Nondestructive testing with thermography

Clemente Ibarra-Castanedo; José Ricardo Tarpani; Xavier P V Maldague

doi:10.1088/0143-0807/34/6/S91

1. What is thermography?

Thermographic testing, thermal imaging, infrared (IR) thermography or simply thermography is a nondestructive testing (NDT) imaging technique that allows the visualization of heat patterns on an object or a scene. The theoretical and experimental aspects of thermography are covered in [1–6]. Vollmer and Möllmann [1] present a complete chapter dedicated to describe a number of applications oriented to teaching physics. Indeed, thermography constitutes a valuable tool for teachers and students willing to understand a wide variety of physical phenomena, such as the transformation of mechanical energy into heat or the selective absorption and emission of gases, among several others [1, chapter 5].

In the field of NDT, Meola et al [2], Holst [3] and Maldague [4] widely cover the basic theory and the possible applications for the inspection of materials. Interested students and professors can benefit from these resources either as introductory or advanced courses of thermography. In addition, [5] presents a formal review of the different techniques and applications from several of the most recognized researchers in the field.

The basic equipment includes an IR detector, a monitor to display images and a PC to record (and sometimes process) data. There are different types of detectors, from mono-element sensors to multi-element (focal plane arrays or FPAs) cameras, with the latter becoming more and more popular given that they provide directly a picture or a video of the scene being observed and that their operation is relatively straightforward. (Some IR cameras are close to digital cam recorders with specific functionalities for thermal patterns.)

Thermography is among the seven methods referred by the European (EN 4179) [7] and American (NAS-410) [8] standards for NDT personal qualification. Hence, qualified personnel in areas such as aerospace are able to choose and use thermography for the NDT of components at different stages (production, operation and maintenance).

A formal definition has been proposed as follows [5 6]: 'Infrared thermography is a nondestructive, nonintrusive, noncontact technique that allows the mapping of thermal patterns, i.e., thermograms, on the surface of objects, bodies or systems through the use of an infrared imaging instrument, such as an infrared camera'.

This definition covers a wide variety of applications and many possibilities exist to carry out a thermography inspection as explained in the following section.

1.1. Inspection scenarios

Figure 1 presents a diagram summarizing the different elements to take into account when designing an NDT inspection scenario. It is divided in six columns (from A to F), which contain the different possibilities that can be employed in order to define an inspection scenario after answering to some basic questions as discussed hereafter.

1.1.1. Approaches: is an external energy source required?

The first column in figure 1 includes the two possible approaches in thermography:

passive, in which materials and structures are naturally at different (higher or lower) temperature than the background, e.g. the human body is normally at a temperature higher than the ambient, hence it is easily detected by an IR camera without any additional stimulation and
active, in which an external stimulus is needed in order to produce a thermal contrast in the object surface, e.g. an object containing internal defects (such as voids, delaminations, foreign material inclusions, etc) will require submission to a thermal disequilibrium in order to produce distinctive surface thermal patterns between the defects and the sound material that can be detected with an IR camera.

Figure 2 schematizes the configuration for these two approaches. For the passive approach (figure 2, top), an object or a system ${\epsfbox{images/ejp474034un01.eps}}$ , a human hand in this case, having a distinctive thermal contrast with respect to the environment can be monitored using an IR camera $\smash{{\epsfbox{images/ejp474034un02.eps}}}$ without any additional energy input. A computer system ${\epsfbox{images/ejp474034un03.eps}}$ is used to display and process images. The setup is very similar for active IR thermography (figure 2, bottom) with the difference that an energy source ${\epsfbox{images/ejp474034un04.eps}}$ is required to generate a thermal contrast between the feature of interest and its surroundings, two internal foreign material inserts on a tile specimen in the case.

The passive approach is often qualitative, such as the diagnosis of the presence of a given abnormality or hot/cold spots with respect to the immediate surroundings. Typical applications are the monitoring of electrical and electronic components or the detection of humidity or insulation problems in buildings. Quantitative analysis is also possible such as in the inspection of civil engineering and cultural heritage structures using solar loading cycles and water ingress characterization in aircraft structures upon landing among others.

Active thermography finds a large number of applications in NDT since practically any form of energy can be used to stimulate the inspected object, provided that the thermo-physical properties of the eventual defects are different enough to the non-defective areas in order to produce a measurable thermal contrast. Besides, the time of application of the external stimulus can be synchronized with the acquisition, providing the possibility of developing quantitative data analysis, e.g. the time of appearance of a subsurface defect is proportional to its depth.

1.1.2. Configurations: is the camera, the source or the object in motion?

It is also possible to adopt two different configurations, as shown in the second column of figure 1, these are as follows:

static, where the camera, the object and the source remain fixed during the inspection and
dynamic, where at least one of these three elements is in motion.

There are some differences in the manner in which data are acquired and processed depending on the selected configuration. Most of the signal processing techniques that are discussed below are based on a static configuration since, in this case, the time history for every point at the surface can be straightforwardly tracked.

The static configuration, although very convenient for data analysis, is not always practical. Complex-shaped, rounded or large components are not easily inspected from one fixed spot. In these cases, dynamic inspection could provide a complete portrait of the component's condition faster and better. For instance, a laser beam could be employed to scan (heat) a surface point by point, while the object and the camera are fixed or a camera and a lamp could be mounted on a robot to inspect (in motion) the surface of a static object. Furthermore, sometimes the objects to be inspected are already in motion such as in people tracking or in industrial production lines where halting the process for inspection might be proscribed. It is also possible to reorganize data obtained from a dynamic configuration in order to produce pseudo-static data sequences to which static signal processing techniques can be applied (if needed).

1.1.3. Modes: where the energy is transferred or generated from?

In the third column of figure 1, there are three modes in which the energy is transferred to the inspected object or generated from it with respect to the camera, these are as follows:

reflection, in which energy is delivered to the specimen from the same side from which data are recorded;
transmission, in which energy is delivered to one side of the specimen while observing from the opposite side and
internal, in which regardless of the location of the energy source, heat (or cold) is generated internally while data are collected from either side of the specimen.

Reflection and transmission are applied when using energy sources (see below) such as optical, conductive and convective. Generally, the reflection mode is best suited to detect defects located close to the heated surface while the transmission approach allows detection of defects closer to the rear surface (deeper).

Ultrasounds and eddy currents (see below) are examples of typical internal heat generation energy sources. In these cases, heat is produced at the interface of defects from where it travels to the object's surface by conduction.

1.1.4. Scanning: how the energy is transferred to the surface?

In addition, there are three scanning methods as indicated in the fourth column of figure 1:

point scanning, in which only a point of the object's surface is heated at a time, e.g. the whole inspected surface can be dynamically scanned (point by point) with a laser;
line scanning, in which a vertical or horizontal line is projected to the surface being inspected, e.g. a robotic arm moving across a large component with a line lamp as a heating source and
surface scanning, in which an area of the inspected object is heated at once, e.g. using halogen lamps to heat the whole surface of a component.

Point and line scanning are interesting for the dynamic configuration since heating (or cooling) can be carried out on a point-by-point or line-by-line basis while the camera acquires data. Surface scanning is preferred in the static configuration to take advantage of the possibility of inspecting large surface at once.

1.1.5. Sources: what type of energy is being used?

There are several manners in which energy can be delivered to the surface of the object being inspected, from a hot or cold fluid (water, air, etc) to microwave heating. Often, a heat source is preferred over a cold one, since it is more easily controlled. The fifth column in figure 1 presents some examples of energy sources, the most commonly adopted in the active approach can be subdivided as

optical; such as photographic flashes, halogen lamps or IR lamps,
mechanical, such as sound or ultrasound excitation,
inductive, such as eddy currents or
other, for instance, heating blankets are frequently used to heat (by conduction) large aeronautical components, whilst heat guns are also employed in the aeronautical industry to rapidly inspect (by forced convection) relatively smaller components.

In the passive approach, although external energy sources are not employed, an energy source can still be identified in many cases. For instance, building inspectors take advantage of solar energy in order to detect humidity in a building's exterior envelopes: they wait for the right moment of the day, usually shortly after either sunrise or sunset when the thermal gradients are at their maximum, to perform a survey. The inspection is carried out passively since no external (artificial) stimulus is used with the Sun as the energy source.

1.1.6. Sources: how the surface is being stimulated?

Finally, the sources can deliver energy using different waveforms as enumerated in the sixth column in figure 1:

modulated, thermal waves modulated at a given frequency;
pulse, a short pulse of a few milliseconds;
square pulse, a long pulse of several seconds or minutes; and
step, continuous heating from a few to several seconds.

Inspections using modulated waveforms require several cycles to reach a steady state, after which, at least one cycle is recorded and processed. By contrast, inspections using the last three waveforms are carried out in transient state.

The selected waveform will ultimately dictate the technique to be employed. For instance, data of interest are recorded and processed (if required) during the heating stage as in step heating thermography (SHT), during the cooling stage as in pulsed thermography (PT) or during both stages as in square pulse thermography (SPT). These techniques are discussed in the following section.

1.2. Techniques

Several thermography techniques have been proposed using specific scenarios adapted to a particular application, these are as follows [1–4]:

lock-in thermography (LT) [9–11], in which periodic heating at a given frequency is used in steady state to measure the amplitude and/or phase delay of the thermal response;
PT [12–14], in which a short (a few milliseconds) energy pulse is utilized to heat the inspected object, surface temperature is monitored under the principle that defective areas cool down (or heat up) at a different rate than non-defective areas;
SPT [15–17], which is similar to PT with the difference that a longer (from a few seconds to several minutes) pulse is applied, the surface temperature is monitored during both heating and cooling, or only during cooling (if defects appear during this stage);
SHT [18, 19], in which the object is heated for several seconds or minutes, similar to SPT, with the difference that only the heating stage is of interest; and
point or line scan thermography [20], in which a point (laser) or a line heating source (IR lamp) moves along the surface of the inspected specimen, while it heats up, the surface the IR camera follows at a fixed distance and at the same (or at a known) speed.

This classification is based on the waveform being used for the inspection. It is also customary to categorize techniques according to the type of energy source being employed [21]:

optical thermography (OT) [22, 23], most of the time, when no term related to the source type is included, it is referring to one of the two classical optical thermography inspection scenarios, either optical PT using photographic flashes (commonly referred simply as PT or flash thermography) or optical LT halogen lamps (commonly known as LT or modulated thermography);
ultrasound thermography (UT) [24–26] also known as vibrothermography or thermosonics, an ultrasonic (or sonic) pulse is modulated at high frequencies (15–40 kHz, typically), hence the term burst VT is preferred over pulsed VT; and
inductive thermography (IT) [27, 28] also known as eddy current thermography, either pulsed or burst is frequently adopted.

Strictly speaking, a terminology including the type of energy source and waveform should be used to avoid confusion, e.g. optical PT, ultrasound LT, inductive line scanning thermography, etc, although in practice, this is rarely the case.

These techniques were developed for the active approach, where proper control of the energy sources is possible. Nevertheless, they might also be adopted in the passive approach as illustrated in the second inspection scenario example presented next.

Example 1: inspection scenario for optical pulsed thermography

Figure 1 is intended to cover most of the situations that may arise during an inspection. For instance, one of the most commonly employed NDT IR techniques is PT using photographic flashes (optical source). The typical setup is presented in figure 3.

**Figure 3.** Typical setup for optical pulsed thermography.
Download figure:
Standard image High-resolution image

From figure 1 and the above discussion, this scenario can be represented as in figure 4(a). This scenario corresponds to the first application example presented in section 2.

Alternatively, halogen lamps can be employed instead of flashes. Since, in contrast to photographic flashes, halogen lamps are not conceived to produce highly energetic short pulses (of a few milliseconds), the inspection scenario in this case will rather correspond to the optical square pulse or optical step heating techniques. The main advantages of this arrangement are that (a) it is easier to acquire low cost lamps to implement the technique as a first experiment in universities and research laboratories, (b) the amount of energy and the duration of the stimulation can be easily controlled (with a dimmer and a switch, or electronically) and (c) it is still a good choice for the inspection of a variety of materials that require long stimulation, e.g. plastics, building materials and thick composites.

Example 2: inspection scenario for solar lock-in thermography

Several passive scenarios can also be imagined from figure 1. For instance, a promising technique for the inspection of historical buildings consists of a passive approach taking advantage of solar natural cycles. Indeed, there is an inverse correlation between the modulation frequency f and the maximum depth z that can be attained depending on the thermal diffusivity α, of the material being inspected, which is proportional to the thermal diffusion length µ: z = 1.8 µ = 1.8(α/πf)^1/2.

This relationship has been conveniently exploited in active LT for the detection of defects in a variety of applications (aerospace composite components, artworks, etc). On the other hand, the use of day and night solar cycles, with a very low modulated frequency, f = 1/(3600 × 24) = 0.000 0157 Hz, roughly approximate a sinus modulation that confers the possibility to detect internal structure or defects (cracks, humidity, delaminations, etc) located deep in the material.

The inspection scenario for an historical building long-lasting (several days) survey would be as depicted in figure 4(b). This scenario could also be implemented in the framework of academic teaching. Although no external stimulation is required, the success of an experiment will depend on the weather conditions (sunny days are the ideal situation, clouds and particularly rain will negatively affect the acquired data) and on the acquisition system (thermographic camera and PC) that should be able to acquire and store thermograms at regular intervals during at least one day (ideally more). The acquired data sequences can be processed afterwards using one of the advanced signal processing techniques presented below.

All these techniques share some interesting advantages and limitations with respect to other NDT techniques, as discussed next.

1.3. Advantages and limitations of thermography

Thermography is gaining in popularity due to several factors, mainly

constant progress and continuous price decline in computers, IR detectors and cameras technology,
production of standards practices for thermography methods by recognized organizations (such as the ASTM [29]),
development of tests methods by the industry, e.g. aeronautical, wind power, petrochemical, to cite a few and
increase of certified personnel (levels I, II and III) to carry out inspections, although still far behind other NDT methods such as ultrasound testing and radiography.

All NDT techniques have strengths and weaknesses. In the case of thermography, some of the advantages and limitations with respect to other NDT methods are as follows [4–6].

Advantages: •fast inspection rate; •contactless, no coupling needed as in the case of conventional ultrasounds. It should be noted, however, that UT requires a coupling media between the transducer and the specimen and that induction thermography coils have to be relatively close to the inspected surface; •security of personnel, there is no harmful radiation involved as in the case of x-ray radiography. However, high power external stimulation (such as powerful flashes) requires eye protection and heat-induced ultrasound requires protective ear plugs; •imaging capabilities, results are relatively easy to interpret, since they are (often) obtained in image or video formats; •applications are varied and numerous; •unique inspection tool for some inspection applications as in the case of open micro-cracks in thermally sprayed coatings difficult to inspect with other NDT approaches; and •the number of training hours required for level I certification is half the requirement for other NDT techniques such as ultrasounds and x-rays [7, 8].
Limitations: •non-uniform heating, difficulty of obtaining a fast, uniform and highly energetic thermal stimulation over a large surface (particularly for the optical pulsed scenario). For this reason, several processing techniques have been proposed as discussed below; •thermal losses by convection and heat radiation that might induce spurious contrasts affecting the reliability of the interpretation; •cost of the equipment, IR camera and thermal stimulation units for active thermography, although this is relative, active thermography is more expensive than some NDT techniques (visual inspection, some ultrasound devices, etc), however, it has very competitive costs when compared to more sophisticated equipment such as phase array (ultrasound or eddy currents) and x-rays systems; •capability to detect only defects resulting in a measurable change of the thermal properties from the inspected surface; •ability to inspect a limited thickness of material under the surface, although detection of defect several centimetres under the surface is sometimes possible, notably by LT using very low modulation frequencies; and •emissivity variations, low emissivity materials strongly reflect thermal radiations from the environment, surface painting can be employed to increase and equalize emissions when possible.

1.4. Qualitative versus quantitative inspection

In many applications, the user is interested in qualitatively detecting the presence of anomalies in order to establish if the part being inspected is defective or not. Conversely, there are some applications for which it is important to go a step further and quantify or characterize the detected defects, i.e., to determine their size, depth and/or thermal properties:

qualitative thermography: defect detection and
quantitative thermography: defect characterization: •defect sizing: determine the size and shape of the detected anomalies; •depth retrieval: calculate the depth at which the anomalies are located; and •thermal properties: estimate the thermal properties (diffusivity, thermal resistance) of the anomalies.

In either case, qualitative or quantitative, it is often necessary to apply some sort of processing in order to improve the signal-to-noise ratio (SNR) to increase defect contrast and/or to characterize defects.

Different signal processing techniques have been developed or adapted from other fields to process thermographic data. The next section presents a summary of techniques with some representative bibliographic references.

1.5. Advanced signal processing

The term advanced signal processing is employed in order to distinguish this group of methods from more basic image and signal processing such as image averaging, subtraction, division, filtering, edge detection, etc that are mainly applied as pre- or post-processing steps.

Advanced signal processing can be categorized in different groups according to the manner in which data are handled. In the following sections, a non-exhaustive list of techniques is presented. Literature on most of these methods is abundant, some representative works can be found in [1–6, 30–34] and are summarized hereafter.

Thermal contrast-based techniques: •basic thermal contrast: absolute, running, normalized, standard, •early time, •peak slope, •maximum contrast, •differential absolute contrast, •thermal tomography and •thermographic signal reconstruction.
Statistical techniques: •statistical analysis using informative parameters, •statistical moments and •higher order statistics thermography.
Matrix factorization techniques: •principal components thermography (PCT), •non-negative matrix factorization and •archetypal analysis.
Signal transforms: •Fourier transform: -phase-sensitive LT and -pulsed phase thermography (PPT); •Wavelet transform: -PPT with the wavelet transform; •Hough transform and •Laplace transform.
Artificial intelligence: •artificial neural networks and •evolutionary computations.

Thermographic signal processing is a field in constant evolution and is expected to continue to grow in the next few years.

In the following section, an example of application is presented to illustrate the experimental aspects of three different inspection scenarios and the subsequent data processing.

2. Application example: impacted carbon fibre specimen

2.1. Inspected specimen

A carbon fibre-reinforced plastic specimen (150 × 100 × 5 mm³) was inspected using different thermography techniques. The test coupon was manufactured by hot compressing (∼300 °C) 16 plies of bidirectional (0/90) 5-Harness Satin weave fabric of T300 continuous carbon (C) fibres semi-impregnated with thermoplastic polyphenylene sulfide (PPS) resin. The consolidated PPS-C composite laminate displayed the following stacking sequence: [(0/90), (±45)₂, (0/90)]₄.

The angles 0°, +45°, −45° and 90° refer to the four orientations in which the reinforcing carbon fibres were disposed in the composite laminate. The subscript 2 indicates that two consecutive ±45° oriented satin fabrics were placed in between two 0°/90° positioned woven fabrics. The subscript 4 denotes that four subsequent sets of [(0°/90°), (±45°)2 and (0°/90°)] fibre array were employed. These characteristics render the whole composite laminate symmetric and balanced, thus exhibiting in-plane quasi-isotropic mechanical behaviour.

The specimen was fully clamped along its borders and subjected to impacts with three different energy levels (respectively 5, 10 and 20 J, as illustrated in figure 5) by employing a falling weight containing a 16 mm diameter steel ball tip, as per ASTM standard [35].

**Figure 5.** Specimen USP-05: (a) map of impacts, (b) front side and (b) back side.
Download figure:
Standard image High-resolution image

2.2. Inspection scenario $\#$ 1: optical pulsed thermography in reflection

The inspection scenario is illustrated in figure 6. In reference to figure 1 is an active approach, performed on a static configuration, in reflection mode, using a surface scanning method and an optical energy source (photographic flashes) with a pulse waveform, as depicted in figure 4.

**Figure 6.** Experimental setup for pulsed thermography (reflection) and step heating thermography (transmission), Inspection scenarios $\#$ 1 and $\#$ 2, respectively.
Download figure:
Standard image High-resolution image

The specimen was stimulated from the front side (the impacted side) using two photographic flashes (Balcar, 6.4 kJ/flash) and data were recorded in reflection using a mid-wave IR (MWIR) camera (FLIR Phoenix, InSb, 3–5 µm, 640 × 512 pixels) at a frame rate of 55 Hz.

Data were stored as a 3D matrix as illustrated in figure 7(a), where x and y are the spatial coordinates and t is the time.

**Figure 7.** Inspection scenario $\#$ 1: optical pulsed thermography: (a) 3D thermogram matrix and (b) 1D temperature profiles for an impacted zone (dotted line) and sound area (continuous line).
Download figure:
Standard image High-resolution image

Figure 7(b) shows actual temperature profiles for a small area (mean value) over the 20 J impacted zone and a sound area next to it. A logarithmic scale is used for the time coordinates for sake of clarity. Temperature decreases approximately as t^−1/2 (at least at early times), as predicted by the 1D solution of the heat equation (surface temperature evolution for the case of a Dirac pulse in a semi-infinite isotropic solid): ΔT = T-T₀ = Q/[e/(π t)^1/2], except for the defective areas, where the cooling rate is different.

As can be seen, the acquisition lasted 17 s; the 20 J impacted area is not seen right after the flash since the temperature for both the impact and the reference area is practically the same. A significant thermal contrast appears a few milliseconds later (∼0.07 s) and fades after a few seconds of cooling (∼4 s). Similar temperature profiles could be traced for the 10 J impacted area, although thermal contrast would appear with lesser contrast, and there would be not enough thermal contrast signature for the 5 J impacted zone. Advanced processing might help to detect these areas as will be discussed below.

2.3. Inspection scenario $\#$ 2: optical step heating thermography in transmission

For the second inspection scenario, a halogen lamp was utilized to heat the sample in the transmission mode (heating from the back side, inspection from the front side). The experimental setup is schematized in figure 6. In this case (from figure 1), the active approach was used on a static configuration, in transmission mode, using a surface scanning method and an optical energy source (halogen lamp) with a step waveform, as depicted in figure 8.

**Figure 8.** Inspection scenario for optical step heating thermography in transmission.
Download figure:
Standard image High-resolution image

The sample was heated for 60 s using a 1000 W halogen lamp, while data were recorded in transmission with an MWIR camera (the same camera employed for the inspection scenario $\#$ 1) using an acquisition frame rate of 6.5 Hz. The recorded 3D thermogram matrix is presented in figure 9(a), the thermal profiles corresponding to approximately the same areas around the 20 J impacted area selected for the inspection scenario $\#$ 1 (figure 7(b)) are shown in figure 9(b). Temperature is presented in arbitrary units since temperature calibration was not available. The fact of using non-calibrated data, however, does not have any effect on the processing techniques that will be presented below.

**Figure 9.** Inspection scenario $\#$ 2: optical step heating thermography: (a) 3D thermogram matrix and (b) 1D temperature profiles for an impacted zone (dotted line) and sound area (continuous line).
Download figure:
Standard image High-resolution image

**Figure 9.** Inspection scenario $\#$ 2: optical step heating thermography: (a) 3D thermogram matrix and (b) 1D temperature profiles for an impacted zone (dotted line) and sound area (continuous line).
Download figure:
Standard image High-resolution image

In this case, the impacted areas are more difficult to see than for the PT scenario. Thermal contrast starts to be evident only after 40 s of heating (with lesser contrast than in scenario $\#$ 1). Although the sample was heated during 60 s, the surface temperature continued to increase for a few more seconds (until ∼76 s).

The areas damaged with less energetic impacts (10 and 5 J) present practically no thermal contrast in the raw thermograms. As for the case of optical pulsed thermographic data (inspection scenario $\#$ 1), signal processing techniques will be applied to increase impact damage detection in these areas.

2.4. Inspection scenario $\#$ 3: ultrasound lock-in thermography

For the third inspection scenario, an ultrasound transducer (Branson 200b, 15–25 kHz) was employed to stimulate the specimen from the back side. Heat is generated internally while data are recorded from the front side. In this case, the active approach was used on a static configuration in which heat is generated internally, using a surface scanning method, and a mechanical energy source (ultrasounds) with a modulated waveform, as depicted in figure 10.

**Figure 10.** Inspection scenario for ultrasound lock-in thermography.
Download figure:
Standard image High-resolution image

The experimental setup is presented in figure 11. The sample was heated for 5.3 s using a 25 kHz ultrasound wave modulated at a frequency of 0.6 Hz and a high velocity MWIR camera (Telops FAST-IR 1000 MW, InSb, 3–5 µm, 320 × 256 pixels) using an acquisition frame rate of 1000 Hz.

A sample of the recorded 3D thermogram matrix is presented in figure 12(a), the thermal profiles corresponding to approximately the same areas around the 20 J impacted area selected for the inspection scenario $\#$ 1 (figure 7(b)) are shown in figure 12(b). The temperature scale is in arbitrary units (non-calibrated data) for which the scale is considerably different than the data presented in figure 9(b), since a different camera was employed in each inspection scenario. In this case, the affected area corresponding to the 20 J impacted area can be clearly distinguished (see last recorded thermogram in figure 12(a)).

**Figure 12.** Inspection scenario $\#$ 3: lock-in vibrothermography: (a) 3D thermogram matrix and (b) 1D temperature profiles for the 20 J impacted zone (red curve showing cyclic modulation) and sound area next to it (black quasi-steady line with only a minor temperature rise).
Download figure:
Standard image High-resolution image

Areas affected by lower energy impacts require further processing, as discussed in the following section.

2.5. Comparative advanced signal processing results

2.5.1. Inspection scenario $\#$ 1

Figure 13 presents some comparative results from the three inspection scenarios investigated herein. Figure 13(a) shows a close-up view from the front side of the test coupon where surface damage corresponding to the 20 and 10 J impacts is clearly seen by visual examination. The impacted area corresponding to the 5 J impact presents no sign of damage. Figure 13(b) presents a raw thermogram from optical PT (inspection scenario $\#$ 1) in which the surface thermal signature of only the two more energetic impacts (20 and 10 J) are noticeable, but not for the less energetic impact (5 J).

**Figure 13.** Comparative results: (a) close-up view of the front side of the test coupon; (b) raw thermogram at t = 1.75 s; (c) optical pulsed thermography (inspection scenario $\#$ 1) processed by PPT: phasegram at f = 0.17 Hz; (d) optical step heating thermography (inspection scenario $\#$ 2) processed by PCT: EOF₄; and lock-in vibrothermography (inspection scenario $\#$ 3) processed by PCT: (e) EOF₂ and (f) EOF₃.
Download figure:
Standard image High-resolution image

Figure 13(c) shows a phasegram obtained by processing data by PPT (see [6] for more details). In this image, the dark areas around the 20 J impacted zone appear to correspond to internal damaged, while the bright areas to surface damage. Only slight indications of surface damage (bright areas) could be seen around the 10 J impacted area, and no signs of damage are detected around the 5 J impact. Usually, severe damage caused by a transverse impact in fibre-reinforced polymer composite laminates will present some clear visual indications in the surface, but no signs of internal damage. At least for the case of the most energetic impact (20 J), inspection by optical PT (inspection scenario $\#$ 1) proves to be advantageous with respect to the visual inspection. Nevertheless, it appears that the applied energy was not enough to be able to detect damage of less severe impacts (10 and 5 J).

2.5.2. Inspection scenario $\#$ 2

The image in figure 13(d) was obtained by processing the SHT data (inspection scenario $\#$ 2) by PCT, it corresponds to the fourth empirical orthogonal function (EOF) (see [6] for more details). Given that data were recorded in a transmission configuration (the thermal front propagated from the front to the back side of the sample where data were collected), the thermal signature is likely to be due to the internal damage caused by the impacts. Internal damage can be seen with good contrast in the area corresponding to the 20 J impact with less contrast for the area around the 10 J impact and only a hint can be observed around the 5 J impact. Comparing the EOF image in figure 13(d) with the phasegram in figure 13(c), a certain correlation between internal damaged areas can be inferred in the zone around the 20 J impact.

2.5.3. Inspection scenario $\#$ 3

The images in figures 13(e) and (f) are the result of processing the ultrasound LT data (inspection scenario $\#$ 3) by PCT. Given the great number of images in this case (acquisition lasted 5.3 s at a frame rate of 1000 Hz = 5300 thermograms), the image was subdivided in three regions of interest (ROI) to facilitate processing, since PCT requires a great amount of computer resources. A very fast frame rate was employed given that the heat generation process with ultrasounds is very fast and a large amount of images are required in order to improve the SNR. The areas corresponding to the damaged areas around the two more energetic impacted zones can be clearly seen in figure 13(e) and around the three impacted areas in figure 13(f). In figure 13(e), the ROIs around the 20 and 10 J impacts show good indications of internal damage (dark colours), although no signs for the 5 J impacts are seen. In figure 13(f), surface or near the surface damage (bright colours) as well as internal damage (dark colours) are observed for the two most energetic impacts (20 and 10 J), whilst in the case of the 5 J impact, only internal damage can be observed. This result is interesting in the sense that it allows to clearly identify the impacted areas corresponding to the three different energy levels (20, 10 and 5 J) and to classify them according to their degree, surface or internal.

3. Conclusions

Thermography is an attractive technique gaining popularity for NDT applications, thanks to the constant technology improvement in different areas (computers, thermal sensors and cameras) and also because of the progressive acceptance of the technique by the industry (development of standards practices and test methods and the increase of certified personnel).

A wide variety of scenarios can be conceived in order to carry out a thermographic inspection. The most formally established scenarios can be divided by the energy source being used, e.g. optical, ultrasound and inductive, although many other types of source can be employed. The manner in which energy is transferred to the inspected part is also of great importance, which can be modulated or pulsed (short pulse, square pulse, or step). In either case, advanced signal processing is very useful in order to improve the signal-to-noise ratio, increase defect contrast and to characterize defect.

In addition, thermography can be conveniently used as a teaching tool for undergraduate and graduate university programs to help students to better understand basic physical phenomena related to heat transfer or in many areas related to the generation of heat.

Some interesting remarks could be made from an impacted test coupon that was inspected employing three different inspection scenarios.

Optical stimulation is simple and economical. However, the heat front is produced at the surface and then propagates through the material by conduction. The whole inspected area needs to be heated and results are therefore greatly affected by non-uniform heating. This was the case of optical pulsed thermography in reflection (inspection scenario $\#$ 1) and optical step heating in transmission (inspection scenario $\#$ 2).

Advanced signal processing helped to increase defect contrast and to detect internal damage. In the case of optical pulsed thermography, the photographic flashes did not provide enough heat to detect damaged areas impacted with energies below 20 J. On the contrary, optical step heating in transmission allowed clear detection of impact internal damage from 10 J and more and a slight indication could be detected for the 5 J impacted area.

Detection of low-impact (5 J) internal damage was only possible by ultrasound lock-in thermography (inspection scenario $\#$ 3) given that, in this case, heat is generated internally, providing thermal signatures of increased defect contrast with respect to optical techniques. Still in this case, advanced signal processing was necessary to retrieve the information of internal impact damage in the case of medium (10 J) and low (5 J) energy impacts.

Acknowledgments

The authors want to thank the support of the Canada Research Chair in Multipolar Infrared Vision (MIVIM), the Ministère du développement économique, innovation et exportation du Québec and the São Paulo Research Foundation (FAPESP) for supporting JRT through the processes 2006/50110-3 and 2010/08552-4.

Nondestructive testing with thermography

Article metrics

Permissions

Author e-mails

Author affiliations

Dates

Abstract