple image averaging, subtraction or slope operations. Quantitative contrast methods, based on comparison to a defect free reference point or region, have also been used to a lesser extent. Despite their widespread use, all of these methods are highly susceptible to noise, nonlinearity of the IR camera response, and the presence of surface features on the sample. More importantly, the ability of any of these methods to significantly improve the ability to retrieve deep or weak subsurface features beyond the original unmodified image is limited. In a previous paper, we introduced the concept of Thermographic Signal Reconstruction (TSR) as a means of enhancing defect to background contrast while reducing the amount of data that must be stored by an order of magnitude. The TSR method increases the depth range over which pulsed thermography can be applied, and also reduces the amount of blurring due to lateral diffusion that is typical of thermographic imaging. In this paper we compare TSR with conventional thermographic approaches and consider the mechanisms for the resulting performance improvements.
KEYWORDS: Signal processing, Thermography, Cameras, Nondestructive evaluation, Data storage, Image processing, Interference (communication), Signal to noise ratio, Data acquisition, Data conversion
Until recently, thermographic methods for NDE have generally been quantitative, relying heavily on operator interpretation of image data. Although quantitative methods have been developed, they have generally required a priori knowledge of the sample physical properties, or identification of a defect free region within the field of view. Recent advances in pulsed thermography allow reference-free measurement of defect size, sample thickness and material properties without operator intervention or a priori knowledge of sample properties. An essential component of these advances are new signal processing methods based on both the spatial and temporal thermal response of the sample surface temperature to an instantaneous heat pulse. These methods provide a significant reduction in noise and blurring due to lateral diffusion of heat, and effectively increase the maximum penetration depth and spatial resolution beyond that of conventional thermography.
Although acceptance of pulsed thermography as a tool for nondestructive inspection continues to increase, standards and metrics for performance assessment and procedure development have been slow to follow. Consequently, practical application development is often left to the experience and intuition of the supervisor, and decisions such as heating input power, surface preparation, acquisition time, camera wavelength, integration time, and sensor type are made on a subjective, qualitative basis. Furthermore, quantitative thermographic methods that have been reported generally rely on the ability of the operator to identify a defect free region of the sample, as well as high-speed laboratory cameras that are not practical for many field inspection applications. Because of these limitations, practical application of pulsed thermography is often limited to use as a qualitative complement to conventional point inspection methods (e.g. ultrasound). We have developed a metric for characterization of active thermographic system performance that defines a Thermal Modulation Transfer Function (TMTF), which allows complete characterization of the IR NDE system performance for a given sample type. The TMTF approach allows accurate prediction of peak contrast times, and best-case contrast for defects of a given diameter and depth. We will use the TMTF approach to demonstrate how detection limits and ranges can be established, and how these limits can be improved using TMTF based signal processing algorithms.
In current thermographic NDE practice, the detection limits of various methods are typically undefined, beyond citing the deepest flat bottom hole that has been successfully detected in a particular material. Little distinction is made between the ability of a thermographic system to detect the presence of a subsurface defect, to measure its physical properties, or to resolve adjacent defects. Although many practitioners rely on a `rule of thumb', which states that the aspect ratio (diameter/depth) of a detectable defect must be greater than 1, it has been shown to be unreliable as a basis for determining the feasibility of a particular inspection. It is possible to characterize the performance of a thermographic system using a simple procedure, so that the ability of the system to detect, resolve, or measure defects in a particular material or sample type can be successfully modeled. The intrinsic detection limits derived from this approach effectively defines a thermal point spread function that defines the minimum detectable defect size at a particular depth. This information can be used to remove blurring due to lateral heat flow, and extend the depth at which resolution is possible.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.