view as required for maintenance                  Emissivity changes can cause severe
inspections. Single chip arrays of            errors in radiometric detection methods,
128 X 128, 256 X 256, 512 X 512,              unless something is done to keep
640 X 512 , 800 X 500, 1200 X 1000 and        emissivity constant. This has been
larger elements apiece are for full field     successfully accomplished primarily by
imaging systems. The most common array        coating the test surface with materials
detectors are made of platinum silicide or    that possess uniform, high emissivity
indium antimonide; mercury cadmium            values (typically c = 0. 7 to 0.9). Uniform
telluride, common for single element          emissivity is necessary for temperature
detectors, has also been introduced for       measurement accuracy while high
multiple arrays.                              emissivity is desirable to provide a larger
                                              radiant intensity (per the
Other Radiometers. A great variety of         Stefan-Boltzmann lmv). It was once
radiometer designs and readouts exist.        thought that coatings needed to be dark
One radiometer detects a thermographic        in color, such as lampblack suspended in
image by means of the                         a polymeric binder. However, transparent
evaporation/condensation nature of a          polymeric coatings also have been
thin oil film on a membrane at the focal      successful. 1\•fetal surfaces especially must
plane. Some scanning radiometers              be coated and nonmetal surfaces to a
produce C~scan and multiple A~scan            lesser degree; however, high emissivity
recordings of thermographic information       coating will help. After testing, the
rather than the optical image. This type of   coatings are usually removed from the
readout gives an accurate value for           material surface.
radiation intensity at each scan point.
High speed nne scanners with a scope              The emissivity problem can also be
readout (instead of the slower mechanical     reduced by special design of the
recorder) display the temperature profile     radiometer system. One quite ingenious
of a "line" across the surface.               method uses two radiometers, where the
                                              infrared signal from one radiometer is
   The infrared microscope scans an           delayed, a constant signal added to it and
extremely small spot (only about 25 pm        the sum divided by the other radiometer's
fl x 10-3 in.] in diameter) over the          TABLE 1. Total radiation emisslvities at all wavelengths.
surface of very small materials. The                       Material                                   Emissivity Range
resulting image is magnified and                                                                          (0 to 1.0)
displayed on some form of cathode ray
tube, such as a television tube. Infrared     Metallic
nondestructive testing microscopy is
suited for intricate surfaces in specialized  Highly polished aluminum, silver, gold, brass, tin 0.002 to 0.04
applications such as the inspection of
integrated circuits. Nonmkroscopic            Polished brass, copper, steel, nickel,                  0.03 to 0.08
systems are also common.                         chromium, platinum, clean mercury
Emissivity Variables                          Dull, smooth, dean aluminum and alloys, copper, 0.08 to 0.20
                                                brass, nickel, stainless steel, iron, lead, zinc
Infrared thermography is most successful
for surfaces with high emissivity. High       Rough ground or smooth machined castings;               0.15 to 0.25
emissivity provides several important
effects. First, as seen in the                steel mill products, sprayed metal, molten metal
Stefan-Boltzmann law, surfaces with a
high emissivity e emit a higher intensity     Sooth, slightly oxidized aluminum, copper, brass, 0.20 to 0.40
radiation at a given temperature, thereby       lead, zinc
providing a larger signal for the infrared
detector. Second, high emissivity surfaces    Bright aluminum, gilt, or bronze paints                 0.30 to 0.55
are, by definition, poor reflectors. Low
emissivity surfaces tend to reflect           Heavily oxidized and rough iron, steel, copper,         0.60 to 0.85
radiation from other sources. The infrared      aluminum
detector thus senses energy unrelated to
the temperature of the object being           Nonmetallic
examined. This contributes to the noise of
the test and reduces the sensitivity to       White or light colored paint, plaster, brick, tile,     0.80 to 0.95
details of interest. Third, high emissivity     porcelain, plastics, asbestos
surfaces also absorb more radiant energy.
Radiant sources can therefore be effective    Snow at 263 K (-1 0 "'Cor +14 °F)                       0.85
in inducing a thermal gradient in the test
sample, a condition required for many         Red, brown, green, buff and other colors of paint,      0.85 to 0.95
infrared thermal tests. Table 1 provides a      tile, inks, clays or stone; glass and translucent
list of typical surface emissivities for a      plastics; glass fiber composites; ice crystals; oil;
variety of materials in a temperature range     varnish
of about 273 to 313 K (0 to 40 oc; 32 to
104 °P).2·'                                   White bond paper, sand, wood (planed oak)               0.90
                                              Carbon black, asphalt, carbon fiber composites,         0.90 to 0.97
                                                matte black paints, tar
                                              Concrete                                                0.92
                                              Brick (common red)                                      0.93
                                              Water                                                   0.96
                                              Human skin                                              0.98
36 Infrared and Thermal Testing
output. Sensing delay time between the          Melting point coatings melt at some
                    two radiometers is accomplished by           specific temperature. Anomalies are
                    scanning one ahead of the other. The         usually associated with a temperature
                    result is an emissivity independent          increase, so the materials melt first over
                    readout (even on very rough surfaces) for    anomalies. Melting point compounds also
                    regions of constant temperature. Another     arc comparatively insensitive and require
                    method uses a single radiometer that         relatively high surface temperatures.
                    samples the reflected and emitted infrared
                    radiation. Although considerably more           Liquid aystals are cholesteric liquids
                    simple, this design is suited only for very  whose optical properties cause them to
                    smooth surfaces.                             reflect vivid spectral colors for
                                                                 temperature changes. Their adjustable
                 Thermosensitive Indicators                      response is quite sensitive and can be
                                                                 made to change from red to blue over a
                    As mentioned above, the other basic          temperature gradient as small as 1 K
                    method for observing thermograms
                    involves the application of                  (1 oc = 1.8 °F). An additional feature is
                    thermosensitive materials directly on the
                    surface. This approach conducts the          that their responses can be adjusted to
                    thermal pattern on the test surface into     occur at temperatures only slightly al10ve
                    the thennosensitive material, usually a      ambient. Tests on honeycombs and
                    thin coating sprayed, painted or held        laminated structures \\'ith liquid crystals
                    against the surface. A variety of coatings   have been quite successful. Figure 5 shows
                    have been developed, including               the invisible thermogram of irregularities
                    permanent color change coatings,             in a light gaged honeycomb structure.
                    phosphor coatings, melting point coatings    Notice that the thermal gradient can be
                    and liquid crystals.                         intensified by contact heating one surface
                                                                 while cooling the near surface ·with cool
                      Perrnanent color change coatings, such as  air.
                    creosote, react chemically and change
                    color permanently over temperature               Both passive and active infrared
                    gradients; these materials are               nondestructive testing find a variety of
                    comparatively insensitive and normally       applications throughout diverse
                    require that the surface be heated weH       industries. The development of
                    above the ambient temperature.               satisfactory (sensitive) thermograms
                                                                 depends on several important factors. It
                      Pllvspllor coatings are materials whose    must be remembered that, because of
                    fluorescent radiation intensity, under       transient heat flow, a thermogram of a
                    ultraviolet light, is a function of          given area changes with respect to
                    temperature.                                 observation time (heating time). \.Yhen
                                                                 heat is applied to an inspection surface, as
FIGURE 5. liquid crystal testing of light gaged honeycomb        shown in the common setup in Fig. 6, a
structure.                                                       thermogram will develop that is a
                                                                 function of the material, the nature of the
                                Cool air                         discontinuity, the observation time and
                                                                 the heat intensity. A complicated
                                                                 relationship exists among these variables.
                                                                 However, a practical summary follows.
Green                                                            FIGURE 6. Common radiometer setup for
                                                                 active testing.
                                                                                             Radiometer
                                                                 Infrared lamp                 Infrared lamp
Aluminum honeycomb with                                                         Black coating
0.3 mrn (0.013 in.) thkk skin                                                   Test material
                                                                                 Thermogram
                                                                 Fundamentals of Infrared and Thermal Testing 37
I.Transient temperature differences over       given depth, infrared testing is less
    discontinuities in materials having          sensitive to smaller discontinuities
    higher thermal diffusivities (metals)        than to larger ones.
    are shorter m duration and lower in        S.Discontinuities parallel to the test
    magnitude than they are in materials         surface are usually easier to detect
    having lower thermal diffusivities           than ones perpendicular to the
    (nonmetals).                                 surface.
  2.The magnitude of the surface             Emissivity
    temperature difference over a
    discontinuity decreases with the depth   Emissivity is a variable defined as a ratio
    of the discontinuity. Near surface       of the total enefb'Y radiated by a given
    anomalies are much easier to detect      surface at a given temperature to the total
    than deep ones.                          energy radiated by a blackbody at the
                                             same temperature. A blackbody is a
 3.The observation time required for a       hypothetical radiation source that yields
    temperature difference to reach its      the maximum radiation energy
    maximum value over a discontinuity       theoretically possible at a given
    increases as the thermal diffusivity     temperature. Also a blackbody will absorb
    decreases, as the depth of the           all incident radiation falling upon it.
    discontinuity increases and the size     Blackbodies have an emissivity of 1.0 and
    increases. At a given depth, small       aH real materials have emissivities
    discontinuities reach their peak         between 0 and 1.0.
    intensities on the thermogram before
    larger discontinuities reach their peak     Figure 7 represents the
    intensities.                             Stefan-Boltzmann law for blackbodies at
                                             various intensities. Note that the
  4.The duration of the transient            wavelength envelope shifts toward the
    temperature difference over a            visible range for increasing blackbody
    discontinuity decreases as the           temperature, according to \.Yien's
    discontinuity size decreases and as the  displacement la'~N. As shown by \.Yien's
    discontinuity depth increases. At a      equation (Eq. 1), the wavelength of
                                             maximum intensity is computed simply
fiGURE 7. Stefan~Boltzmann radiation law     by dividing 2897 by the temperature of
for blackbodies (total energy E= e7",        the surface:
maximum" 3000· r-1).                                                  b
                                                                   T
                                             where b is the ·wien displacement constant
                                             (2897 pm·K-1), Tis temperature (kelvin)
                                             and },1nax is maximum wavelength
                                             (micrometer).
                                                The effect of emissivity on the
                                             radiation curve is given in Fig. 8. As
                                             fiGURE 8. Emissivity effect on radiation from surface of
                                             emissivity c: with hypothetical intensity.
                                               ,.-,            c = 1 (blackbody)
                                                I\                        £ = 0.9 (graybody)
                                               I\            '\
                                              I\
                                             I                '''
                                             I r varies
                                                       (not
                                               graybody)
                    10 100                   Wavelength (relative unit)
                       Wavelength j.Jm
legend
  1. 300 K ;;+27 °C "'-t80 °F
  2. 195 K;; -78 oc;; -108 °F
  3. 126 K = -147 oc = -233 Of
38 Infrared and Thermal Testing
shown, it acts somewhat like a filter or         In the most general case, incident
valve. Graybody materials have emissivity     infrared waves are reflected and
values that are less than that of a           transmitted in addition to being absorbed.
blackbody at all temperatures and             Because the whole must equal the sum of
wavelengths. Some materials, called           its parts, the fractions of absorbed,
spectral radiators, have a spectral           reflected and transmitted energy must
emissivity that varies in a characteristic    equal the total incident energy:
fashion over the range of emitted
wavelengths.                                  (3) a + r +  1
   Emissivity is a surface phenomenon         where a= absorptivity coefficient;
depending on the surface condition and        r = reflectivity coefficient; and
composition. Smooth materials have            t = transn1issivity coefficient.
lower emissivities than rough materials.
Freshly polished metals have lower               If the transmissivity is low enough to
ernissivities than oxidized or corroded       be neglected (as is usually the case with
metal surfaces. Nonmetals usually have        infrared testing), then a = 1 - r and
higher emissivities than metals.              r = 1 - a. Another important relationship,
Lampblack or certain metallic powders         known as Kirchoff's law, states that the
yield very high emissivities. Special         ratio of radiation intensities for two
blackbody cavities for calibrating            surfaces iS equal to the ratio of their
radiation equipment produce the highest       absorptivities. This means that
                                              absorptivity equals emissivity or
emissivity available, within a fraction of a  emissivity equals 1 minus reflectivity.
                                              Several deductions from the preceding
percent of that for a blackbody.              radiation laws are useful: (1) efficient
   Because the object of infrared testing is  emitters arc efficient absorbersi
                                              (2) inefficient absorbers are inefficient
to measure surface temperature changes,       emitters; (3) efficient reflectors are
emissivity can be an uncontrolled             inefficient emitters; and (4) inefficient
variable. Variations of emissivity across     reflectors arc efficient absorbers.
the surface of a material can cause false
indications. \Vhen the emissivity
decreases in a localized region, the
radiation intensity decreases, falsely
indicating a localized reduction in
temperature and vice versa. Also, surfaces
with low emissivity values, such as
polished n1etals1 are more difficult to test
than high emissivity surfaces.
   Because all materials are continuously
radiating infrared energy, it seems they
might eventually cool down to absolute
zero. This would happen if materials did
not pick up energy from other sources by
means of radiation, conduction or
convection. All materials are continuously
and simultaneously radiating and
absorbing infrared energy. \Nhen a
material is wanner than its surroundings,
its radiation emission will exceed its rate
of absorption1 causing the temperature of
the material to drop. The opposite is true
for material cooler than its surroundings.
\-\7hen the material and its surroundings
reach the same temperature, the thermal
equilibrium point, each body is emitting
radiation at the same rate that it absorbs
it (without a net change in energy).
   The basic relationship describing the
above radiation interchange is defined by
Prevost's l<lW of exchanges. For the case of
one body surrounded by the interior walls
of another body, Prevost's law yields:
where£ is emissivity (of materials 1
and 2); K11 is Boltzmann's constant;
Tis absolute temperature (K); and H' is
net gain or loss of radiation intensity.
                                              Fundamentals of Infrared and Thermal Testing 39
1.
PART 2. General Approaches and Techniques of
Infrared and Thermal Testing6
Thermography is one of several                 Emissivity is a unitless surface property
techniques used to see the umee11. ?,8 As the  that describes the ability to emit energy. It
name implies, it uses the distribution         is a unitless quantity and on a scale from
(~graph)') of surface temperatures (thermo-)   0 to 1, where E = 1 for a blackbody.
to assess the structure or behavior of what    Generally, emissivHy E depends on
is under the surface. Traditionally the        wavelength 'A, temperature T, viewing
term thermography has denoted a contact        angle 8 and surface conditions such as
technique to record a distribution of          roughness, oxide layers and physical and
surface temperatures whereas infrared          chemical contamination. Objects whose
thermography is a contactless technique        emissivity is independent of the
with distinct advantages. In the 1970s,        wavelength are called gray bodies while
the term thermography came to usually          colored bodies refer to full dependence
mean noncontact, infrared thermography.        emissivity.
Contact thermograph}' can be deployed
with liquid crystal paints applied on the         The fundamental equation of infrared
surface of interest and monitored with <l      thermography relates the irradiance Ncam
conventional video camera: cholesterol         (that is, spectral radiant po·wer incident
esters, under temperature effect change        on a surface per unit area) received by the
orientation and reflect colored light from     camera to the radiance emitted from the
red to violet when illuminated with white      surface under consideration Nsur at a given
light. Deployment with an array of             temperature T, neglecting the atmosphere
thermocouples is also possible but not         contribution as in most nondestructive
practical \\'hen numerous grouped              testing applications:
readings are needed. In nondestructive
testing, thermography is generally             (6) Ncam = ENsur + {I - c) N{'Jl\'
deployed contactless.
                                               with Nenv being the radiance emitted by
Physical Basis                                 the surrounding environment considered
                                               as a blackbody. If emissivity of the surface
Planck's law describes the distribution of     is high, Eq. 6 reduces to Ncam = Nmr and
the spectral radiance L'} b the rate at which  knowledge of the calibration curve of the
energy is emitted by a blackbody bat a         camera linking environmental radiance
given temperature T, per unit surface, per     Nsur to temperature T permits retrieval of
unit of solid angle and as function of the     the surface temperature, assuming there is
wavelength k                                   no radiometric distortion. Various
                                               techniques can help to solve low or
where h is Planck's constant (6.626076 x       uneven emissivity problem: (1) covering
                                               the inspected surface with a high
w-34 }s), cis the speed of light               emissivity paint, (2) a reflecting cavity's
                                               artificially increasing the emissivity
(-3 x 10s m·s-1) and KB is Boltzmann's         through multiple reflections,
constant (1.381 x J0-23 U·K-')). A             (3) performing simultaneous observation
blackbody is also an instrument that, as a     of the surface in different spectral bands
perfect absorber, totally absorbs energy       (as in two color pyrometry), (4) relying on
coming from any direction and from any         thermal imprint transfer on a high
wavelength while, as a perfect radiator, it    emissivity material on which the
follows Kirchkoff's laws and reemits this      observation is performed and (S) taking
energy until the thermodynamic                 into account in Eqs. 4 and 6
equilibrium is reached with the                measurements of local emissivity values
surrounding environment. For normal            and radiance from the surrounding
bodies, Eq. 4 becomes Eq. 5 with               environment.
correction factor~ (emissivity):
40 Infrared and Thermal Testing
Instrumentation                                    detectors in which photon
                                                    interactions either change
  Infrared thermography uses two different
  kinds of infrared cameras: scanning               conductivity (photoconductive) or
  radiometers and focal plane arrays                generate voltage. Detectors that
  available either in one or two dimensions         generate voltage are called plwtuelectric
  (one-dimensional is useful when                   or plwtovoltaic.
  inspecting moving objects, the second           2. In pltotoemissive photonic detectors,
 dimension then provided by the
 displacement). Infrared images are called          the signal observed is constituted by
 lhennogmms.                                        measurement of an electron current i).
     Scanning radiometers are generally           Because no heating is needed as for
 equipped with an internal temperature
 reference seen by the detectors during           thermal detectors, response time is short.
 scanning so that the output signal is
 directly calibrated in temperature.              Because of their solid state structure, these
 However, the electromechanical scanning
 by the detectors through prisms or               detectors are compact, reliable, robust and
 mirrors limits the frame rate and
 sometimes corrupts the output signal. In         much employed. Common materials in
 focal plane arrays or starring arrays,
 because the infrared image of the scene is       photoelectric devices (photodiodes and
 directly imaged on the detector matrix,
 no scanning is needed. All pixels are            phototransistors) are silicon, indium
 acquired simultaneously and high speed
 frmne rate is achieved but temperature           arsenide, indium antimony and cadmium
 calibration is done externally- for
 instance, by pointing the focal plane array      mercury telluride. Quantum detectors
 camera to a blackbody of known
 temperature.                                     require cooling to reduce the noise to an
    'f\\'o families of infrared detectors exist.  acceptable level. The most common
 In thermal detectors the incoming
 radiation heats the surface and this             approach is liquefied gas stored in a
 heating affects a property of the heated
 material that then translates into               vacuum vessel called a dewar. For
 variations of the signal output. In
 bolometers, the electrical conductivity          example, liquid nitrogen provides cooling
changes. Thermal detectors do not require
cooling and their response is independent         at a temperature of 77 K (-196 "C ~
of the wavelength so that an interference
filter is added in the optic window to            -321 oF). Other cooling approaches are
limit spectral sensitivity. Because detector
temperature changes are required,                 joule-thompson gas expansion, stirHng
response is slow. Recently,
micromachining technology has made                cycle engines or thermoelectric elements
microbolometer arrays available in focal
plane arrays and especially aimed at              based on the peltier effect.
qualitative applications. Thermopiles and
thermocouples generate a voltage                  Because the atmosphere lacks perfectly
difference through thermoelectric
                                                  flat spectral transmission properties, the
thompson effect. In pyroelectric detectors
electric charges are generated by incident        selection of the operating wavelength
radiation absorption (heating),
pyroelectric elements can be made as              band is conditioned by the application
point or image detectors. Pyroelectric
tubes are similar to standard vidicon             and detector type. Among the important
television cameras except for the face
plate and pyroelectric target material.           ~riteria are operating distance,
   In photonic detectors, the signal is           mdoor/outdoor operation, temperature
obtained by measuring directly the
excitation generated by the incident              and emissivity of bodies of interest.
photons. Heating of the sensitive surface
is unnecessary. Photonic detectors are of         Following Planck's Jaw (Eq. 4), high
two types: (1) quantum (photoconductive
and photoelectrk/photovoJtaic) and                temperature bodies emit more in the short
(2) plwtoemissh'e.
                                                  wavelengths, thus long •wavelengths ·will
  I. Quantum detectors are solid state
                                                  be of more interest to observe near room
                                                  temperature objects. These are also
                                                  preferred for outdoor operation where
                                                  signals are less affected by solar radiation.
                                                  At operating distances restricted to a few
                                                  meters (about ten feet) in absence of fog
                                                  or water droplets, atmospheric absorption
                                                  has little effect. lvfost common bands in
                                                  infrared thermography are 3 to 5 pm
                                                  (short waves) and 8 to 12 pm (long waves)
                                                  because these match the <Itmospheric
                                                  transmission bands.
                                                  Another important point to consider is
                                                  the detectivity D (or normalized
                                                  detectivity  D*) of the  d°Cete=c~to3r21usoeFd/  for
                                                  instance a   77 K (~196                          cooled
                                                  indium antimony detector operating in
                                                  the 3 to S pm range has a detectivity
                                                  seven times higher than does a 77 K
                                                  (~ 196 °C = -321 oF) cooled cadmium
                                                  mercury tellmium detector operating in
                                                  the 8 to 12 pm range. Detailed studies
                                                  have concluded that for temperatures
                                                  from 263 to 403 K (-10 "C to +130 "C;
                                                  14 to 266 oF), measurements can be done
                                                  ·without much difference in both the 3 to
                                                  5 pm band and the 8 to 12 pm band, ·with
                                                  however a slight preference for 3 to 5 pm
                                                  devices for which the errors in
                                                  Fundamentals of Infrared and Thermal Testing 41
temperature measurements are generally                   speed sewing of seat cushions and airbags
smaller. For particular applications (such               in the automobile industry. Modeling
as mi1itary), bispectral cameras operating               makes it possible to optimize sewing
simultaneously in both bands are                         operations through needle redesign and
deployed to characterize target thermal                  needle cooling- with significant
signatures more accurately.                              economic and quality benefits because of
                                                         the million of products sewn daily.
Passive Thermography                                     Active Thermographic
                                                         Techniques
The first Jaw of thermodynamics expresses
the principle of energy conservation and                 Pulsed Thermography
states that an important quantity of heat
is released by any process consuming                     Pulsed thermography is one of the most
energy because of the law of entropy                     common thermal stimulation techniques
(Eq. 4). Temperature is thus an essential                in infrared thermography. One reason for
parameter to measure in order to assess                  this test's popularity is its quickness. It
proper operation (Fig. 9).                               relies on a short thermal stimulation pulse
                                                         lasting from a few ms for high
   In passive thermography, abnormal                     conductivity materials such as metals to a
temperature profiles indicate a potential                few seconds for low conductivity
problem and key words are the                            specimens (such as plastics and graphite
temperature difference with respect to the               epoxy laminates). Brief heating at a few
surrounding, often referred to as the                    degrees above initial component
delta T (t.'J) or the hot spot. A t.T of                 temperature prevents damage to the
I to 2 K (I to 2 oc ;c 2 to 4 oF) is generally           component.
found suspicious while a 4 K (4 °C =: 7 ol:)
value is a strong evidence of abnormal                      Basically, pulsed thermography consists
behavior. Generally, passive thermography                of briefly heating the specimen and then
is rather qualitative because the goal is                recording the temperature decay curve
simply to pinpoint anomalies. However,                   (Fig. 9). Qualitatively, the phenomenon is
some investigations provide quantitative                 as follows. The temperature of the
measurements if thermal modeling is                      material changes rapidly after the initial
available so that measured surface                       thermal pulse because the thermal front
temperature (isotherms) can be related to                propagates, by diffusion, under the
specific behaviors or subsurface                         surface and also because of radiation and
discontinuities.                                         convection losses. The presence of a
                                                         discontinuity modifies the diffusion wte
   Dedicated modeling helps process
control researchers, for instance, to
understand needle heating during high
FIGURE 9. Schematic setup for infrared thermographic nondestructive evaluation. Thermal stimulation is only
needed in active procedures. Drawing shows reflective scheme.
                                                                          Subsurface structures of interest
                Thermal stimulation control                                          ~·                   ~    .     Specimen
                     in active procedures
                                                                                                          euc     '~
 O::::LJControl
                                                                                                                  .•u,
   >igo•l> IPo»iv•""'hoiq"'
                                                                                                                   0
.----~>-1 lockin technique 6..2j 1 - r - - - - - - - /
                                                                                                                  ~
                                                                                     ~·· ~
                                                                                         ~ .•
                                                                                                          e
                                                                                     .... •~· 3e
                                Digital image recording  Infrared camera             ~                   ~E
Computer (visualization,                                                                                  ~
        processing)                                                                  .... .·.•5.,
                                                                                          •~
                                                          lamp or other thermal
                                                         stimulation devices such
                                                          as hot air jets in active
                                                                 procedures
42 Infrared and Thermal Testing
so that, when the surface temperature is         monitored during the application of a
                                                 step heating pulse (Fig. 9). Step he~ling
observed, discontinuities appear as areas        thermography finds many apphc<ttw_ns
of different temperatures with respect to 1      such as for coating thickness evaluatiOn
surrounding sound area, once the thermal         (including multilayered coatings),
front has reached them. In a first
approximation, observation time t is             inspection of coating-.to-substrate b?nd or
function of the square of the depth z and        evaluation of composite structures. !I
the Joss of contrast C is proportional to
the cube of the depth:                           Lockin Thermography
(7)                                              Thermal waves were already investigated
                                                 by Fourier and Angstrom in the
(8) c   1                                        nineteenth century. Lockin thermography
                                                 is based on such waves generated inside
       z'                                        the inspected specimen and detected
                                                 remotely.IO \'\lave generation, for instance,
where a is the thermal diffusivity of the        is performed by periodically depositing
material.                                        heat on the specimen surface (for
                                                 example, through sine modulated lamp
   These relations indicate two limitations      heating) while the resulting osciHating
of infrared thermography: observable             temperature field in the stationary regime
discontinuities are generally shallow and        is remotely recorded through its thermal
contrasts are generally weak. An empirical       infrared emission (Fig. 9).
rule of thumb says that tile radius of the
smallest detectable discontinuity should be at      Lockin thermography is also called
least as large~ and preferabl)' two or more      photothermic radiometry'. The word lockin
times larger, than its depth under the surface.  refers to the necessity to monitor the
This rule is valid for homogeneous               exact time dependence of the heating,
isotropic material. In case of anisotropy, it    modulated between the output signal and
is more constrained.                             the reference input signal. This is done
                                                 with a lockin amplifier in a point-by-point
   Various configurations are possible:          laser heating or by computer in full field
point, line or surface inspection. Pul~ed        deployment so that both phase and
heating is achieved by laser beam, h1gh          magnitude images become available.
power photographic flashes, lamps with           Phase images are related to the
mechanical shutter and hot air jets. In          propagation time and, because they are
some instances, a cool pulse is preferred,       relatively insensitive to local optical
for instance if the temperature of the part      surface features (such as nonuniform
to inspect is already higher than ambient        heating), they are interesting for
temperature due the manufacturing                nondestructive testing. Modulation
process. In that case, a cold thermal            images are related to the thermal
source such as a line of air jets is used        diffusivity and found some uses for the
while such a cold thermal source docs not        characterization of electromagnetic fields
induce spurious thermal reflections into         in antennas. 11
the infrared camera as in the case of a hot
thermal source (Eq. 6).                              Recording of four raw thermograms 51
                                                 to 54 located equidistantly on the sine
   Obsen'ation is possible either in             modulation cycle yields to the phase
reflection mode, ·where the thermal source       $(ro)and magnitude A(m) images:
and detector located on same side of the
inspected part (Fig. 9) or in transmission       (9)  $(w)   ~       s, - s,
mode, where the source and detector are
on opposite sides of the part. The                              atan · .
reflection approach is best suited to detect                           54 - 52
discontinuities close to the heated surface
while the transmission approach permits          The depth range of magnitude image is
detection of discontinuities close to the
rear surface. The transmission approach is       roughly given by thermal diffusion
not always possible- sometimes the rear                         r;2-k
surface is not accessible. Also, in              length !t:
transmission mode, the discontinuity             (11) l'
depth can not be estimated because the
travel distance is the same regardless of                    \ OlpCp
discontinuity depth.
Step Heating Thermography                        with thermal conductivity k, density p,
The step heating technique is sometimes          specific heat CP and modulation .
referred to as long pulse tlzennogmplly or       frequency (1}, In the case of phase unages,
time resolved infrared radiumel1y. Here the
increase of surface temperature is               the depth range is about twice larger.
                                                 Equation 11 indicates that higher
                                                 modulation frequencies restrict the
                                                 analysis in a near surface region.
                                                      Fundamentals of Infrared and Thermal Testing 43
Loss Angle Lockin Thermography. Other                within the field of view, the temporal
                stimulations are possible. For instance, an
                ultrasonic transducer (shaker) can be                decay f(x) is extracted from the
                attached to the specimen, or the specimen            thermogram sequence, where x is the
                can be partly immersed in an ultrasonic              index ill the thermogram sequence. Next
                bath. In these cases, the high frequency             from ((x) the discrete Fourier transform
                ultrasonic signal (typically about 40 kHz)
                is modulated with a low frequency signal.            F(w) is computed, where {I) is the
                The lmv frequency modulation creates a
                thermal wave of desired wavelength as in             frequency variable. Finally, from the real
                photothermal lockin thermography while               R(w) and imaginary /(co) components of
                the high frequency acts as a carrier                 F(w), the amplitude A(w) and phase ¢(ro)
                delivering heating energy inside the                 are computed:
                specimen.12 This technique is referred to
                as the loss ansle lockin thermography and            (12) A(ro)
                has been applied with success for
                detection of corrosion, vertical cracks and          (13) ¢(w)   a t a n - 1-Rl((ww-))
                delaminations.
                                                                     In pulsed phase thermography as in
                Pulsed Phase Thermography. An ideal                  lockin thermography, it is possible to
                Dirac pulse has a flat frequency spectrum,           explore the various frequencies. However
                thus a thermal pulse in pulsed                       in pulsed phase thermography the
                thermography launches under the                      analysis is performed in the transient
                specimen surface a mix of frequencies                mode while in lockin thermography, the
                that can be unscrambled by performing                signal is recorded in the stationary mode.
                the Fourier transform of the temperature
                decay on a pixel-by-pixel basis. This                Vibrothermography
                enables computation of phase images as
                in lockin thermography. Such a                       Vibrothemography is an active infrared
                processing technique that combines                   thermographic technique where, under
                somehow advantages of both pulsed                    the effect of mechanical vibrations (20 to
                thermography and lockin thermography                 SO Hz) induced externally to the structure,
                is called pulsed phase thermograplly. 13             because of direct conversion from
                                                                     mechanical to thermal energy, heat is
                   The process is as follows. After pulsed           released by friction precisely at
                heating the specimen, the temperature                discontinuities such as cracks and
                decay is recorded and for each pixel (i,j)
\ •'-.u-.--·
TABlE 2. Applications of infrared thermographic techniques.
     Technique                      Process Control                      Discontinuity Detection              Material Characterization
Passive
                Carton sealing line inspection, automobile brake     Walls, moisture evaluation, roofs,   Glaze thickness on ceramics,
  thermography    system efficiency, heat dissipation of electronic    assemblies                           crush tests investigation
                  modules, recycling process identification,
lockin            printed circuit boards, glass industry (bottles,   liquid level in tanks
  thermography    bulbs), welding process, metal (steel) casting
  (active)                                                           Crack identification, disbanding,    Adhesion strength, anisotropic
                Bearings, fan and compressors, pipelines, steam        impact damage in carbon fiber        material characterization,
                  traps, refractory lining, rotating kilns, turbine    reinforced plastics                  coating thickness in ceramics,
                  blades, electric installations, gas leaks                                                 moisture evaluation
                                                                     Metal corrosion, crack detection,
                Aircraft structural component inspection, loose        disbanding, impact damages in      Depth profile of thermal
                  bolts detection                                      carbon fiber reinforced plastics,    conductivity or diffusivity
                                                                       turbine blades, subsurface
                Plastic pipe inspection                                defect characterization (depth,    Thermophysical properties
                                                                       size, properties) in composites,     (diffusivity etc.), underalloyed
                Radar absorbing structure investigation                wood, metal, plastics                and overalloyed phases in
                                                                                                            coatings on steel, moisture,
Pulsed          Aircraft structural component inspection, solder     Defects in adhesive and spot           anisotropic material
  thermography    quality of electronic components, spot welding       welded lap joint                     characterization
  (active)        inspection
                                                                     Coating wear, fatigue test, closed   Thermal conductivity
                Water entrapment in buildings and fresco               crack detection                      measurement in c.:1rbon fiber
                  delamination                                                                              reinforced plastics
Step heating    Degradation of erasable programmable read only                                            Coating thickness mea~urement
  thermography    memory chips
  (active)                                                                                                Variations in viscoelasticity and
                Paper structure (cockling)                                                                  emissivity
Vlbrothermography Failure analysis
  (active)
44 Infrared and Thermal Testing
FIGURE 10. Application of active pulsed thermography to detect subsurfac:;:e discontinuities in
carbon fiber reinforced plastic specimens. Thermogram sequence reveals impact damage on
bottom right corner. Progression of the thermal contrast is clearly seen after initial thermal
pulse (flashes are fired at t = 0 s). Maximum thermal contrast image clearly shows extent of
delaminated area.
t=O.s    I"" 0.98 s                          t"" 1.25 $  t= 1.51 .s
f=l.78s  I= 2.06 s                           t=2.32s     t;; 2.87 s
1=3.42s  t=4.5ls                                         Maximum contrast at
                                                         1""' 1.44 s(see Fig. 11)
delaminations. Discontinuities are excited     (focal plane arrays with more than
at specific mechanical resonances: local       512 x 512 pixels), greater pixel
subplates formed from delaminations            response uniformity (> 99.5 percent)
presence resonate independently of the         and faster acquisition, no longer tied
rest of the structure at particular            to video standards of 25 or 30Hz.
frequencies. 14 Consequently, by changing      Some infrared cameras, for example,
(increasing or decreasing) the mechanical      have a 30 kHz frame rate.
excitation frequency, local thermal          2.Computer hardware and sofhvare can
gradients might appear or disappear.           be improved to process the infrared
                                               signaL For example, wavelet transform
Common Applications and                        processing could be used with both
Limitations                                    time and frequency capability, and
                                               thermograms could be evaluated
Common applications of infrared                automatically.
thermographic techniques are listed in       3.lnfrared machine vision applications
Table 2. Their limitations and capabilities    ofinfrared thermography include
are listed in Table 3. Figures 10 and 11       integrated processing and, for active
show results from active pulsed                procedures, particular heating
thermography on a carbon fiber                schemes. ror example, lateral heating
reinforced plastic specimen: an impact         may be applied to detect cracks on
damage specimen is detected. Dedicated         concrete structures, and performance
processing such as thermal contrast            from uncooled microbolometer based
computing permits extraction of                infrared cameras may be improved for
quantitative information about the             continuously monitoring various
discontinuity.                                 industrial processes.
   Future developments can be
anticipated in three directions.
  !.Hardware can be improved to acquire
    the infrared signaL For example,
    infrared cameras with low noise
    (< 20 mK), higher spatial resolution
                                             Fundamentals of Infrared and Thermal Testing 45
TABLE 3. Advantages and limitations of infrared thermographic techniques.
Technique             Advantages                                           Disadvantages
All thermographic     Fast, surface inspection                    Variable emissivity
  techniques
                      Ease of deployment                          Cooling tosses (convection/radiation causing perturbing contrasts)
                      Deployment on one side only
                      Safety (no harmful radiations)              Absorption of infrared signals by the atmosphere (especially for distances
                      Ease of numerical thermal modeling            greater than a few meters [about 10ft])
                      Ease of interpretations of thermograms      Difficulty to get uniform heating (active procedures)
                      Great versatility of applications (see
                                                                  Transitory nature of thermal contrasts requiring fast recording infrared
                        Table 2)                                    cameras
                      Sometimes unique tool (corrosion around
                                                                  Need of straight viewing corridor between infrared camera and target
                        rivets)                                     (although it could be folded through first surface mirrors)
                                                                  limited contrasts and limited signal to noise ratio causing false alarms-
                                                                    measurement of a few degrees above background at around 300 K
                                                                    (27 'C ~ -80 'F)
                                                                  Observable defects generally shallow
Passive               No interaction with specimen                Works only if thermal contrasts naturally present
  thermography        No physical contact
lockin                No physical contact                         Require modulated thermal perturbation
  thermography        large inspected surface- several m2         Require observation for at least one modulation cycle (longer observation
  (active)
                        (30 ft2) simultaneously                     with respect to pulsed thermography)
Pulsed                                                            Thickness of inspected layer under the surface
  thermography        Phase and modulation images available       related to the modulation frequency (unknown defect depth might require
  (active)
                      Modulated ultrasonic heating (for some        multiple experimentations at different frequencies)
                        applications, might require physical
                        contact or bath immersion)                Requires apparatus to induce the pulsed thermal perturbation
                                                                  Computation of thermal contrasts require a priori knowledge of defect free
                      No physical contact
                                                                    zone in field of view
                      Quick (pulsed thermal stimulation: cooling  Inspection surface limited (~0.25 m2 maximum).
                        or heating)
                      Phase and modulation images available
                        with frequency processing (as in pulsed
                        phase thermography)
Step heating          No physical contact                         Require apparatus to induce the thermal perturbation
  thermography        Reveal dose cracks                          Risk of overheating the specimen
  (active)
                                                                  Difficulty to generate mechanical loading
Vibrothermography                                                 Thermal patterns appear only at specific frequencies
  (active)                                                        Physical contact to induce thermal stimulation
FIGURE 11. Logarithmic progression of thermal contrast over
white line passing through center of delaminated area of
Fig. 10. Contrast value over discontinuity free areas is 1.
      1.5                                           Time (s)
2c 1.4
 ~
·~ 1.3
g 1.2
,e;; 1.1
~c
u0
      0.9
     150
      Pixel position
       (increment)
46 Infrared and Thermal Testing
PART 3. Calibration for Infrared
Thermography1s
               Correction of Vignetting                               transmit oblique rays and a part of the
               Effect                                                 light cone may be cut off, causing an
                                                                      amplitude reduction at the edges of the
                      Before quantitative analysis of                 image. In Fig. 12, the central bright area
                      thermograms, it is necessary to convert         corresponds to the portion of sensor fully
                      the raw image sequence into temperature         reached by radiation and the dark area is
                      images. This involves generally one or two      caused by a loss of radiation caused by the
                      steps such as image restoration (for            limited lens aperture (the three dots in
                      example, correction for vignetting effect       Fig. 12 correspond to locations of three
                      and/or noise suppression) and conversion        reference points in Fig. 14). Vignetting is
                      of raw pixel values to temperature              also more severe if expansion rings restrict
                      following a calibration procedure. In the       the field of view because of the limited
                      case of focal plane arrays, the image           effective aperture obtained in this case.17
                      restoration is generally limited to the
                      vignetting effect (if present) because the          As predicted by the theory,ls
                      noise level is low. Here, uniform heating       experiments carried out with the focal
                     and emissivity are assumed.                      plane array camera shmved that this effect
                                                                      depends both on pixel location and
                         An example of vignetting is shown on         temperature difference between the target
                     Fig. 12. This phenomenon is explained as         and the ambient. Figure 13 clearly
                     follows. 16 If a cone of rays from a point in    illustrates the vignetting effect on three
                     the object space, limited by the                 thermograms recorded at uniform
                     diaphragm of the lens, is formed and             temperature, respectively 282, 295 and
                     intercepted with the image plane
                     perpendicular to the lens axis, the              323 K (9' 22 and so °C; 48, 72 and 122 °F)
                     intercept is a circle if the object lies on the  for af band c. Belmv ambient temperature
                     optical axis and more generally an ellipse
                     if the object is laterally displaced.            (295 K [22 oc = 72 °F]), vignetting has an
                     !vforeover, for many lenses, the front and
                     rear apertures are too small to fully            opposite behavior than above ambient
                                                                      temperature: the curvature direction
FIGURE 12. Vignetting effect is visible in image of uniform           changes. At ambient temperature,
                                                                      vignetting is not visible (Fig. 13b). In
temperature target (323 K[50 oc = 122 °F]) acquired with              Fig. 13, the left drawing in each pair is an
                                                                      uncorrected plot whereas the right
focal plane array camera. Three squares correspond to                 drawing is obtained after correction for
reference points (see Figs. 13 and 14).                               vignetting.
                                                                          In Fig. 14, the difference between the
                                                                      signal at the central reference point
                                                                      (corresponding to the center of the
                                                                      brightness area in Fig. 12) and three
                                                                      points placed at different distances from
                                                                      it, is shmvn (starting at the center of the
                                                                      FIGURE 13. Side view w·lre frame representations of raw and
                                                                      corrected images showing vignetting effect: (a) 282 K
                                                                      (9 oc =48 of), (b) 295 K (22 oc = 72 of), (c) 323 K
                                                                      (50 oc = 122 oF). Left drawings are before correction for
                                                                      vignetting; right drawings, after correction.
                                                                      (a) (b) (c)
                                                                      Fundamentals of Infrared and Thermal Testing 47
image, the three dots shown in Fig. 12          optical elements.20 Because in infrared
             correspond to the*, +, o plots of Fig. 14).
             This difference is expressed by:                thermography (depends on the
                                                             temperature, a software approach is
             (14) ti;,;(G},y) = Cil,r ~ G/;
                                                             necessary. The idea is to create an
             where d;,;(G},.) is the difference in gray      (M x N x 2) matrix file (where M x N is the
             level at reference G}.r; G/; is the gray level
             of the pixel i,j at the temperature t; and      image format) containing the coefficients
             G},,.is the gray level of the reference at
             temperatme t.                                   a;,j and b;,; for every location (i,j) in the
                 In the case of a sequence of                image.
             thermograms taken at different                      By combining Eqs. 14 and 15, solving
             temperatures, this difference is a linear
             function of the temperature in gray level       for Gf,jrr and adding the effect of
             which may then be expressed as
                                                             temperature derived with GJ~~~v.o, the
             (15) d;,;(c.!._,.) = a;.; c.!,, + b;,;          correction formula then becomes:
             where a;,; and bi,j are constant coefficients   ,,, ,,,,.o(16)
             for every location (i,j) in the image.          , (crdG.c.~m =               .Ten\'~ (1'ref ) + bij
                                                                         b;  '1·  ~a;,·
                 The ambient temperature for this test                                 '
             corresponds to gray level 985. For such a       1'1 I - a··                       l'    '
             level the correction is negligible and does
             not depend on the position in the image         where GJ~~· is the gray level at the
             (see Fig. 14, around gray level 985, where      reference point location corresponding to
             the three curves get close together).           the actual ambient temperature (at the
                                                             time the correction is computed) and
                 In the visible spectrum, a possible         GJ~~·.o is the temperature (in gray level) of
             hardware solution to this vignetting            the room when the correction matrix was
             problem is to add an additional lens in         created. As an illustration of the
             front of the objective. 19 If the original      effectiveness of this procedure, it was
             optics of the camera introduces a               applied to the left plots of Fig. 13, and the
                                                             plots on the right of Fig. 13 are obtained
             r-Jdistortion {then using a lens with a         after application of Eq. 15. The
                                                             improvement obtained for Fig. 13b (the
             distortion function corrects tht'               ambient temperature case) is attributed to
             response of the global system. Of course        the noise filtering effect of the fitting
             such an approach would attenuate the            process with Eq. 16.
             signal. This approach is possible if the
                                                             Noise Evaluation and
             function f does not depend on the               Temperature Calibration
             features of the scene, such as temperature      The technique proposed by Haddon21 and
             because of emission originating from
                                                             Lee22 and reported in Maldague2:l was
                                                             used to chawcterize the noise content
                                                             present in infrared images. Two images, A
                                                             and B, of the same scene were recorded
                                                             and a third image C was obtained by
fiGURE 14. Linear behavior of signal loss due to vignetting  subtraction of A and R. Because the scene
for three reference points of Fig. 12.
                                                             in A and ll is the same, image C
                                                             represents the noise introduced by the
                                                             camera. The standard deviation is
    120                                                      computed with the formula:
"'- 100                                                                  1[ · · - · · · · - - -
                                                                              2}:(i,i) - Cav]'
~                                                            (17) oc ~ ·.         I
c                                                                                     ZMN
~ 80
a_ 60
70                                                           where C;;Jv is the average of C,
                                                             M =maximum row, N =maximum
 ~  40                                                       column, i = (0, 1, ..., M-1), and j = (0, 1,
~                                                            ..., N-1).
~
                                                                l listograms of image C at about 296 K
>, 20
                                                             (2:l oc ~ 73 °F) for the three different
~
s"' 0                                                        infrared cameras 'i\'ere computed (see
                                                             Table 4 for figures about the three
gw -20                                                       cameras). Newer advanced technology
                                                             cameras {focal plane array) exhibit a noise
    -40      1000  1500  2000    2500           3000         content restricted in contrast to the frame
        500                                                  grabbing system (whose technolot.'Y dates
                                                             back to the 1980s). To compute the
             Gray levels at reference position
48 Infrared and Thermal Testing
standard deviation in degret's celsius, the                                     For the other cameras, calibration
calibration function for each camera is                                         functions were obtained directly from the
also needed (see Eq. 18 below).                                                 manufacturers' data sheets.
Concerning the focal plane array camera,
experiments have been carried out to                                               It is important to point out that such a
estimate this function and to verify its                                        calibration procedure is only valid for a
stability in time; measurements performed                                       specific experimental setup. If the
through one month showed variations                                             experimental conditions change, it is
Jess than one percent.                                                          necessary to repeat the process. For
                                                                                instance, such changes may involve a
   The next step consists of comierting                                         change of the objective or a change of the
gray level values into temperature.                                             operating range of the camera (see the
Figure 1S shows calibration curves for                                          case in Fig. 15 for the stirling cooled
three different infrared cameras. For                                           camera (camera 3) at 273 to 523 K (0 to
instance, in the case of the focal plane                                        250 "C; 32 to 480 "F) and 123 to 353 K
array, the following relationship is                                            (-150 to +80 "C; -240 to +175 "F). As
obtained where sis the gray level value                                         mentioned before, focal plane arrays do
(linear best fit with a third order                                             not have internal temperature references;
polynomial function):                                                           it can then be of interest to include a
                                                                                blackbody in the field of vie\\' to
(18) 1tc) ~ - 13.4 + o.o5 g                                                     recalibrate the focal plane array as the
                                                                                experiment progresses.
                      - 1.6 X to-S g2
                                                                                   Table 4 indicates for the three tested
                      + 2.2 X 1()-9 g3                                          cameras, the standard deviation (at
                                                                                ambient temperature) according to Eq. 17
TABlE 4. Comparison of infrared cameras.                                        in both gray levels crG and temperatures aT
                                                                                (with images converted to temperature
                             Standard Deviations                       Noise    using a calibration function such as the
                                                                    Equivalent  one of Eq. 18 for the focal plane array).
                             Gray                                    Material
                                                                                   The reference to camera 2 with a digital
     Infrared Camera         Level  crT at T~mb                         Loss    board in Table 4 means that the digital
          (percent)                                                             signal is obtained directly from the
                             oc K or <>c "F                            5 to 6   camera electronic unit through a
Camera 1, focal plane array                                                     dedicated board (single analog-to-digital
                             1.5 0.04 0.072                                     conversion inside the electronic unit)
                                                                                whereas the acquisition through the
Camera 2 with frame grabber 4.3" 0.17h 0.306b 18 to 19                          frame grabber for the same camera is done
Camera 2 with digital board 2.03 O.Q8b 0.144b 9 to 10                           from the analog video output of the
                                                                                electronic unit by a frame grabber_ Thus
Camera 3, stirling coo!ingc 1.3 0.13 0.234 14tol5                               this makes reference to a triple acquisition
                                                                                that degrades the signal as indicated
a. Thermal range= 10.                                                           (analog to digital to analog to digital). See
b. Assuming a linear function temperature (gray level value).                   elsewhere24 for additional details on this
                                                                                digital board.
c. Range 273 to 523 K (O to 250 ~c = 32 to 482 ~f).
FIGURE 15. Comparison among temperature calibration                             TABLE 5. Noise standard deviation for focal plane array.
functions for three tested infrared cameras. For focal plane
array camera, experimental points are plotted too.                                   Temperature                                                    Standard Deviation
                                                                                                                                                 O"c O"T (K or 0 C) O"T (<>F)
      433 (160) (3201                                                                                                                     ~----
      413 (140) (2841                                                           (K) ("C) ("F)
[L 393 (120) [248}                                                              282.4   9.2   48.6                                               1.47  0.054  0.097
                                                                                284.8  11.6   52.9                                               1.43  0.050  0.090
G'-..' 373 (100) (2121                                                          290.8  17.6   63.7                                               1.49  O.D45  0.081
                                                                                296.2  23.0   73.4                                               1.50  0.039  0.070
"-' 353 (80) (1761                                                              300.8  27.6   81.7                                               1.48  0.033  0.059
                                                                                304.2  31.0   87.8                                               1.48  0.029  0.052
               (60) (1401                                                       307.3  34.1   93.4                                               1.54  0.027  0.048
                                                                                309.5  36.4   97.5                                               1.62  0.025  0.045
313 (40) (1041             _ J_ __,__ _L _ __t__                    _j          311.3  38.2  100.7                                               1.62  0.023  0.041
293 (20) (681               1000 1500 2000 2500                                 312.9  39.8  103.6                                               1.61  0.022  0.039
273 (0) (321                                                          3000      314.7  41.5  106.7                                               1.61  0.020  0.036
                                                                                315.9  42.8  109.0                                               1.60  0.019  0.034
253 (-10) HI                                                                    317.4  44.2  111.6                                               1.60  0.019  0.034
                                                                                318.4  45.3  113.5                                               1.63  0.019  0.034
233 (-20) (-40j t___                                                            319.4  46.2  115.2                                               1.64  0.019  0.034
                                                                                323.7  50.5  122.9                                               1.69  0.021  O.D38
                 soo
                                     Gray levels (arbitrary units)
legend
  - - - - "'900 rilnge (0.250}
  ---a- =focal plane array camera
  ------ =900range(-150.80)
                                                                                Fundamentals of Infrared and Thermal Testing 49
In column 2, the values of standard
                   deviation in gray level for the focal plane
                   array camera and stirling cooled camera
                   are very close, \\'hereas the corresponding
                   values in column 3 exhibit a threefold
                   difference. This difference is explained by
                   the different temperature responses from
                   the cameras which have calibration curves
                   ·with very different slopes, as seen in
                   Fig. IS.
                       VVith respect to noise, the focal plane
                   array camera was further studied and
                   standard deviation was computed, in gray
                   levels G and celsius degrees T, in a range
                   of temperature from 282.4 to 323.7 K (9.2
                   to 50.5 oc [48.5 to 123 oF]) (Table 5).
                       This discussion illustrates issues
                   addressed in calibration of equipment for
                   infrared thermographic signal acquisition.
SO Infrared and Thermal Testing
References
 1. Section 6, "Thermal and Infrared         11. Balageas, D. and P. Levesque. uEMII{:
     Nondestructive Testing."                     A Photothermal Tool for
    Nondestructive Testing Handbook,
     second edition: Vol. 9, Special              Electromagnetic Phenomena
    Nondestructive Testing Methods.              Characterization.'1 International
     Columbus, OH: American Society for          journal of Thermal Sciences- Revue
     Nondestructive Testing (1995).              GCm?rale de Thermique. Vol. 371
                                                  No. 317. New York, NY: Elsevier
 2. Wolfe, W.L. and G.J. Zissis. 1"1le            Science (September 1998): p 725-739.
     lnfi·ared Handbook. \'\7ashington, DC:
     Office of Naval Research, Department    12. Rantala1 ]., D. VVu and G. Busse.
                                                  "Amplitude Modulated Lockin
     of the Navy (1985).
 3. Cohen,]. Elements of Thermography for         Vibrothermography for NDE of
                                                  Polymers and Composites." Research
    Nondestructh-e Testing. NRS Technical        in Nondestructit'e Evaluation. VoL 7,
     Note 1177. \'Vashington, DC: National        No. 4. Columbus, OH: American
                                                  Society for Nondestructive Testing
     Bureau of Standards (1983).                  (1996): p 215-228.
 4. ASTM E 1316, "Standard Terminology
                                             13. lvialdague, X. and S. Marinetti. upuJse
     for Nondestructive Examinations:
     Section], Infrared Examination."             Phase Infrared Thermography."
    Annual Book ofASTM Standards:                journal of Applied Physics. Vol. 79,
     Vol. 03.03. \Vest Conshohocken, PA:          No. 5. College Park1 MD: American
                                                  Institute of Physics (1996):
     American Society for Testing and             p 2694-2698.
     Materials.
                                             14. Tenek1 L.H. and E.G. Henneke. 11Flaw
 5. Bolz, R.E. and G.L. Tuve, eds.                Dynamics and Vibro-Thermographic
    Hand/Jook of Tables for Applied               Thermoelastic NDE of Advanced
     E11gineering Science. Cleveland, OH:         Composite Materials.11 Tlzermosense
     CRC Press (1973): p 211.                     Xlll. Proceedings SPIE Vol. 1467.
 6. Maldague, X. "Thermographic                   Bellingham, VVA: International
    Techniques for NDT.'' Encyclopedia of         Society for Optical Engineering
    Materials Science a11d Teclmolog}'.           (1991): p 252-263.
     Oxford, United Kingdom: Elsevier
    Science Limited (to be published).       15. Marinetti, S., X. :Maid ague and
 7. Maldague, X.P.V. Nondestructive               M. Prystay. "Calibration Procedure for
    Evaluation ofMaterials by Infrared            1:ocal Plane Array Cameras and Noise
     Thermography. London, United
     Kingdom: Springer-Verlag (1993).             Equivalent Material Loss for
                                                  Quantitative Thermographic NDT."
 8. Maldague, X.P.V., ed.lnfrared                Materials Evaluation. Vol. 55, No. 3.
    Metllodologr and Teclmolog)'.                 Columbus, OH: American Society for
     langhorne, PA: Gordon and Breach             Nondestructive Testing (March 1997):
     (1994): p 525.                               p 407-412.
                                             16. Ballard, D.H. and C.M. Brown.
 9. Osiander, R., ].W.M Spicer and               Computer l'ision. Upper Saddle River,
    ].C. Murphy. 11Analysis Method for            NJ: Prentice Hall (1982).
     Full-Field Time Resolved Infrared       17. Maldague, X., ).-C. Krapez and
     Radiometry." Thermoset1se XVIII. SPIE        P. Cielo. "Temperature Recovery and
     Vol. 2766. Bellingham, WA:                   Contrast Computations in NDE
     International Society for Optical
                                                  Thermographic Imaging Systems,"
     Engineering (1996): p 218-227.               journal ofNondestructive EwJluation.
                                                  Vol. 10, No. !,january 1991, p 19-30.
10. Busse, G. "Nondestructive Evaluation     18. Kingslake, R. Applied Optics and
     of Polymer Materials." Nondestructive       Optical Engineering. Vol. 2. New York,
    '!!'sting and Evaluatiou International.       NY: Academic Press (1965):
    Vol. 27, No. 5. Oxford, United                p 212-213.
                                             19. Cicio, P. Optical Techniques fOr
     Kingdom: Elsevier Science Limited           Industrial lmpection. San Diego, CA:
     (1994): p 253-262.                           Academic Press (1988).
                                             Fundamentals of Infrared and Thermal Testing 51
20. Hamrelius, T "Accurate Temperature
                          Measurement in Thermography: An
                          Overvie'w of Relevant Features,
                          Parameters and Definitions."
                        Tllerrnosense XIII. Proceedings
                          SPIE Vol. 1467. Bellingham, WA:
                          International Society for Optical
                          Engineering (1991): p 448-457.
                     21. Haddon, j.F. "Generalised Threshold
                        Selection for Edge Detection.'1 Pattem
                        Recognition. Vol. 21, No. 3. Oxford,
                          United Kingdom: Pergamon Press for
                          Pattern Recognition Society (1988):
                          p 195-203.
                     22. Lee, D.)., 1~F. Krile and S. Mitra.
                          "Digital Registration Technique for
                          Sequential Fundus Images."
                        Applications o(Digital Image
                          Processing X. SPIE Vol. 829.
                          Bellingham/ V\7A: International
                          Society for Optical Engineering
                          (1987): p 293-300.
                   23. Maldague, X. Non-Destmctive
                       Evaluation o( Materials by Infrared
                        Thermography. London, United
                          Kingdom: Springer Verlag (1993).
                    24. Jalbert, L. and X. Maldague. "Design
                          of an Integrated Conversion/Control
                          System for Infrared Image and Signal
                       Acquisition." Vision Interface 1995
                          {Quebec, Canada]. Toronto, Canada:
                          Canadian Image Processing and
                          Pattern Recognition Society (1995):
                          p 148-152.
52 Infrared and Thermal Testing
CHAPTER
      Heat Transfer
Vladimir P. Vavilov, Tomsk Polytechnic University,
Tomsk, Russia (Parts 1, 2 and 3)
Roberto Li Voti, University of Rome, Rome, Italy
(Part 4)
Mario Bertolotti, University of Rome, Rome, Italy
(Part 4)
Douglas D. Burleigh, San Diego, California (Parts 1, 2
and 3)
Grigore L. Liakhou, Technical University of Moldavia,
Kishinau, Moldavia (Part 4)
Stefano Paoloni, University of Rome, Rome, Italy
(Part 4)
Concita Sibilia, University of Rome, Rome, Italy (Part 4)
PART 1. Fundamentals of Heat Transfer
Heat Transfer Mechanisms                         where 7~ is surface temperature, T1 is fluid
                                                 temperature and hr is convection heat
Heat transfer occurs in a medium or              transfer coefficient (\·V·nr2·K-').
between bodies in three different ·ways:
conduction, com•ecliuu and radiation                The maximum radiation flux emitted
(Fig. 1). Conduction is the propagation of       by a blackbody is given hy the
heat energy whenever a temperature               Stefan-Boltzmann law:
difference exists between two solid bodies
in contact or among parts of a body.             where a is the Stefan-Boltzmann constant
Convection involves the mass movement            (5.67 x JQ---8 \'V·m-2 ·K-4 ). Graybodies are
of gas or liquid molecules over large            characterized ·with an emissh'it)' E that
distances. Two solid bodies will exchange        varies from zero to unity and determines
energy by convection if they are in              the energy emitted by radiation:
contact with a fluid. Radiation is a process
of heat transfer and is characteristic of all    FIGURE 1. Heat transfer mechanisms:
matter at temperatures higher than               (a) conduction; (b) convection; (c) radiation.
absolute zero. Radiated energy may be            (a)
transported over large distances through
gases or a vacuum with no conduction or                                           Heat
convection medium.
                                                 (b)
    Heat energy is transferred from an
object with a higher temperature to an                                                                          Fluid
object with a lower temperature. fn a
general case, the temperature distribution
in a body depends on three coordinates
X,)~Z and on time t. The temperature
distribution is expressed T(X,)~z,t). Heat
transfer occurs bet\veen bodies until
thermodynamic equilibrium is reached.
Temperature differences in a medium or
in a body tend to decrease with time
because of heat transfer.
   The analysis of three-dimensional heat
transfer problems is extremely difficult. It
is common to simplify the mathematics
by considering a less complicated
two-dimensional or one-dimensional case.
    In a one-dimensional case, the
conductive heat transfer 0cd (\·V·m-2) is
given by the Fourier lav·::
(1)
where k is a material property called            ~(c)
tllrrmal conductivity (\·V·m-1·K-1). Tlms, in a
                                                      t. t. t r,
plate of thickness L with the stationary
                                                  r, c=J
surface temperatures T1 and T2 where '12 >
                                                   legend
T1, the rate of heat transfer is equal to
                                                      T = temperature
(2) Q,d  Tz-1j
            R                                       r, =fluid temperature
where thermal resistance R "" L·k 1                  T, = surface temperature
(m2·W-I.!().
   Convection heat transfer is described
with Newton's law of cooling:
54 Infrared and Thermal Testing
Two bodies 1 and 2 emit radiant              Heat Transfer in Gas Filled
energy independently. The resulting heat        Voids
flux between them is of the form:
                                                ~vfany typical voids in materials, especially
where G is a dimensionless geometric            voids in composite laminates, can be
factor dependent on the shape and               approximated as thin, gas filled gaps. The
orientation of two bodies {for two large        conductive heat transfer that occurs in
size parallel planes G = 1). Over relatively    such voids because of a differential
large distances, radiation energy may be        temperature T 1-T2 on opposite surfaces of
partially absorbed in the gaseous medium        the void is given by Eq. 2. Convection
bet\veen two bodies. In thermal                 may be neglected if the product of grashof
nondestructive testing, the distances are       number Grand prandtl number Pr fits the
small and radiation energy losses are also      following condition:
small.
                                                (10) Gr· Pr < 1000
   In the case of a single surface, heat
exchange between a warm body '/~ and a             The grashof number has the same role
cold ambient Tamb occurs by combination         in free convection that the reynolds
of convection and radiation:                    number has in forced convection. The
                                                grashof number indicates the ratio of the
(7) Q  QC\.+Qrd                                 buoyancy force relative to the viscous
                                                force acting on the fluid. The prandtl
       hcv (Ts ~ Tamb)                          number relates the relative thicknesses of
                                                the hydrodynamic and thermal boundary
       + crG(£1~4 ~ £·Ta~th)                    layers. It is a ratio of the (molecular)
                                                momentum diffusivity to the thermal
   In thermal nondestructive testing, a         diffusivity.
temperature difference (T5 - Tamh) is
typically small and emissivities are               Verification of Eq. 10 for temperature
rtypically high (< ~ 1):                        differences T1 ~ T2 < 100 oc yields the
(8) ll,d (T, -J;,nb) 4aG ( T, +Jemb £           maximum detectable void thickness
                                                typical in materials evaluation:
                          X (J~-1~mb)
                                                (11) d < 6 mm (0.25 in.)
                                                In thin voids of large lateral size, the
                                                radiation heat flux is approximated:
    \h/ith this assumption, Eq. 7 is identical  TABLE 1. Recommended values of the
to Eq. 3 with 11 = ll,v + llrd combining both
convection and radiation. Typically, the        surface heat exchange coefficient (E = 0.9;
value of 11 is not known with high
accuracy because it depends on the body's       T,mb = 293 K (20 "C = 70 "F); h,v =
shape and orientation, composition and
roughness, as well as on the temperature        1.7 x (T5 - Tamb) 113; hro is determined
difference T5 - Tamb that varies with time.     by Eq. 8. Values for hare in W·m-2.K-1.
The recommended values of II arc
presented in Table 1, where E = 0.9; Tamb =          Ts- Ta   hiT  h,d  IJ
293 K (20 oc = 68 °F); ll,v is determined by    K or oc ('F)
(9) hcv = 1.7~1~-Tamb                           1 (1.8) 1.7        5.2 6.9
                                                                   5.3 8.2
and 11rct is determined by Eq. 8.               5 (8) 2.9          5.4 9.1
   Nonadiabalic heat transfer involves all                         5.7 10.3
                                                10 (18)       3.7  6.0 11.3
three mechanisms above. In active                                  6.3 12.1
thermal nondestructive testing, an              20 (36)       4.6  6.6 12.9
external heat source is used to provide a                          6.9 B.6
heat flux to a test body. If an external        30 (54)       5.3  7.3 14.3
heat flux is much more powerful than the                           7.6 14.9
heat flux between a body and the                40 (72)       5.8  8.0 15.6
environment, the case is referred to as                            8.4 16.3
adiabatic. Only conduction must then be         50 (90)       6.3
considered. Additionally, the
corresponding analytical adiabatic              60 (108)      6.7
solutions have a simpler form.
                                                70 (126)      7.0
                                                80 (144)      7.3
                                                90 (162)      7.6
                                                100 (180)     7.9
                                                                            Heat Transfer 55
where Tis defined by Eq. 13:                                                 The iriternal heating component H' in
                                                                          Eqs. 15 and 16 is usually not needed for
(13) T ~                                                                  the analysis of tlwrmal nondestructive
                                                                          testing, so it is assumed that w = 0.
Hence, the ratio between a conductive
and a radiation component of heat                                            Internal heat generation is not a part of
transfer in a thin gas filled void is as                                  the fourier equation:
follows:                                                                                               I aT
                                                                                                       u. at
(14) Q,d        -     k
         Q,d    -  4dcrT3                                                    A steady state heat flow with no heat
                                                                          generation is described by the laplace
for air  (f5il7leodcv~oid13s 5(k°F~);0q.0_3d.qW,ct·-n1r>1·1K7-1f; oTr  ~  equation:
330 K
                                                                          (18) a"T +  ~0
d < 0.5 mm [0.02 in.J). The consideration                                      ax2
above implies that, in typical material
voids, heat transfer occurs mostly by                                        Steady state techniques are not used for
                                                                          thermal nondestructive testing, because
conduction; the contribution from                                         the temperature disturbances that exist in
                                                                          a solid around interilal discontinuities
radiation is negligible. This would be                                    diminish because of the diffusion of heat
                                                                          through the surrounding material. As a
predicted merely by the fact that the J'o...T is                          result, no thermal signature is generated
                                                                          on the surface that indicates the presence
low and the ambient temperature is low.                                   of a subsurface discontinuity.
Differential Equation of                                                      For the transient process of thermal
Heat Conduction                                                           nondestructive testing, the most
                                                                          appropriate equation is the fourier
The theoretical simulation of thermal                                     equation (Eq. 17), which is typically
nondestructive testing scenarios requires
the prediction of three·dimensional                                       analyzed.
temperature distributions in a solid body.
This is achieved by analyzing the dynamic                                 Material Thermal
heat conduction governed by the                                           Properties
following differential equation:
                                                                          The basic material thermal properties are
(15) cPaaTr                                                               thermal conductivity k, specific heat C
                                                                          and density p (see Table 2). 1•2 Thermal
          + (a)z (k7. a()Tz ) + w(x,y,z)                                  conductivity is a material heat transport
                                                                          characteristic responsible for the
where w{X,}~Z) is rate ofenergy generation                                attenuation of heat flux~ especially in
per unit volume in the medium (W·m-3),                                    the steady state case. Thermal
Cis specific heal U·kg-1·K-1) and pis density                             conductivity values in 1~1ble 2 depend on
(kg·m-3), Equation 15 is a general                                        the processing parameters of the
                                                                          manufacturer and are given for reference
expression of the conservation of energy                                  only.
for the medium in which heat is                                              The nvo dynamic characteristics are
                                                                          thermal diffusivity a and thermal
generated and propagated \Vhile                                           efhisivity e:
experiencing diffusion and absorption.
                                                                          (19) e ~ ~kCp
    Equation 15 accounts for the
anisotropy in the thermal properties of                                   See Table 2. Thermal diffusivity is a
                                                                          measure of the rate at which heat is
materials such as composites and is                                       diffusing through a material. Generally,
expressed with three different thermal                                    materials that have high thermal
                                                                          conductivity k also have high thermal
conductivities k.vkpkr The coordinates                                    diffusivity and respond more quickly to
                                                                          changes IY.T in their thermal environment
correspond to the three principal                                         than do materials with low thermal
directions. In the case of an isotropic                                   conductivity.
material, in ·which kx = ky = k1, the                                         Thermal effusivity e is a measure of the
expression \'•till be a simpler form of                                   ability of a material to increase its
Eq. 15. Using the fact that"~ k(Cp)· 1 is                                 temperature as a response to a given
                                                                          energy (heat) input. It is often referred to
thermal dif(ush'il)' (m2·s··l), Eq. 15 becomes:                           as thermal inertia and is used in calculating
      I aT            a'T a'T
                      ax'- + Oyz
(16)
         a ilt
                   +  a'T        + -w
                      az2
                                      k
56 Infrared and Thermal Testing
a thermal mismatch factor r that               Thermal diffusion length J.l is related to
characterizes a thermal contact between        the heat stimulation frequency f (Hz).
two bodies, 1 and 2:                           This parameter is used in the theory of
                                               thermal waves and expresses the fact that
(20) r ~  el - ez                              lower frequency thermal waves penetrate
                                               deeper into a material than high
          e1 + e2                              frequency thermal waves. Conversely a
                                               thermal wave of a fixed frequency will
r == 0 means no thermal mismatch, that         propagate deeper into a material of higher
                                               diffusivity than into a material of lower
is, the interface of two materials is not      diffusivity. In thermal testing, thermal
                                               diffusion length serves as an estimate of
detected on the surface; r == 1 specifies the  the depth at which a buried discontinuity
                                               could be detected in a one-sided test by
case when the second material is a perfect     applying heat stimulation of different
                                               frequencies.
conductor; conversely r = -1 is realized
                                                  One more thermal parameter is thermal
when the second material is a perfect          resistance R:
insulator.                                     (22) R ~ L
   Another dynamic thermal parameter is                           k
thermal di{fi1sion lmgth ~~ (meter):
(21) ~
- ...,,-
TABlE 2. Thermal properties of common materials in order of Increasing thermal conductivity. 1•2 Values
depend on manufacturer's processing parameters and are given for illustrative purposes only.
Conversions: 1 kg ~ 2.2 lb; 1 m ~ 39.4 in.; 1 mm ~ 0.039 in.; 1 K ~ 1 "C ~ 1.8 "F.
                           Density                                      Thermal  Dlffusivity a  Effusivity e
                          p (kg·m-3)
        Material                               Specific Heat Conductivity        (1o-<' m'·s-1) [W·>I(s)·m-2·K-1]
                                               C (J·kg-1·K-1) k (W·m-1·K-1)
Air (thin gaps)                 1.2            1005     0.070                     58.0                9.19
Rubber (soft)              1100                2010     0.130                       0.0588         536
Po!yisoprene                 913               1905     0.134                       0.0770         483
Polyvinyl chloride                                      0.140                       0.11           422
Polyaramld (1.)            1330                1047     0.142                       0.102          445
Polyaramid (II)            1330                1047     1.69                        0.121          485
Pine (II)                                      2512     0.174                       0.126          490
Epoxy resin                  550               1700     0.200                       0.09           667
Glass fiber plastic (1.)   1300                1200     0.30                        0.13           832
Glass fiber plastic (I!)   1900                1200     0.38                        0.17           922
Plaster                    1900                1005     0.233                       0.211          507
Water                      1100                4193     0.586                       0.140        1570
Graphite epoxy (.1)        1000                1200     0.64                        0.52           888
Graphite epoxy (II)        1600                1200     1.28                        1.04         1260
Zirconia                   1600                         0.65                        0.219        1390
Brick (red)                5100                 582     0.755                       0.505        1060
Glass                      1700                 879     0.879                       0.430        1340
Concrete                   2442                 837     1.51                        0.752        1740
Nickel superalloy          2400                 837     9.5                         2.60         5890
Steel AISI 316             8300                 440   13.4                          3.47         7190
Silicon nitride            8240                 468   16.0                          9.65         5150
Titanium                   2400                 691   21.9                          9.32         7170
Zirconium                  4500                 522   22.7                                       6440
Aluminum oxide             6570                 278   46.0                        12.4          11800
Bronze                     3970                 765   52.0                        15.2          13900
Steel AISI 1010            8800                 420   63.9                        14.1          14700
Graphite                   7830                 434  116                          18.8          13400
Silicon                    2300                 670  148                          75.3          15700
Tungsten                   2330                 712  174                          89.2          21100
Aluminum 2024-T6 alloy    19300                 132  177                          68.3          20700
Gold                       2770                 875  313                          73.0          28600
Copper                    19450                 134  365                         120.0          36500
Silicon carbide            9000                 406  490                         100.0          32300
Diamond                    3160                 675  660                         230.0          34100
                           3516                 502                              374.0
                                                                                                Heat Transfer 57
The heat flux Q through a wall of
                  thickness L is related to the overall
                   temperature difference across the wall ['}.T:
                  (23) Q ~ /',T
                                      R
                      Equation 23 is analogous to Ohm's law
                  in electric circuit theory. An analogy to
                  electricity is often used to solve complex
                  problems in heat conduction. In thermal
                  nondestructive testing, a discontinuity is a
                  thermal resistance site across which
                  temperature experiences a significant
                  change. This is the same as an electrical
                  resistance causing a drop in voltage.
                  Similarly, a multilayer wall behaves as N
                  thermal resistances in steady state in a
                   series:
                                                       N
                          "L..1k2·
                                                                         ioo} /
58 Infrared and Thermal Testing
PART 2. Heat Conduction in Sound Solids
Basic Models                                               and on a surface, where z:::::: 0, this is
                                                           simplified to:
In the classical heat conduction theory, a
solid is modeled with one of four simple                   (26) T(z ~ O,t)
geometric bodies: a semiinfinite body, a
plate, a cylinder or a sphere (l'ig. 2).                   where H' is the absorbed heat energy
A heat pulse or function can be simulated
with (1) an imtantaneous (Dirac) pulse,                    O·m-2), z is the depth into the material
(2) a square pulse, (3) a st<1' function or (4)
a periodic (harmonic) function (Fig. 3).                   and Eq. 27 expresses the fourier
Taking into account that surface heat                      number Fo:
*transfer could be either adiabatic (h ::: 0)              (27) Fo,,
or nonadiabatic (11 0), this yields 32 (4 x                   Square pulse heating of a material as a
4 x 2) possible combinations for the                       function of depth z is described by:
thermal testing of solids. Those most                      (28) T(z,t) = ~'[2k'r~,
useful for thermal nondestructive testing
are discussed below. Details of other
combinations are available elsewhere.3
Semiinfinite Body                                                     x  exp(-1-       )
(Thermally Thick                                                                 4Foz
Specimen)
                                                                      - 2 Foz -Fozh
The solutions of the fourier equation for                                          n
the adiabatic heating of a thermally thick
specimen are as follows. Instantaneous                                                    1
pulse heating of a material is described as                           x exp( 4Fo,- Fo,, )
a function of depth z:
                                                                      - Ertc----b-
                                                                               z,rFoz
exp(--(25) T(z,t) ~ Wa 1       1                                      +  Erfc      1           ]
                    kz )n·FOz4Foz )
                                                                               2~Fo,,
FIGURE 2. Solid models: (a) semiinfinite body; (b) plate;  FIGURE 3. Types of heating: (a) instantaneous pulse;
(c) disk; (d) sphere.                                      (b) square pulse; (c) step function; (d) periodic function.
(a) (c)                                                    (a) (c)
                    [~_]
                                                                                      (b) (d)
( )(b) (d)
                                                                                                       Heat Transfer 59
On a surface where l = 0, this is                 (36) T(z,t)  g    [Erf2c-1-~
simplified:
                                                               lr
(29) T{z ~ O,t) ~ 2;L (ft - h )
                       e,l n                                   - exp(Biz+Bi~Fo2 )
Here Q is the absorbed heat power                              xEric(2ft·o, + m,~)]
(W-rn-2) and
                                                  On the surface, where z = 0, Eq. 36
(30) Fo,
                                                  becomes:
   Step function heating of a material is         (37) T(z ~ O,t) ~ [1 - exp(H2at)
described as a function of depth z:
(31) T(z,t)  Qz                  [2 ~.I:_F:IoT ,                   X Erfc(HFt)]
             k
                                                  The biot criterion Biz is defined by Eq. 38:
                                                  (38) Bi,     lrz
                                                               k
                                                  H expresses the ratio of heat transfer
                                                  coefficient h to thermal conductivity k:
and on a surface, where z = 0, this is            (39) H ~ 1_1_
simplified to:
                                                                       k
/I(32)
T(z ~ O,t)   ~                   2Q                  The contribution of surface heat
                                  e ~n            transfer to a temperature signal could be
                                                  evaluated by the ratio of transient
    Harmonic heating (the alternating             nonadiabatic ('/~1ad) and adiabatic (Tad)
solution component only) of a material is         temperatures on the surface of a
described as· a function of depth z if z > 0.     semiinfinite body given by Eqs. 32 and 37
                                                  for step function heating:
(33) T(z,t) ~ _zeQ,J_;
                                                                    -[;. l-exp(x2)Erfcx
      x ex+~)                                                        2X
             x ex+(wt-;- %)]                      where
                                                  (41) X ~ H .,fai
where e is thermal effusivity. On a surface          The ratio is shown in Fig. 4 as a
where z = 0, this is simplified to:               function of x. The divergence between the
                                                  two solutions is less than one percent if
                                 Q                H\l(at) < 0.011. As an example, in a
zej;(34) T(z~O,t) ~
                                                  graphite epoxy composite, the analysis
                                 expHwt-%)]
                                                  time in laboratory conditions must be
                                                  within about. 1 s to neglect surface heat
                                                  exchange with a one percent loss of
                                                  accmacy.
The variable ro is angular modulation             Plate
frequency (snr1) and Q is defined by:
                                                  The solution of the fourier equation for
(35) Q ~ ~n [1 + cos(cot)j                        the adiabatic heating of a solid plate by
                                                  an instantaneous pulse as a function of
where ~n is maximum absorbed heat                 depth z is:
power (\'V·m-2).
   The solution of the fourier equation for
IWtWdiabatic step function heating is given
by the following expression:
60 Infrared and Thermal Testing
(42) T(z,t)  lVa + [lVa                            (44) )nro
                           Lk tk
                           Lx 2 cos(nnZ)                                       t,[1 2x +  exp(-n2 n2 ro)]
                                   11=1                          where the subscripts pi and sib denote a
                                                                 plate and a semiinfinite body respectively.
                           x exp(-n2 .n2 ·Fo)]                   The latter function is shown in Fig. S to
                                                                 clearly illustrate the fact that the
              and on a front (heated) surface, where             divergence between the temperatures of a
              z = 0, this becomes:                               plate and a semiinfinile body made of the
                                                                 same material diminishes for thicker
              (43)                             lVa  +  [  -W-o:  plates and at earlier times. To ensure that
                                               Lk          Lk    such divergence is less than one percent,
                                                                 the fourier number must be less than
                           x 2I                                  0.19. As an example, for a graphite epoxy
                                                                 composite, the time of analysis must be
                                   11"'1                         less than 1.5 s for a 2 mm (0.08 in.) thick
                                                                 plate and less than 9 s for a S mm
                           x exp(-n2 .rt2 ·Fo)J                  (0.2 in.) thick plate.
              Here Lis the plate thickness and Z = z.L-1
              is the dimensionless in~depth coordinate.
              Plate versus Semiinfinite                          FIGURE 5. Comparison of adiabatic solutions for plate and
              Body Solutions                                     semiinfinite body with instantaneous pulse heating.
              The solutions of the fourier equation for a                1.25      7
              semiinfinite body are simpler than those                                                   -
              for a plate. Jt is obvious that the                        1.2
              divergence between these solutions                                 I
              increases in time, with more heat                  '·,_:" 1.15
              penetrating into the material. It is                              I
              convenient to show the corresponding               >--"
              condition as a function of time and                        1.1   1/
              thermal properties in the case of
              instantaneous pulse heating. Combining                     1.05
              Eqs. 26 and 43 produces:
                                                                               0.2 0.4 0.6 0.8 1.0
FIGURE 4. Comparison of nonadiabatic and adiabatic                                                                      Fo
solutions for semiinfinite body (step function heating).
                                                                 legend
              ~                                                     Fo-= fourier number (Ht ·L-1), where Lis material thickness (rnm), tis time
                                                                         (second), and u is thermal diffusivity (ml·s--1)
                                                                    Tpl = temperature of plate
                                                                   T,,b"' temperature of semiinfinite body
        0.98        ~
',u_, 0.96                 f'-.._
 ~                          -.......
he                             ~
       0.94                                         ~
       0.92
              0.02 0.04 0.06 0.08 0.1
legend
     h =heat transfer coefficient (W·m-l.K-1)
      k = thermal conductivity (W·m-l.K-1)
  r.d = adiabatic temperature
  Tmd = transient nonadiabatic temperature
    a = thermal diffusivity (m2·s-1)
     t = time (second)
                                                                                                            Heat Tran~fer 61
PART 3. Heat Conduction in Solids with Buried
Discontinuities
Coating on Substrate                               (48) T"t•  r.- [I 2~Qc ~ + :f(-r)"
Heat transfer in solids with buried                                               11=1
discontinuities is not easily understood.
Some useful information may be obtained                       l FJ]x(exp-t-•
by modeling the in-depth propagation of
pulsed thermal energy as the                                          1
superposition of thermal waves. Then, by
analogy 1.vHh optical waves, refraction and                   ~-n1t-• Erfc'/l
reflection phenomena may be introduced
at the boundary between two materials.                Here t* = n2·L2·ac-1 is the thermal
The simplest solutions arise for a coating
on a substrate material, with the latter           transit time for the nth reflection of a
being represented by a semiinfinite body.4
                                                   thermal wave, the subscript c specifies a
    For instantaneous pulse heating:
                                                   coating and a thermal mismatch factor r
(45) T F(t) ~ ~W
                                                   is determined for the interface between a
                     en/nt                         coating and a substrate.
       + [__!:I'_                                     The thermal mismatch factor r is a
               e,.{rrt
                                                   good indicator of the thermal contact
        x 2IJ-r)"
                                                   between two materials. \'\1hen r = 0, this
                                            Jlo=]
                                                   indicates no thermal mismatch between
                 x exp(<)]
                                                   two identical materials. Cases \Vhere r > 0
    For square pulse heating:
                                                   correspond to a more effusive substrate;
(46) T'(t) ~
                                                   values of r < 0 mean that a substrate is of
                                                   lower effusivity. The graphical
                                                   presentation of these cases is shown in
                                                   Fig. 6 for step function heating. During
                                                   the initial period, the whole structure
                                                   behaves as a semiinfmite body and the
                                                   corresponding function is a straight line
                                                   in a graph ofT versus ~t. During this time
                                                   the interface between a coating and a
                                                   substrate will not be detected. During
                                                   thermal transit time, the surface
                                                   temperature of the coating starts to
           + 2jt ,L}r)" Ft                         FIGURE 6. Influence of substrate thermal properties on
                                                   behavior of surface temperature (step function heating of
                          11=1                     thermally thick sample).
              2Jl-0,                                                        Insulating coating -------:;.•••.....
                                                   ... .... ... "...'...Thermally thick specimen
where:     X I,(-r)" r(t - th )]                                                        •••'                  ~'
(47) F(t)        11=1                                                              • .,.. """ ..,. - -
            t:)Exp(_                                                           •••• *"" '
            p- (:t;-rrt-* Erfc ~t*                                                              Conducting
            t \1 t                                                                              coating
For step function heating:                                            Thermal transit tinlf•
                                                                  Square root time (\1s)
62 Infrared and Thermal Testing
deviate from the temperature that would          This restriction limits the choice of a
appear in the absence of a coating. The       temporal step, once a spatial ·step is
sign of the difference between a real and     fixed. In two-dimensional and
expected temperature depends on thermal       three-dimensional models, this may lead
mismatch factor magnitude.                    to a very long computation time. The
                                              other type of calculation scheme is called
Numerical Modeling                            implicit and makes it possible to overcome
Principles                                    this difficulty.'
Analytical solutions are of little help       Transition Criteria for
when simulating practical test situations     One-Dimensional to
where a material is anisotropic and, more     Three-Dimensional Models
importantly, where discontinuities have
complex shapes.                               The discontinuities that may be
                                              analytically simulated must be very large
   With growing computer power,               laterallyi otherwise lateral heat diffusion
numerical techniques are becoming             will reduce the temperature signature that
increasingly common in the analysis of        would reveal a discontinuity and reduce
three-dimensional discontinuities. As         the length of time during which the
required by the geometry of a test part,      discontinuity may be observed. Lateral
the heat diffusion equation may be            heat diffusion is affected by the type of
written either in cartesian coordinates as    the material, the discontinuity depth and
in Eqs. 15 to 18 or in cylindrical or         time. The ratio rd·f-1 between
spherical coordinates. 'r\rith a transient    discontinuity radius and depth appears in
numerical technique, the second partial       a rule of tlzumb to define whether a
derivative with respect to coordinates and    discontinuity may be regarded as
the first partial derivative with respect to  one-dimensional (infinite) or
time are replaced with their finite           three-dimensional (finite). Approximate
difference substitutes:                       rd·I-1 values are given in Table 4 and Fig. 7
                                              for two different materials and two
      a2 r  1i+l + 1f-t - 2T;
(49) ax2
                 (Lix)z
(50)  ar    rt+1  -  rP                       FIGURE 7. lateral discontinuity size:
      at     '       '                        (a) one-dimensional cartesian model;
                M                             (b) two-dimensional cylindrical model.
where subscript i denotes spatial mesh        (a)
points (the x dimension), superscript p
specifies time mesh points (current time),
t1X is spatial step and At is time step.
   Equations 49 and 50 may be
substituted into the one-dimensional
presentation of Eq. 51:
                                                   Discontinuity
This simple numerical scheme allmvs the       (b)  It
calculation of the temperature in the
mesh point T1at a time t + At if all                  ~ !I                              I
temperatures at a time tare known. An
extension of Eqs. 50 and 51 to two and             @--:----!o-,,
three dimensions can easily be made.s              I~
   Equation 51 specifies an explicit          legend
calculation scheme that for mathematical
reasons requires stability:                      d = discontinuity thickness
                                                 L = sample thickness (meter}
      (L1x)2                                      I = discontinuity depth (meter)
                                                Q = heat transfer or heat flux (W·m-2)
(52) $ 2                                        rd = radius of discontinuity
           aM
                                                                                           Heat Transfer 63
different heating times. It is evident that                                     discontinuity to.Tm and the time of its
one-dimensional discontinuities in metals                                       optimum observation tm(to.Tm) and (2) the
must be laterally larger than similar                                           maximum rwming contrast
discontinuities in materials whose thermal
conductivity is lower. Also, lateral heat                                       (53) c~~n =
diffusion is less at earlier times ~ it
increases with time.                                                            and the time of its optimum observation
                                                                                tm(C,Rm). The maximum temperature
    Moreover, such a criterion can be                                           signal could be expressed in the form of
generalized in the case of anisotropic                                          the normalized contrast:
materials with the weighting ratio Y(krkr1)
where k1 and kr are thermal conductivities                                      where Tmax is the maximum surface
in the I and r directions, respectively. The                                    temperature that occurs above a
criterion ratio thus becomes                                                    discontinuity free area at the end of
(rd·I-1H(kl'k,-1). Note that such extension                                     heating. Both running and normalized
is also valid for isotropic materials such as                                   contrasts are dimensionless and
steel because in such a case k1 ;= kr.                                          independent of heating power because
                                                                                both aT and Tare linearly proportional to
   The material properties in Tables 3                                          Q. (Note that, in transient thermal
and 4 were obtained for an air filled                                           nondestructive testing, Tis not the
discontinuity (void) in a material where                                        absolute temperature but is the
discontinuity depth I= 0.1 L and where                                          differential temperature, the temperature
discontinuity thickness d = 0.1 L. Lis wall                                     in excess of the ambient or initial
thickness.                                                                      temperature.)
Basic Features of Transient                                                     Discontinuity Detection Criteria
Thermal Nondestructive
Testing                                                                         Generally, a temperature signal can be
                                                                                detected if one of its characteristics, such
Informative Parameters                                                          as amplitude, exceeds the noise level.
                                                                                Often this is expressed with the simple
Two pairs of iu{ormative parameters are of                                      formula:
practical interest in transient thermal
nondestructive testing: (1) the maximum                                         (55) s > 1
temperawre signal above or under a
                                                                                   An expression of this form can be
TABLE 3. Material properties for air filled                                     written for any chosen informative
discontinuity.                                                                  parameter or combination of them. In
                                                                                thermal nondestructive testing,
                      Conductivity K Dlffusivlty a                              discontinuities are typically detected by
                                                                                evaluating temperature signals f1T(t) that
    Material (W·m-'·K-1) (m'·s-')                                               occur over discontinuities. Then Eq. 55
                                                                                can be written as follows:
Steel                              32.0     7.3 X lQ--6
                                    0.64    5.2 X 10-7                          (56) > 1
Graphite epoxy                      0.07    5.8x10-5
Air defect
TABLE 4. Approximate values of lateral                                          where <Jnd is the standard deviation of the
defect size corresponding to transition                                         temperature in an area without anomalies
from two-dimensional cylindrical model                                          and K is the coefficient that determines
to one-dimensional cartesian model in                                           the reliability of decision making
material where discontinuity depth                                              (typically 1 ~ K ~ 3).
I= 0.1 L and where discontinuity
thickness d = 0.1 L.                                                               Assume that there are both additive
                                                                                and multiplicative noise:
  Material                          Heating Critical rd·l-1Values
Thickness L                                                                     The magnitude of additive noise <Jadd does
                                     Time Graphite                              not depend on a measured parameter
mm (in.)                               (s) Epoxy Steel                          such as sample temperature. In thermal
                                                                                nondestructive testing, this type of noise
               . . v··~· --~-.-~,  -----.,-·~-~----· -~~--'- -~>·---~~~ ~-~---  is typically determined by the infrared
1.0 (0.04)                            0.01  23  10
1.0 (0.04)                         100.00   30  10
5.0 (0.20)                                  23  10
5.0 (0.20)                            0.01  32  10
                                   100.00
64 Infrared and Thermal Testing
system temperature resolution !lTres and                           Time Evolution of Temperature
some reflective noise O'ref(t) (in the ideal                       Signals
case cradd = !lTres)· Multiplicative noise is                      The variation of temperature versus time
proportional to a measured signal. For                             in transient thermal nondestructive
                                                                   testing is of particular interest because the
example1 variations .6e in emissivity                              analysis of the temporal peculiarities of
influence useful temperature signals so                            the signal is important in reducing noise
that output infrared signals =otleT(t), where                      and in characterizing subsurface
                                                                   discontinuities.
T(t) is the excess specimen temperature
(here it is assumed T(t) is not dependent                             Figure 8 is composed of six plots that
                                                                   show the temperature evolution on the
on local absorptivity). This means that the                        front and rear surfaces of a 5 mm (0.2 in.)
magnitude of the multiplicative noise can                          thick graphite epoxy specimen heated
                                                                   with a 0.01 s heat pulse. The specimen
be approximated as <imult = CnmnT(t)l                              contains an air filled circular
where C11tun is the noise running contrast                         discontinuity that has a radius of 5 mm
independent of time. Then, the simple                              (0.2 in.) and a thickness of 0.1 mm
model of the signal-to-noise ratio (SNR) in                        (0.004 in.), at a depth of 0.25 mm
                                                                   (0.01 in.), except for Figs. Be and Bf. The
transient thermal nondestructive testing                           thermal properties of the graphite epoxy
                                                                   are treated as isotropic. Tables 3 and 4
can be presented:                                                  show discontinuity parameters.
(58) SNR     tlT(t)                                                   I~igures 8a and 8b show temperature
                                                                   versus time for the front and rear surfaces,
                       ['""()j2                                    both over a discontinuity and over sound
                                                                   material (without a discontinuity). W'ith
          K ~1LlTn2; + cr2nf(t) + C., T t                          short heating pulses, a noticeable
                                                                   difference between Tnd and Td appears in
Equation 58 shows that the                                         the cooling stage. For longer pulses or
                                                                   discontinuities nearer the surface, the
signal-to-noise ratio varies in time/                              difference may. also appear during the
reaching a maximum value at a particular                           heat pulse.
time that is the best experitnental
observation time: tm(s ~ maximum). The                                 In Figs. 8c and 8d, the temperatures
maximum signal-to-noise ratio that can                             AT= Tn- Tnct for the front and rear
                                                                   surfaces are shown in comparison with
be achieved in a thermal nondestructive                            the running temperature contrasts:
test is:                                                           (60) c'""
(59) SNRmax                                                        (Below this point in the text the
                                                                   superscript nm will be omitted).
It is controlled by the temperature
                                                                      These informative parameters reveal
resolution of the infrared thermographic                           the maximum temperatures for both
                                                                   sample surfaces but they occur at different
equipment used. In modern infrared                                 observation times lm. On the front surface
                                                                   (Fig. Sc), after heating, the maximum
thermog   r0a.0p0h1icoicm. aWgehresr1e.6tThreesrmcaanl  be as  of  temperature signal ATm occurs earlier than
small as                                                noise      the maximum contrast Cm. Conversely, if
                                                                   both maximums occur during a heat
the environment is higher than 0.1 K                               pulse, Cm surpasses 11Tm· Typically1 the
                                                                   maximum signal-to-noise ratio occurs at
(0.1 oc = 0.18 °F), temperature resolution                         the time tm(Cm) that may be regarded as
                                                                   the optimum obseroation time in a
is not a limiting factor. In fact1 reflective                      one-sided test. On the rear surface, the
additive noise can be very high,                                   maximum contrast appears very early
                                                                   when the temperature signal is rather
partiCularly if temperature is being                               small. Therefore, in a two-sided test
                                                                   (Fig. 8d), the optimum observation time
measured during heating. Even after                                occurs when the temperature signal
                                                                   exceeds the temperature resolution of a
heating, the residual heat energy can                              recording device and meets the conditions
                                                                   of Eq. 55. Notice that, because of the
cause significant indications of reflective                        accepted convention for .6T, rear surface
                                                                   temperature signals are negative.
nature. However, an organized test can
                                                                       From here on the analysis of
often reduce the ratio of the                                      discontinuities will primarily be
multicomponent noise to surface noise
caused by variations in optical properties
such as endssivity and absorptivity.
Unpainted objects max have quite a high
level of noise with Cr~~~tse reaching 20 to
100 percent for corroded and greasy
metals.6 Nonmetallic materials are
characterized with noise of about
C1~1~se""' 4 to 6 percent. The best way to
reduce noise (down to one to two percent)
is generally to apply a flat black high
emissivity coating to the surface. The
coating is commonly applied by using
black spray paint or other coatings. Water
washable paint is available and normal
aerosol spray paint may be removed using
a solvent such as acetone.
                                                                   Heat Transfer 65
FIGURE 8. Temperature evolutions on front and rear surfaces of 5 mm (0.20 in.) thick graphite epoxy
specimen: (a) front surface excess temperature versus time; (b) rear surface excesS temperature versus time;
(c) front surface temperature and running contrast cruo versus time; (d) rear surface temperature and running
contrast cruo versus time; (e) front surface temperature for three discontinuity depths, versus time; (f) rear
surface running contrast for three discontinuity depths, versus time. Parameters: discontinuity thickness d =
0.1 mm (0.004 in.); discontinuity depth I~ 0.25 mm (0.01 in.); heat transfer Q = 100 kW.m-2; discontinuity
radius rd = 5 mm (0.2 in.); heat pulse duration th = 0.01 s.
(a)                                                                      (d)
  9"      2.S                                                                 0 \•                       /'[run
  c15
                \•.                                                      9" -0.05                 \I  I
   ,_~E.
          2.0 \                                                          15 -0.10                 v      ······..........................
   ,_'
          1.S                                                            c,._~E. -0.15            /\
                                                                                 -0.20            J !J.Tx 100
               \ \\D··..
                                                                         'h -0.25
          1.0
                                                                                 -0.30
          0.5
                                                    .............                            0 5 10                                        1$
                                                                                                         Timet (s)
          0                                      2                 3
                0
                                 Timet (s)
                                                                         (e)
                                                                                        0.8
(b)
              0.16             /.              ~D                                       0.6
     9" 0.12                   /"            . ...                            ~ 0.4
     15                    /                 10 15                                      0.2
     c 0.08                           5                                                 0
     ,._~E.                                                                                  0
     ',_ 0.04
              0
                     0
                                 Time t (s)                                                                       2                        3
                                                                                                         Time t (s)
(c)                                                                      (f)
                1.2                                                                  0
                 !\
                     i :,
                0.8
               0.6               ,~·· .....~                                        -0.004
     ~                                ·········........ ~                     ~
               0.4                                                                  -0.008               5 10                              15
                                                                                                             Timet (s)
               0.2                                                                  -0.012
                                                                                               0
                0
                     0                          2                     3
                                 Time l (s)
legend
  cwn "' running contrast
     D "' having measured discontinuity
       J = defect depth (meter)
   ND = having no measured discontinuity
      T ~ temperature
  Tamb = ambient temperature
66 Infrared and Thermal Testing
FIGURE 9. Temperature evolutions on front surface of 5 mm (0.20 in.) graphite epoxy
specimen: (a) front surface temperature in sound area versus time (80 cycles); (b) front
surface temperature in both sound and discontinuity areas versus time (three cycles); (c) front
surface temperature signal and contrast versus time (three cycles). Parameters: air filled
discontinuity thickness d ~ 0.1 mm (0.004 in.); stimulation frequency f ~ 0.2 Hz, close to
optimal; air filled discontinuity at depth I~ 1.0 mm (0.04 in.); heat transfer Q ~ 10 kWm-2;
discontinuity radius rd = 5 mm (0.2 in.); modulated heating.
(a)
               200                        ;~
     9                       ~----
     0                                         Sound area
                                      200 300 400
     ~
     f-,0
     <1(,
     ~<;.; 100
     /-~6ge
      ~u            /
     1 /;":J"o
     ~
      ~
     0.
     E0
     ~ 100
                             Timet (s)
(b) , - - - - - Q"' (Oml2) [1 ~cos (2nfi)J
     9              //"'·..                                                   Phase lag
     0 30           /\
     ~
     f-,C
             20
     <1u2
     ~t\
     c~
     .~c"-
     u§
     ~u
     ;":J"o
     ~
     ~ 10
     0.
     E
     ~
             0 ,•                                                                        15
                                                  5 10
                                                         Timet (s)
(c)
                                                                           ,
             3.0 ' ' ' ........ ' ' '
             1.5
                           0                                                  10         15
                                                                   5
                                                                                                   Heat Transfer 67
                                                                   Timet (s)
Legend
   cnm "' running contrast
     Td = temperature in area with measured discontinuity
    T00 =temperature in area without measured discontinuity
    tJ.T =temperature change
performed with cylindrical geometry to       (Fig. 9b). On the surface of a semiinfinite
permit reasonable accuracy in simulating     body this phase lag is -45 degrees. The
real discontinuities while keeping the       phase difference between sound and
analysis relatively simple.                  discontinuity areas enhances
                                             discontinuity detectability because a
   Figure Be shows the front surface         higher signal·to·noise ratio appears in the
temperature difference versus                phase domain. Because a phase lag can be
                                             expressed in terms of time, the thermal
discontinuity depth, and Fig. Bf shows the   wave approach uses the advantages of
rear surface running contrast versus         phase analysis in the same manner that
discontinuity depth. Figures 8e and 8f       the pulsed approach takes advantage of
show the temperature evolutions when         time domain treatment.
the discontinuity is at three different
depths, 0.25, 0.5 and 1.0 mm (0.01, 0.02        The concept of an absolute
and 0.04 in.). Because of the diffusive      temperature signal and dimensionless
nature of heat conduction, the front         contrast is also valid for thermal wave
surface temperature versus time behavior     techniques. Examples of time evolution of
is very sensitive to the depth of            11T and Care shown in Fig. 9c. Both
discontinuities. Deeper discontinuities      informative parameters experience
produce lower amplitude temperatures at      oscillations. The constant component of
later times (Fig. Be). This difference may   the temperature signal increases in time,
be explained by using thermal wave           at least within the first wave periods.
theory, in which the harmonic                Meanwhile the temperature contrast
components of a heat pulse propagates in     reaches a maximum at about 5 s after
depth depending on wave frequency            turning on a heat source that produces a
while also experiencing phenomena of
interference, reflection and attenuation.    FIGURE 10. Optimal detection parameters in 5 mm (0.2 in.)
Rear surface signals do not vary as much
with discontinuity depth (Fig. 8f), for a    thick graphite epoxy specimen versus discontinuity depth:
heat pulse must pass through the whole       (a) front surface; (b) rear surface. Parameters: discontinuity
specimen. Moreover, the highest              thickness d = 0.1 mm (0.004 in.); heat transfer Q = 106
temperature signals on the rear surface      W·m-2; discontinuity radius rd = 5 mm (0.2 in.); heat pulse
occur for discontinuities located in the
middle of the specimen (see the curve for    duration th = 0.01 s.
I= 2.5 mm (0.1 in.) in Fig. 8f).
                                             (a)                                                                                        20
   When using a thermal wave technique,
a heating function is often described:                   3.0
                                                                                                                          ......
%"(61) Q =                                    ~         2.4                                                            /                16
            [1 - cos(2nft)j                  "c' ~                 \                                       lm... ..··/                  12 3
                                              ~u        1.8                                                                                  ..!'
                                             .Co        1.2
                                                        0.6                                                                                          ~
   Figure 9 is composed of three             a~  0f\8Tm\u"0                          ....······
additional plots that show the                                                    ..····                                                8E
temperature versus time for the same             c               ........····.······.x:.._..·1..-..-··-····                                  F
material and discontinuity type, with the        ~._!'
discontinuity at a depth of 1.0 mm                                                                                                      4
(0.04 ln.). The surface temperature in both  E"<l
materials with and without                   ~                                                                                          0
discontinuities reflect both the sample
thickness and the material. Typically, a                      0       2 34                                                           5
thermal wave technique known as a
classical photothermal technique is applied  (b) 0.16                 Discontinuity depth I (min)                                    8~
to rather thin materials requiring about 5
to 10 cycles to reach a steady state             ~                         ..· .......l.m...... ....                                     ·Ep
condition where amplitude and phase
characteristics of surface thermal waves     "o"' 0.12                           2 34                                                7      c
are further analyzed. Recently, this                                  Discontinuity depth I (min)
technique has been realized as a lockit1       .Co                                                                                          0
thermograpllic technique based on                                                                                                       ·~
recording an alternating thermal wave          u"
component over large areas. In thermal                                                                                                      i:~
nondestructive testing of rather thick          ~0                                                                                      ~~
materials, such as a 5 mm (0.20 in.)                                                                                                 6
graphite epoxy specimen, even the 80         l~~"' 0.08                                                                                 .g.s
cycles shown in Fig. 9a are not enough to
get rid of influence of the constant           E<l 0.04                                                                                     E
component.                                    ~                                                                                             ~
                                                                                                                                     5
   The most remarkable feature of a                          0                                                                              E
surface thermal wave is its phase Jag                           0                                                                       K
relative to a periodic heating function                                                                                                   0
                                                                                                                                     4
                                                                                                                                  5
                                                 legend
                                                     tm = optimum observation time (second)
                                                   llTm = maximum temperature signal above (a) or under
                                                             (b) discontinuity
68 Infrared and Thermal Testing
thermal wave having a frequency of                                                          power is Q = 106 W·m-2 and corresponds
               0.2 Hz. This frequency is close to optimal                                                  to a specimen excess temperature of
               when detecting an air filled discontinuity                                                  -110 K (-110 °C = -200 ol') at the end of
               in carbon fiber reinforced plastic at the                                                   heating. As expected1 the optimum
               depth of 1 mm (0.04 in.).                                                                   observation time increases with
                                                                                                           discontinuity depth reaching tm ~ 10 sat
               Discontinuity Depth                                                                         I ~ 3 mm (0.125 in.).
               An illustration of how discontinuity                                                           The corresponding graphs for the rear
               depth affects optimal detection                                                             surface are symmetric in regard to a
               parameters is shown in Fig. 10 for a 5 mm                                                   discontinuity depth of I= 2.5 mm
               (0.2 in.) graphite epoxy specimen. On the                                                   (0.10 in.) (Fig. lOb). Therefore, a
               front surface1 the temperature signal                                                       two-sided test is optimal for detecting
               decays rapidly for deeper discontinuities                                                   discontinuities in the middle of a
               (Fig. lOa). If the temperature resolution of                                                specimen.
               infrared equipment is assumed to be 0.1 K
               (0.1 oc = 0.2 °F) the depth limit for                                                       Discontinuity Thickness
               detected discontinuities will be about
               3 mm (0.125 in.). Note that the absorbed                                                    The thicker a discontinuity is, the higher
                                                                                                           the relative temperature above it will be,
FIGURE 11. Optimal detection parameters in 5 mm (0.2 in.)                                                  and the later the optimum observation
thick graphite epoxy specimen versus discontinuity thickness:                                              time will be. This statement is Hlustrated
(a) front surface; (b) rear surface. Parameters: discontinuity                                             in Hg. 111 where the data are p'resented
depth I= 2.5 mm (0.1 0 in.); heat transfer Q = 1O' W-m-2;                                                  for an air filled delamination located in a
                                                                                                           5 mm (0.2 in.) thick graphite epoxy
discontinuity radius rd = 5 mm (0.2 in.); heat pulse duration                                              specimen at a depth of 2.5 mm (0.10 in.).
                                                                                                           In this case, the temperature profiles are
th = 0.01 s.                                                                                               similar for both the front and rear
                                                                                                           surfaces.
     0.6                                      .:>-- 11
                                                                    10                                        For discontinuity characterization in
     0.5                                                                                                   perturbing temperature distributions1 the
                                             ..... .....l.m..........                                      discontinuity thickness is less important
     0.4       .k" ~.............. ...                                                                     than the discontinuity thermal resistance:
     0.3                                                     93
            ....··"/                                                                                       (62) R,
     /0.2                                                        8 .f
                                                                                                        ~  Discontinuity lateral Size and
     I0.1                                                                E                                 Configuration
     0
                                                                   7F                                      The diffusive nature of heat conduction
            0 0.2 0.4 0.6                                                                                  makes detection parameters rather
                                                                     6                                     sensitive to variations in discontinuity
                                                                                                           size. Both discontinuity thickness and
                                                              s                                            lateral dimensions influence the
                                                                                                           magnitude of the disturbance in heat flow
                                                0.8                                                        caused by the presence of a discontinuity.
                                                                                                           As discontinuity thickness increases/ the
               Discontinuity thickness d (mm)                                                              surface temperature signals increase and
                                                                                                           appear at later times (l'ig. 11). Lateral
(b)                                                                                                        discontinuity dimensions are often
                                                                                                           confronted with discontinuity depth. It is
P' 0.36        ·y -------I--                          1m  9.S 3                                            intuitively obvious that a sufficiently large
                             .............                                                                 discontinuity could make heat flow
0 0.30                                       .......      9.0 .f                                           one-dimensional. This is illustrated in
                                                                                                           Fig. 12 where variations in tlTm and tm are
~              '/_...·········  ···········                  ..E~,                                         shown in as a function of rd·f-1. It may be
                                                                                                           observed that, for graphite epoxy1 lateral
...!' 0.24                                                8.S c                                            heat diffusion stops if 2rd·l-1 > 5. For the
                                                                                                           more isotropic aluminum/ this condition
<1                                                           ·g0                                           becomes Zrd·/-1 > 10. The accepted rule of
                                                                                                           thumb states that, in a one-sided thermal
 ~                                                        8.0                                              nondestructive test1 the lateral
;-··0c> 0.18                                                                                               discontinuity size should be twice as large
                                                                ~                                          as the discontinuity depth (2r,ri-I > 2).
.~c                                                             .lS                                        Fig. 12b demonstrates the validity of this
u 0.12                                                                                                     rule of thumb. Because time parameter tm
~           l                                             7.S    0                                         is stable compared to variations in lateral
iel 0.06                                                         E
a~.
E0                                                        7.0 ·Eg"_
,". 0 0.2 0.4 0.6 0.8
                                                          6.S 0
               Discontinuity depth I (min)
legend
     tm "' optimum observation time (second)
  ATm =maximum temperature signal above or under discontinuity
                                                                                                           Heat Transfer 69
discontinuity size, domain treatment is                              are illustrated by Fig. 13, which shmvs
                    preferable in characterizing discontinuity                           thermal images taken at three different
                    depth.
                                                                                         times. The three discontinuities simulate a
                        Another aspect of lateral heat diffusion                         50 percent reduction in thickness caused
                    is how temperature distributions are
                    influenced by the planar configuration of                            by corrosion in a 2 mm (0.08 in.) thick
                    discontinuities and how neighboring
                    discontinuities interact. These phenomena                            steel specimen and each has the same area
                                                                                         of 25 mm2 (0.06 in. 2) (Fig. 13a). The most
                                                                               ,--.,-:,
                                                                                         accurate representation of discontinuity
FIGURE 12. Optimal detection parameters versus lateral                                   shape and the best resolution of the
discontinuity size on front surface, where                                               separation of neighboring discontinuities
llTm(rd'l-1)/llTm(rd'l-1 ==):(a) schematic; (b) maximum
temperature signal versus rd.J-1; (c) optimum observation                                occur in the thermal images taken at
time versus rd.J-1.                                                                      earlier times (Fig. 13b) because of weak
(a)                                                                                      lateral heat diffusion. At the optimum
                                                                                         observation time (Fig. 13c), some
                                                                                         temperature signals reach maximum
                                                                                         values but the effect of heat diffusion
                                                                                         becomes noticeable. At this time, a five
                                                                                         fold change in discontinuity size causes
                                                                                         about a five fold change in fiT1111 even if
                                                                                         the discontinuity area is constant
                                                                                         (compare the signals for discontinuities 1
                                                                                         and 3). Increased heat dissipation at later
                                                                                         times will significantly distort the thermal
                                                                                         signatures of the discontinuities
                                                                                         (Fig. 13d).
td==O.l mm
(b)            0.75            Graphite epoxy                                            FIGURE 13. Influence of discontinuity configuration on
               0.50                                                                      surface temperature distribution. Detection of 50 percent
   ~                                                                                     corrosion in 2 mm (0.08 in.) steel specimen when Q = 1Q6
               0.25                                                                      W-m-2 and t11 = 0.01 s: (a) discontinuity location where
           II                                                                            discontinuity 1 measures 1 x 25 mm2 (= 25 mm2 =
               0                                                                         0.04 in. 2), discontinuity 2 measures 2.5 x 10 mm2
     I.              0 5 10 1S                                                           (= 25 mm2 = 0.04 in.2) and discontinuity 3 measures
                                                                                         5 x 5 mm2 (= 25 mm 2 = 0.04 in.2); (b) temperature
  ;"!'                                                                                   distribution at 0.1 s; (c) 0.5 s; (d) 1 s.
                                                                                         (a) (c)
     ,;:\
                                                                                                           2
     ~                                                                                                                    3
   ;"!'                                                                                         D
         <l                                                                              (b) (d)
                                                                          20
                        Ratio of discontinuity size to distance, 2rd·f-1
(c)
               0.75 f/!5t,mln"m
                            ·  Graphite epoxy
               0.50
               0.25
               0                                           '              20
                     0         s 10 1S
                        Ratio of discontinuity size to distance, 2rd·J-1
legend
   d == discontinuity thickness
    I = discontinuity depth (meter)
   th = heat pulse duration (second)
70 Infrared and Thermal Testing
Heating Process                                      same brief contrast rise occurs at the end
                                                                      of any heating pulse. Notice that, at later
                      Each discontinuity type may require the         times, both the front surface temperature
                     optimization of a heating process that will      signal and the contrast may become
                     best reveal that type of discontinuity. The      negative because of faster cooling of the
                     signal-to-noise ratio can serve as a             material above a discontinuity (Fig. 14).
                     practical means for comparing the
                     effectiveness of different types of heating.        To summarize, three points should be
                     This usually leads to maximizing the             kept in mind.
                     temperature contrast C. Because the
                     contrast value is independent of Q, the            1. Heat pulse duration should be short
                     absorbed heat energy must meet the                    enough to ensure an adequate
                     requirements of Eq. 55:                               temperature contrast C above a
                                                                          discontinuity, to allow detection.
                      (63) ti.T(t) ~ si\ T,"
                                                                       2. Total heat pulse energy must be high
                    The heat must not damage the specimen,                enough to create a detectable
                    or the test would not be nondestructive.               temperature signal!'!T.
                    The maximum front surface temperature
                     Tmax occurring at the end of the heating          3. Total heat pulse energy should not
                     process must be less than the destruction            damage the surface.
                    threshold temperature Tdestr·
                                                                         The comparison between pulsed
                        Some aspects of the heating technique         heating and thermal wave testing is
                    are illustrated with Fig. 14 for a one-sided      presented in Table 5 for a one-sided
                    pulsed heating test. The highest contrast         thermal test of an air filled discontinuity
                    is achieved by a short duration heating           located I mm (0.04 in.) deep in aS mm
                    pulse (I" ~ 0.01 s). Following a longer           (0.2 in.) thick graphite epoxy specimen. A
                    heating pulse T11 = 25 s, the maximum             flash test results in higher contrast. In
                    contrast appears at about 7 s. Stopping           practice, hmvever, other factors may be
                    the heating pulse at 7 s causes a small rise      important to the optimization of a test
                    in the contrast whereas the absolute              procedure, such as hardware limitations
                    sample temperature starts to drop. This           and interference between a heater and
                                                                      infrared camera.
FIGURE 14. Influence of heat pulse duration on time
evolution of running contrast crun in graphite epoxy                  Type of Material
specimen. Parameters: discontinuity thickness d = 0.1 mm
(0.004 in.); sample thickness L ~ 5 mm (0.2 in.);                     Theoretically, subsurface discontinuities
discontinuity depth I~ 1.0 mm (0.04 in.); discontinuity               are detected best in materials that create
radius rd ~ 5 mm (0.2 in.).                                           maximum temperature contrasts. It is
                                                                      clear that, for materials of low isotropy,
                                                                      heat does not penetrate deeply into the
                                                                      structure and lateral diffusion is minimal.
                                                                      On the other hand, for materials of high
                                                                      isotropy, heat will penetrate more deeply
                                                                      into the material. Also, lateral heat
                                                                      diffusion becomes significant and may
                th"' 0.01 s
                      1'.                                             TABLE 5. Front surface maximum
    0.20                                                              temperature contrasts and their optimum
                   I\                                                 observation conditions in inspection of air
l \c               I       \                                          filled discontinuity in graphite epoxy,
                   I        \ :'•,
u2                                                                    where L ~ 5 mm (0.2 in.), I~ 1 mm
t;              I          \
                I                                                     (0.04 in.), d ~ 0.1 mm (0.004 in.) and
~   0.15                   \:
                                                                      rd =  5  1momc  (0.2 in.).  Conversion:
~c                                                                    1K              ~ 1.8 OF.
  K.._I '.4u0 I \i th = 7 s                                                 ~
l"
                I
~
~                                                                                                         Optimum
                I                                                                                        Observation
~                                                                                                        Time tm(Cm)
a~. 0.10
                                                                                                               (s)
--......E
ij~
              p ;I \ ...•,     \\                                           Heating   Temperature
                                                                             Time     Contrast Cm
                               \ '\
                                                                                        (K ~ oq
    0.05                            '  '     •,·..  '•,               Pulsed Heating              0.230        2.6
             I                                                             th := 0.01 s           0.185        8.5
             I                            '  " ....  '•'•,  ···.....                              0.127        7.0
                                                                           th""7s
         I;                                          ....                  th""25s                0.181        9.7
    0                                                                 Thermal Wave Heat           0.208        5.3
                                                                          t::o:O.l Hz             0.167        6.2
          0 10 20                                                         t"" 0.2 Hz
                                                                          t"" 0.5 Hz
                                    Time l (s)
                                                                                                                    Heat Transfer 71
FIGURE 1S. Dependence of optimal detection parameters on type of material: (a) schematic;
(b) maximum temperature contrast versus thermal conductivity; (c) optimum observation
time versus thermal diffusivity. Parameters: heat transfer Q = 1os W·m-2; heat pulse duration
th = 0.01 s. 1 mm = 0.04 in.
(a)
(b)
     2
     ~.                                             /
     \,)                                          v
     gt:                                       /  ~-- --A---·                              --.
                          v --/
     c                                                                                 100 300
                      ~ -·,..o-~ ~-
     8
                               1-o--
     ~
                                    0.5           5 10                             50
     il
     ~
      ~
      0.
     E
     ~
     E
      ~
     :·E•;;;
            0
              0.1
                                         Thermal conductivity kin W·m-1·K-1
(c)
                  10
     3                                   I
                                         I
     .1'
                                            I
     ..~E,
                                            Q
     .cg                                    \
     ~5                                                  \',
     1l                          ~ ~:':a.. ' ' ' ' '       !:- -------- --..
     0
     E
      ~
     ·Eg_
     0
             0
                                               10 100 1000
                                            Thermal diffusivity o: (1 Q-7 m2.s-1)
legend
      d = discontinuity thickness
       I = discontinuity depth
      rd = radius of discontinuity
    ll. = rubber
    0 = graphite epoxy
    0 =concrete
    .6. = titanium
     • =steel
     • =aluminum
  -=l=lmm
  - - = 1=2.5 mm
72 Infrared and Thermal Testing
cause a loss of resolution in the image of  thermal signature fade quickly. As a result,
the discontinuity. In practice, a thermal   if testing thin aluminum is necessary, a
nondestructive test procedure should be     snapshot thermographic camera with
optimized by evaluating the                 high frame rate must be used to monitor
signal-to-noise ratio. Therefore, in some   the high amplitude but brief signals.
materials, such as metals, high             Furthermore, unpainted aluminum is
temperature contrasts may be hidden by
high noise. Also, the best observation      characterized by high reflectance and low
time should be within the performance       emissivity; this results in a thermal image
capability of the infrared equipment used.
Thus, rear surface corrosion in thin        with considerable noise. Therefore, to
aluminum panels produces significant        obtain valid results when testing
temperature signals but the high thermal    unpainted aluminum, the radiation
diffusivity of aluminum makes the           properties of the aluminum must be
                                            enhanced and made uniform by applying
                                            a high emissivity coating, such as paint.
The influence of the type of material     0 '-! . -·· ...                   ,. ·;·'.;··  - \·;
on optimal detection parameters is
illustrated in Fig. 15. It is seen that the  TABLE 7. Thermal testing parameters in the detection of
greatest thermal contrasts appear in the
materials having higher thermal              material loss In steel. Steel properties: a= 7.3 x 10--6
conductivity (Fig. 15b) although the         m2.s-1; k::: 32 W·m-l·K·l. Conversions: 1 m = 39.4 in.;
maximum contrast occurs at relatively
short times (Fig. lSc). This tendency is     1 mm = 0.039 ln.; 1 K = 1 oc = 1.8 °F.
independent of discontinuity depth
although for deeper discontinuities lateral                                    Maximum Optimum
heat diffusion is more significant in
materials of higher thermal conductivity.     Specimen Material Defect         Running Observation
                                             Thickness L loss Radius fd        Contrast c:r,un Time lm·Cf:t"
Heating Power and Surface Heat                                                 (K = oq
Flux                                            (mm) (percent) (mm)                         (s)
 The surface temperature signal above a      Short pulse heating: th = 0.01 s; Tmax o. 8.4 K for Q = 106 W·m-2;
subsurface discontinuity is Jinearly         maximum specimen temperature at end of flash heating is
proportional to the absorbed thermal         weakly dependent on specimen thickness
energy, whereas the temperature contrast
is independent of it. Typically, more        1 90 5                            8.79 0.11
powerful heating provides a better image                                  10   8.89 0.15
of discontinuities because of a higher                                    20   8.89 0.15
signal·to.noise ratio SNR = 11T·11Tres                                         0.99 0.14
determined by the noise level of an                        50 5                1.00 0.19
infrared detector. However, the                                           10   1.00 0.19
signal·to·noise ratio that appears because                                20   0.11 0.16
of variations in surface optical properties                                    0.11 0.22
is a not straightforward function of                       10 5                0.11 0.23
heating power. It may increase because of                                 10
strong reflections or fluctuate because of                                20
some nonlinear phenomena. In general, it
is recommended that the heating should       5 90                          5   4.56         0.61
be sufficiently powerful to create a                                           4.78         1.23
maximum 1\T signal but not sufficient to                                   10  8.51         2.05
destroy the surface of the material.                                           0.58         1.06
                                                                           20  0.89         1.57
   The magnitude of the surface heat flux                                      0.99         2.51
impinging on the front and rear surface is   50 5                              0.055        1.49
described by using coefficients hF and hR.                                     0.097        1.97
Temperature contrasts remain at the same                                   10  0.11         2.96
level within a wide range of 11 for thin
materials that have a high or medium                                       20
thermal conductivity and that meet the
condition:                                   10 5
(64) Bi ~ IlL < 0.1                                                        10
                      k
                                                                           20
For a one.sided test of thick materials
with low thermal conductivity, increasing    10 90                         5   2.24         0.86
the fronl surface heat flux slightly                                           4.68         2.42
decreases temperature contrasts. In a                                      10  7.52         4.73
two·sided test, rear surface contrast is                                       0.24         3.14
practically independent of h.                                              20  0.58         4.26
                                                                               0.89         6.25
Examples of Optimal                          50 5                              0.017        4.70
Detection Parameters                                                           0.053        5.51
                                                                           10  0.093        7.82
As discussed above, one of the main
purposes of theoretical analysis is to                                     20
optimize a thermal nondestructive testing
process and to increase the probability of   10 5
discontinuity detection. During heating,
any discontinuity located at a particular                                  10
depth in a material of specified thermal
properties will produce a maximum value                                    20
of a chosen thermal parameter at a
                                             long pulse heating: th = 1 s
                                             10 90                         5   1.47          5.22
                                                                               4.12          5.97
                                                                           10  7.21          7.88
                                                                               0.19          6.61
                                                                           20  0.53          7.60
                                                                               0.86          9.38
                                             50 5                              0.014         7.88
                                                                               0.049         9.06
                                                                           10  0.090        10.90
                                                                           20
                                             10 5
                                                                           10
                                                                           20
                                             Thermal wave heating at quasi optimum frequency f (Hz) for
                                             particular material loss
                                             10 90 (I= 4) 5                    8.60          0.26
                                                                           10  8.78          0.26
                                                                           20  8.78          0.26
                                                                               0.19          7.34
                                                            50(1=0.14)5        0.50          7.59
                                                                           10  0.72         14.69
                                                                           20  0.013        10.43
                                                                               0.043        10.93
                                                             10(1=0.1) 5       0.073        11.26
                                                                           10
                                                                           20
74 Infrared and Thermal Testing
particular time. Three examples of            TABLE 8. Thermal testing parameters in detection of
thermal nondestructive testing procedures
are described in Tables 6 to 8. Many I        material loss in aluminum. Aluminum properties:
previously discussed thermal testing          a= 8.6 X lo~s m2·s-1; k = 210 w.m-1·K-1• Conversions:
scenarios are illustrated by data in these
tables.                                       1m~ 39.4 in.; 1 mm ~ 0.039 in.; 1 K ~ 1 •c ~ 1.8 "F.
    In a graphite epoxy composite, the                                       Maximum Optimum
lateral dimensions of discontinuities do
not affect ~Tand C values, at least in thin    Specimen Material   Defect    Running Observation
                                              Thickness L loss    Radius rd
samples (Table 6). The best observation                                      Contrast c~m Time tm(q~n)
time is primarily a function of                  (mm) (percent)     (mm)
discontinuity depth whereas the                                              (K = 0 C)  (s)
temperature contrast depends on both
discontinuity depth and thickness.            Short pulse heating: th = 0.01 s. .8.Tmal("' 5.57 K for Q = 106 W·m-1
   In the evaluation of back surface          1 90                5          8.27 0.02
material loss caused by corrosion in steel
or aluminum (Tables 7 and 8), neither the                         10 8.88 0.03
discontinuity depth nor discontinuity
thickness provide accurate physical                               20 8.91 0.04
representations of the material loss.
Hmvever, the basic test features remain       50 5 0.91 0.02
the same: discontinuity lateral dimensions
have the greatest influence on                                    10 0.99 0.04
temperature contrasts but not on
observation times. Also, the data in                              20 1.00 0.05
Table 7 illustrate the fact that each of the  10 5                           0.10 0.03
three common thermal wave techniques
(pulsed, step function and harmonic                               10 0.11 0.04
heating) provides similar detection
parameters that reflect the underlying                            20 0.11 0.06
basic physical principles.
                                              Short pulse heating: th = 0.01 s. 8.Tmax"" 4.S4 K for Q = 106 W·m-2
    Note that the data in Tables 7 and 8
                                              2 90                5          7.48       0.03
(short pulse heating) agree well with the                                    8.78       0.05
one-dimensional predictions on the                                10         8.91       0.07
relationship between material loss and                                       0.83       0.04
dimensionless contrast:7                                          20         0.98       0.06
                                                                             1.00       0.09
                                              50 5                           0.088      0.04
                                                                             0.108      0.07
                                                                  10         0.110      0.10
                                                                  20
                                              10 5
                                                                  10
                                                                  20
        t!.L    c
(65) L        1+ c
(66) c           M
                  L
              1-M
                     L
   It is important to note that Eqs. 65 and
66 are independent of material thermal
properties. For the three levels of material
loss (90 percent, SO percent and 10
percent) presented in Tables 7 and 8, the
corresponding running contrasts are
predicted to be 9, 1, and 0.11 respectively.
These values are close to those obtained
numerically.
                                                                             Heat Transfer 75
PART 4. Heat Diffusion in Periodic Regime
Thermal Waves                                where the analogy between Eq. 71 and
                                             the helmholtz wave equation has been
The term thermal 1vave commonly refers to
the typical ·wavelike temperature field      put into evidence by introducing the
induced by a harmonic heating process.       wave number~= -l(jw-D-1). Consequently,
These waves were used by Lord Kelvin and
A.]. Angstrom to investigate the thermal     the well knmvn theory of wave physics
diffusivity of bodies. Only since the early
1980s, have the waves been called            may be applied to the harmonic thermal
thermal, arousing debatc8·11 in the          field.
scientific community.
                                                As an example it is useful to study the
   The fundamental point to be               Green function, the solution of Eq. 71 for
understood is how the heat conduction
could show wave behavior. Heat               the unitary harmonic point source in the
conduction is well known in the classical    origin 0:1•3;12
case of a diffusive process, as shown by
the fourier equation:                        (72) G2(r,m)      exp[-~r]
                                                                  4nkr
                                                               exp[-{~]
                                                               4nkr
where k is the thermal conductivity of the   (73) G2 (r,t)     )jRe[G2 (r,w) exp(iwt
medium, D is the thermal diffusivity of                        !...)exp[-{] cos(wt-
the medium and w -is the heat supplied
per unit time per unit volume. The green                       4nkr      f
function solution of Eq. 67 shows clearly
                                             where r is the distance from the origin,
the diffusive behavior for a single heating
pulse at the origin 0 at time t = 0 where    and £ is the thermal diffusion length:
pc is the heat capacity per unit volume
and t "?. 0:3,12
                                 1           ~ ~(74) { =    2
(68) G1(x,)',z,t)
   A different situation is found in the        On first inspection, it may be noticed
case of harmonic heating. If the quantity    that the temperature field G2 in Eqs. 72
w is harmonic in time, with the period       and 73 behaves as a spherical wave of
2n·or1, the temperature field Tis forced to
                                             wavelength 2Jtt and for this reason has
be harmonic with the same periodicity. By    been called thermal wave. The main
                                             characteristic of such a wave is that it
introducing into Eq. 67 the complex
                                             decays strongly, moving away from the
quantities Wand T as found in Eqs. 69        heating source (see the exponential term
                                             in Eqs. 72 and 73). In practice the thermal
and 70, Eq. 71 provides the fourier          wave vanishes within some wavelengths,
equation in the harmonic regime: 1,3,12      so scientists disagree on the term wave.
                                             The damping is driven by the wave
                                 w           number ~ = (l+j)·>-1 . It is a complex
                                 k           quantity with the same real and
                                             imaginary part. As a consequence the
                                             thermal diffusion length I plays a double
                                             role: from one side it is proportional to
                                             the thermal wavelength 2rrf; from the
                                             other side it represents the extinction
                                             length at which the exponential term in
                                             Eqs. 72 and 73 is reduced to e-1 of its
                                             initial value.
                                                Once introduced, the spherical thermal
                                             wave G2 can be used to describe any
76 Infrared and Thermal Testing
harmonic field T that may result from an             the z direction, like a forward damped
                   appropriate superposition of spherical               wave. Note that, at anx time, the
                   thermal waves emitted by the heat source.            temperature oscillates ,\,ithin the two
                   Analogously the plane thermal wave can be            exponential envelopes ±exp {-z·£-1), that
                                                                        limit the thermal wave range. The
                   introduced when the heat source is                   difference between spherical and plane
                                                                        thermal waves is only geometric and does
                   planar. By harmonically heating the                  not affect the general properties just
                   medium along the plane x,y the                       shown; in particular any field T may
                                                                        always be also decomposed in terms of
                   temperature T may be represented by a                plane thermal waves propagating in various
                   plane thermal wave propagating in the z              directions.
                   direction: 1•3•12
                                                                        Thermal Wave Reflection and
                   (75) f(z,m)   Aexp(-Pi')                             Refraction
                                 Aexp(-~·z)
                                                                        Besides the thermal wave generation and
                                 Aexp[-(l-+fi-)z]                       propagation in a homogeneous medium,
                                                                        other basic phenomena should be
                   (76) T(z,t)   Re [f(z,m) exp(imt)j                   observed when at least two media are
                                                                        involved: the reflection and the
                                 A exr(-7)cos(rot-f)                    re(raction.1· 13•14 These phenomena are
                                                                        discussed below for plane thermal waves
                   where A is the amplitude at z == 0 and tis           in general.
                   the wave vector in the direction of
                   propagation. The plane thermal wave in                  When a plane thermal wave
                                                                        approaches the interface (the plane z = 0)
                   Eqs. 75 and 76 is plotted in Fig. 16 as a            between two media, it is partially reflected
                   function of the distance z, for different            back and refracted beyond. By imposing
                                                                        the continuity at the interface of both the
                   values of the time: in particular the curves
                                                                        field f and the vertical heat flux
                   a,b,c,d correspond to rot= 0, rr./3, rr./2, rr.. As
                   the time goes on, the maximum                        [-kaf.(az)-1] yields Eq. 77:
                   temperature rise is reduced but moves in
                                                                        (77)
FIGURE 16. In plane thermal wave, temperature rise is plotted           where z = 0 and where the fields in the
versus distance z (expressed in units t of thermal diffusion            first medium (1\) and in the second
                                                                        medium (f2) have been decomposed in
length) at different times: curves a, b, c, d respectively              terms of plane thermal waves:
correspond to rot= 0, n/3, rr./2, n. Two envelopes represent                   f,(x,z) = A e-P,[>in(o,)x + m>(O,)zj
maximum and minimum values of temperature rise.                         (78) + rAe~~{sln(o;)x-cos(o;)z]
              A                  Positive envelope +e-=J                       f 2(x,z) tA e-P,[>in(o,)x+ws(o,JzJ
Ec"                              Negative envelope ~e--f                where A, rA and tA are respectively the
                                                                        amplitudes of the incident, reflected and
 ~                                                    2   3
                                                                        refracted waves and 81 , 8{ and 82 are
·."5' AJ2              Distance z from the source                       respectively the angles between the
                                                                        directions of the waves and the normal to
i                  (unit l "' thermal diffusion length)                 the interface. By combining Eq. 77 with
.rJ 0                                                                   Eq. 78 the Snell relationships for the
                                                                        angles coefficients are obtained:
i'!
"'~ -A/2
 Q.
E
~
           -A
                0
Legend                                                                                         e,
  A "'amplitude                                                         k-(79)
  a=:curvetot=O                                                         {  r1::-   . (ee;)  =      sin(e 2 )
  b =curve tot= n/3                                                               sm 1      =
  c =curve tot= n/2                                                        vD, vD2
  d =curve (J)t""' rr.
  e =root of natural logarithms
                                                                                                   Heat Transfer 77
and the relationships for the reflection r          Thermal Wave Mirror
                 and the transmission t coefficients are
                 obtained:                                           If the effusivity mismatch between the
                                                                     media is very high, a relevant reflection
                 (80) r ~ e1cos(e!) - e2 cos(Bz)                     phenomenon takes place (see Fig. 17),
                                 e1cos(e,) + e2cos(Bz)               whether in phase for very small e2-e1- 1 or
                                                                     in opposition to phase for very large
                 (81) t        r+1                                   e2·ec1•15 In both cases the second medium
                                                                     behaves as a thermal wave mirror (r ~ ±1);
                 where e ~ k·V(IF1) is the thermal                   but the transmission (always t = r + 1) in
                                                                     one case tends to 2, in the other tends to
                 effusivity. Note that   in the case   of D2 > D  1  0. The interface between a generic gas and
                 Snell's law is valid o                :S arcsin     a generic solid provides an example.
                                         nly until e1                Because any solid is always at least 100
                 (v[D 1·D2- 1]) (see Eq. 79). For larger values      times more effusive than any gas, the
                                                                     effusivity mismatch is extremely high and
                 of 81 the thermal waves in the second               the solid consequently behaves as a
                                                                     thermal wave mirror for the gas and vice
                 medium become heterogeneous; the                    versa for reciprocity but with the
                                                                     follmving differences.
                 planes at constant phase differ by the
                                                                       1. If the incident wave is coming from
                 planes at constant amplitude.t4 Although                  the gas, the effusivity ratio is
                                                                          e2·ec1 >100, the reflection r ~ -1, and
                 such waves have been not yet observed,                   the transmission t ~ 0. In synthesis a
                                                                          destructive interference takes place
                 the theory predicts for them an                          between incident and reflected waves
                                                                          in the gas. At the interface the
                 anomalous attenuation larger than the                    temperature rise is kept to zero.
                 one described in Eqs. 75 and 76, as will be          2. If the incident ·wave is coming from
                                                                           the solid, then the effusivity ratio
                 discussed later. Equations 80 and 81 give                e2 -e1- 1 < O.OL the reflection r ~ 1 and
                 the coefficients rand t and consequently                 the transmission t----) 2. In synthesis a
                 the efficiency of the heat transfer through              constructive interference between
                                                                          incident and reflected waves takes
                 the two media. Generally the reflection r                place in the solid. At the interface
                                                                          there is the n1aximum temperature
                 and transrniss·!On t depend mainly on the                 rise.
                                                                     just to conclude this theoretical
                 effusivity mismatch. However, they                  discussion, it is of practical interest to
                                                                     quantify the appropriate depth for a
                 depend also on the diffusivity mismatch.            thermal wave mirror, which is the lower
                                                                     limit for the thickness of a material and
                 In fact the term cos(82) in Eqs. 80 and 81          beyond which the reflection properties are
                 is linked to diffusivity by Eq. 79; one             lost. An advanced study on layered
                                                                     structures has recently shown how the
                 exception is for normal incidence                   reflection properties of a slab change with
                                                                     its thickness. If R= is the reflection for an
                 (81 ~ 82 ~ 0) where rand t depend only on
                 effusivity. In Fig. 17 both coefficients are        R.·.'·"infinite thick slab, the effective reflection
                 plotted versus the effusivity ratio e2·ec1,         R for a finite slab differs by
                 just for normal incidence.
FIGURE 17. Thermal reflection and transmission coefficients
between medium 1 and medium 2 are plotted versus
thermal effusivity ratio e2·ec1 for normal incidence.
            2
"cU 0            0.1                                                   (82) R
 ~                                                                   where d is the thickness of the slab. Note
                                                                      once again the perfect analogy between
c                                                                     Eq. 82 and the equations for the
0                                                                     Fabry-Perot interferometer in wave
-~                                                                    physics. In thermal radiation, thermal
                                         10 100 wave interference occurs in the slab,
~ -I                                                                  which may either enhance (constructive
                                                                      interference) or inhibit (destructive
           0.01                                                       interference) the effective reflection R
                                                                      \\'ith respect to JL. To study the efficiency
legend                                                                of such a thermaJ·wave mirror, R·R,;;1 is
 e = thermal effusivity
 r = reflection coefficient
 t = transmission coefficient
78 Infrared and Thermal Testing
plotted in Fig. 18 as a function of the           distance (L = 1 mm [0.04 in.]) until it
                            normalized thickness defined as
                            rf.1· 1 -IRJ·(I-R~)-1, for different values of    approaches the aluminum foil, which acts
                            IR~I = 0.2, 0.5, 0.9, 0.99 and 0.999. Note        as a thermal wave mirrbr. The foil forms
                            that all curves merge together in two             an oblique angle 8 with the x axes. As a
                            opposite limits.                                  consequence the reflected wave forms an
                             1. For a high normalized thickness the           angle 28 with the incident one, and the
                                slab is thermally thick and behaves as        temperature rise in air where the waves
                                an infinite medium, so R ~ R'"' (mirror       are superimposed is calculated:l4
                                regime).
                                                                              (83) f,1,(x,z)   A exp(-p,1,z)
                             2. For a low normalized thickness the
                                slab is thermally too thin, the incident                      + RA exp{~air
                                thermal wave is transmitted beyond
                                the slab, without a relevant reflection                       x [cos(2e)z-sin(ze)x]}
                                R ~ 0 (transparent window).
                                                                              where A is the amplitude of the incident
                            Note also that the transition from                \\'ave in the origin 0, and R is given by
                            transparent to mirror regime occurs when          Eqs. 80 and 82. The detection of the
                            the normalized thickness is practically           thermal field in Eq. 83 may be achieved
                            unitary, which for a high effusivity              by using the nzirage technique.Is-2z A
                            mismatch (IRJ ~ I) corresponds to an              probe beam is sent along the )'axis, in air,
                            extremely small thickness d ~ 1·(1 - R~).
                            For this reason, even extremely small             close to the origin 0. The thermal
                            discontinuities, cracks and delaminations
                            behave as thermal wave mirrors and may            gradients encountered along the path
                            be easily detected by thermal wave                produce the beam deflection measured by
                            interferometry. 16 , l 7
                                                                              a remote position sensor. In particular the
                            Evidence of Thermal Wave                          beam orientation is chosen to detect the
                            Reflection
                                                                              FIGURE 19. Scheme of thermal wave cavity for detection of
                            A simple experiment to prove the                  reflected thermal wave.
                            reflection of thermal waves is shown in
                            Fig. 19. A plane thermal ·wave is generated
                            in air by heating periodically a thin
                            absorbing layer with a wide pumped laser
                            beam (S = 4 mm [0.16 in.]). The plane
                            wave propagates in air along z for a short
,_.,f;;:-,~·-;_j'-'i-:<,._               ;·~':'?-'•  ;-,·.;·.~ -~;.·•::~c·,·
FIGURE 18. Normalized thermal reflection IR·R.-11 of slab is
plotted versus normalized thickness for different values of
interface reflection coefficient R. (0.2, 0.5, 0.9, 0.99, 0.999).
Normalized thickness is defined as J.[-1.1R~I·(l-R.2)-1 where I
is discontinuity thickness and l is thermal diffusion length of
slab.
   1.1
"·"' 1.0                                                  0.999                                              Incident wave
                                                     Mirror regime
~ 0.9
~
~ 0.8
                            Transparent                                                                      fr
"·"a. 0.7                     window                                                                           Pump beam
                                                                                                                      5
E
~
   0.6                                                                        Absorbing layer
c
·~ 0.5
"" 0.4
 ~
.," 0.3
.~ 0.2
E 0.1
z0 0                                                                          legend
        O.ol                       0.1 10                           100             L "" cavity length
                                                                                   0 =mirror center
                            IR..INormalized thickness = !__.                        5 = beam width
                                                                                x,y,z =axes
                                              { 1- R.!                              e =- angle of incidence
                                                                                                                            Heat Transfer 79
components of the deflection angle along              where the tilt angle 9 between the mirror
the x and z directions:l4,15                          and the absorbing layer has been set at
                                                      9 = 11 degrees. Both components ibx and
(84) <l>x     J_!__<liz_ aT,,, dy                     iPz have been measured and their ratio .Q
                                                      has been compared ·with Eq. 89. The
              11 dT ilx                               results have been repeated by changing
                                                      the modulation frequency {(for example,
                            )'                        fair)· In Fig. 20 the amplitude ratio IQ1 is
                                                      plotted as a function of if; the symbols
              euax-111 dn L dTair                     represent the experimental data and the
                 dT                                   full Jines are the theoretical values by
                                                      Eq. 89 for the tilt angles e reported on the
(85) <1>,     f_!_<liz_ ilT,;, dy                     right scale (0, 5, 10, 11 and 15 degrees).
                                                      Note that, although there is some noise,
              II dT ilz                               all experimental points lie near the Hne at
                                                      9 = 11 degrees at all frequencies. This
                          )'                          agreement demonstrates the following
                                                      points.
              ;1;        ddnf    L      -d-T-a11;;-,
                                                        1. The reflection phenomenon takes
                                   eff                     place as described by Eq. 83.
where 11 is the refractive index of air,                2. The aluminum foil acts as a thermal
dll·(dT)-1 is the optothermal coefficient of               wave mirror for a wide spectrum lair of
air and Lerr is the effective length useful                the thermal wavelength.
for the beam deflection. Leer depends on
the lateral dimensions S of the incident              Evidence of Thermal Wave
thermal wave. By combining Eqs. 83 to                 Refraction
and 85 Eqs. 86 and 87 are obtained:
                                                      Besides the proof of the thermal wave
(86) <i>x c{Rsin(ze)                                  reflection, an experiment may be set up to
                                                      prove the refraction of the thermal waves
             X ,.,,!cm(2o),- >in(zo)xJ}               as shown in Fig. 21. The pumped laser
                                                      beam propagates along y, is modulated in
              _ c[Rsin(ze)]                           time and is focused onto a solid sample
                                                      (medium 1) by means of a cylindrical
(87) <1>, c[e-~•'•'-Rcos(ze)                          lens. The dimensions of the ellipsoidal
             X e~••lw>(20)Hin(2o)xJ]                  pumped beam spot in the plane x,z (see
              _ c[1- Rcos(ze)J
where e is the root of natural logarithms             FIGURE 20. Amplitude ratio .Q between deflection
                                                      components along x and z is plotted versus modulation
and C is a constant defined as:                       frequency square root: open squares (D) represent
                                                      experimental data; full lines are calculated by Eq. 89 for
CI(dll)(88)~                     L,rrP, 1, A          different values of tilt angle e.
              -- -
                II dT                                            Air thermal diffusion length, mm (1 Q-3 in.)
Note that the simplifications in the                             2.0 1.5 1.0 0.8 0.6 0.5      0.4
exponential terms in Eqs. 86 and 87 are                          (80)(60)(40) (32) (24) (20)  (16)
allowed if the probe beam travels close to
the origin 0, at a distance shorter                         0.3                                                ..15
than lair- In this case the two components                                 0                        •~
are proportional to each other and their              .,.0 0.2                                                                "~'
ratio Q depends only on the angle 8 and                          .:.le~ .r!rr>crJ • rJ•rJrlrJ ••               "'• 11
on the reflection coefficient of the mirror            ~                                                             10 ~
R as follows:                                                                   • • ••
                                                       ~
                                                                                                                              0,
                                                      '0                                                                      •c
                   <~>x                                                                                                       cv~
(89) Q ~                           Rsin(ze)           il                                                          s5           ~
                                 1 - R cos(ze)        'a_
                <1>,                                                                                                          :v2
                                                      «E 0.1
               -sin(ze) ~ -tg(e)                      0 ,, ,,                                                     0
                                                           0
              1 + cos(ze)
Jn Eq. 89 it is assumed that R"' -1 because                      Frequency square root .Y(Hz)
the aluminum foil acts as a thermal wave
mirror for any e (see Eqs. 80, 81 and 82).            legend
An experiment has been performed in the                  0 =experimental datum
thermal wave cavity described in Jlig. 19,
                                                       - =calcula_tion with Eq. 89 for degree of incidence on scale to right
80 Infrared and Thermal Testing
Fig. 19) are adjusted to provide a line                         '!fwave 2 forms an angle (82 - 81) with the
heating source along x. The generated
                                                                z axis (see Fig. 21 and Eq. 90). The
thermal wave propagates along z inside 1
                                                                detection of the thermal field in Eq. 90 is
the solid until it reaches the interface                        accomplished by the mirage technique, as
with the second medium (air), which is at
an oblique angle to the line source (the                        in the previous experiment. The probe
normal to the interface 11 forms the angle                      beam is placed in the second medium
81 with the z axis). The theory predicts
that the incident wave is reflected back                        near the interface, where the heat flux is
and refracted in the secorid medium                             greater. The two components of the
according to Eqs. 78 and 79. Equation 90                        deflection angle along x and z are given
is written for the thermal field refracted in                   by combining Eqs. 84, 85 and 90:14
the second medium: t4
                                                                (91) <i>_,  c(tsin{Bz-91}
(90) f 2(x,z) tA cxp(-Pzi')
                                                                            x cxp{-P2 [sin(B,-B1)x
                 lA cxP{-P2 [sin (B2 -B1)x
                                                                            )z]})+ cos(e2-B1
          Jzl}+ cos(e,-e1
                                                                (92) ci\,   c(tcos{B2 -e1}
where A is the amplitude of the incident
                                                                            x exp{-P2 [sin(B2 -B1)x
J:wave in the origin 0, t is the transmission
                                                                            )z]})+ cos(B2-B1
coefficient given in Eq. 81, is the
                                                                Note that in this case the two
position \Vith respect to 0, p2 is the wave
                                                                components are proportional to each
vector that points the direction of the                         other everywhere, regardless of the values
refracted therm_!!l wave, 82 is the refraction                  of x and z, and their ratio depends only
angle between ~2 and the normaln to the                         on the incident and refracted angles
interface. In Eq. 90 the quantity 72 is
expressed as a function of reference axes x
and z. Note that in such a reference
system the direction of the refracted
FIGURE 21. Schematic setup for detection of refracted thermal   The analogy between Eqs. 89 and Eq. 93
wave.
                                                                suggests that the amplitude ratio Q is
                            Medium 2
                                                                useful not only to detect the thermal
                                   Refracted wave               wave but also to reveal its direction. The
                                                     ~Rotation  physical reason is in the basic principle of
                                                                the mirage technique: the deflection angle
                                                      .J stage
                                                                is not related directly to the temperature
                                      Incident wave
                                                                field t· but rather to its gradient Vf, which
                                    Pump beam spot
                                             5                  points the direction of the heat flux.
                                                                Therefore If the direction of a thermal
                          Medium 1
                                                                wave is of interest, the knowledge of the
legend
      n = incidence direction normal to interface               scalar function T could be useless, even if
       S "' beam width
                                                                it is known for '!_ny plane, whereas the
  x,y,z =axes
      j3 = wave vector                                          vectorial field VT immediately gives such
     l:l1 = angle of incidence                                  information, point by point. Therefore
     01 = angle of refraction
                                                                the mirage represents the most appropriate
                                                                technique for this purpose. In particular
                                                                the ratio between the components filx and
                                                                ibz gives, as previously shown, the local
                                                                information on the angle between ~ and
                                                                the z axis. Coming back to the differences
                                                                between Eqs. 89 and 93, an important
                                                                point is the sign of Q. In the reflection
                                                                experiment it is always negative whereas
                                                                in the refraction experiment it depends
                                                                on the sine of €12 and 81, which is fixed by
                                                                Snell's law (see Eq. 79). As a result if
                                                                D2 > 1J1, then 82 > q 1 and consequently
                                                                Q > 0; whereas if D2 < D1 then Q < 0. The
                                                                sine of Q helps to discriminate one case
                                                                from the other. Knowledge of Q permits
                                                                calculation of the refraction angle 82 by
                                                                solving Eq. 93 as follows:
                                                                                                     Heat Transfer 81
The results obtained with the setup                                                                                 phosphorus) more diffusive than air. Note
                                                                                                                        that in the first case the linear behavior is
    shown in Fig. 21 provide an example. An                                                                             broken around 91 "" 30 degrees, which
                                                                                                                        corresponds to the limit angle:
    argon laser illuminates a thin solid sample
    (medium 1) not far from one edge. The                                                                               (95) elim = arcsin~ Dt
                                                                                                                                                        Dz
    pumped beam is modulated in time and is
                                                                                                                        Beyond the limit angle Eqs. 79 and 94
    focused by a cylindrical lens only in the z                                                                         become useless and consequently the
    direction and so has a line source 6 mm                                                                             experimental points measured for e1 > 30
    (0.25 in.) wide along x and a few
                                                                                                                        degrees, for the stainless steel sample only,
    micrometers (-1.4 x to-4 in.) wide along                                                                            are meaningless. By using the least
                                                                                                                        squares technique to calculate the slopes
    z. A helium neon laser probe coHinear                                                                               (see full lines in Fig. 23) the diffusivity
    with the pump argon laser is placed in air                                                                          ratios are obtained: D111r·Dair-1 = 2.2 ± 0.1
                                                                                                                        and DFeNI'Dair-1 = 0.25 ± 0.02, where FeNi
    (medium 2) close to the sample edge. A                                                                              refers to the stainless steel and InP refers
    rotation stage allows the movement of the                                                                           to the indium phosphorus material. These
                                                                                                                        diffusivity ratios lead to the values D 111r =
    sample in the x,z plane to change the                                                                               0.44 cm2 ·s-1 and DFeNi = 0.05 cmZ.s-1,
                                                                                                                        when it is assumed that Dalr = 0.2 cm2·s-l,
    orientation of the edge. The pump and                                                                               in perfect agreement with the values
    the probe are fixed. As a consequence the                                                                           given in the literature.23 Has a ne\\'
                                                                                                                        methodology been introduced here to
    incidence angle el is changed by the                                                                                determine the thermal diffusivity of
                                                                                                                        materials? The question is reasonable but
    rotation ~tage. 1!_1. Fig. 22 the amplitudes                                                                        difficult to answer. It is useful to
    of both <Px and cfl2 are shown versus 91 for                                                                        remember three points.
    a semiconductor wafer of indium
    phosphorus.                                                                                                          1. This technique based on Snell's Jaw
                                                                                                                            makes it possible to determine the
        From these experimental data the                                                                                    thermal diffusivity of one medium if
                                                                                                                            that of the other one is well known.
    ratio Q is calculated and eventua1ly, by
    using Eq. 94, the refraction angle e2 is                                                                             2. This technique works whether the
    calculated. Finally plotting the quantity                                                                               heat source is in the first or second
    sin (82) versus sin (81) provides the most                                                                              medium. Therefore it is suitable even
    appealing proof of Snell's law; in fact a
    straight line is expected, with the slope
    equal to Y(D2·D1-') (see Eq. 79). In Fig. 23
    this procedure is reported for two
    materials: a stainless steel (36 percent
    nickel) less diffusive than air and a
    semiconductive material (indium
FIGURE 22. Amplitude of deflection signal (~V) is plotted                                                               FIGURE 23. Sin 92 is plotted versus sin e1 for two different
versus incidence angle (degree). Sample is indium                                                                       materials: nickel iron alloy sample (D) and indium
phosphorus wafer. Pump power is 800 mW; modulation                                                                      phosphorus semiconductor wafer (+). Full lines are derived
                                                                                                                        by least squares method.
frequency I= 16 Hz. Plotted squares (0) are data measured
for component along x; plotted plus signs(+) are data for                                                                                                Incidence angle (degrees)
component along z.
                                                                                                                                                                                                       DD
90 ,-~~~~~~~~~~~~~~-.
                                                                                                                                  0.5
80
70                               •• •
60                                        +++ +
50
40
30
                                                                                                                     D
20
10  ~·~·~·~·~·-·-·_·_·_i·_·____~----~----~----~                                                                                  0
 0
                                                                                                                                     0 0.2 0.4 0.6
    0 10 20 30 40 50 60                                                                                                                                 Incidence angle (sin 91)   0.8
                             Incidence angle (degree)                                                                   legend
                                                                                                                          e 1 =angle of incidence
legend                                                                                                                    e2 "' refracted angle
  + "' datum in z direction                                                                                              0 =datum from staintes5 steel (36 percent nickel) sample
  0 =datum in x direction                                                                                                 + =datum from indium phosphorus semiconduclor
82 Infrared and Thermal Testing
for measuring the diffusivity of                                 FIGURES 24. Numerical simulations for amplitude of thermal
    nonabsorbing materials.
 3. Preliminary results indicate that this                           field. Contour plots of amplitude of 'fare calculated as
    technique guarantees the same
    accuracy as the other well known                                 function of coordinates p, Swhen stainless steel sample with
    techniques. 24 ·27                                               thermal diffusivity D1 = 5 mm2·s~1 is heated by source at
Heterogeneous Thermal Wave                                           oblique incidence 81: (a) homogeneous thermal waves at
                                                                     p   <p4<f241 t8,,1      20  degrees; (b) heterogeneous thermal waves
Snell's law for thermal waves establishes                            at                  =       70 degrees; (c) heterogeneous thermal
that 1-(YD-1) has the same role as the                                                       =
refractive index has for electromagnetic                                                 e,
waves. This analogy suggests the simple
question of whether the same analogy                                 waves at p > 6(2, 81 = 70 degrees, amplitude increasing
pertains to total reflection for thermal                             with height 1;.
waves. The answer is in the refracted
thermal field when the incident angle is                             (a}
9t > 8nm·· The refracted thermal wave is
still planar but no longer
homogeneous: 13•14
where B is a constant, p is the variable                             (b) \                                             Heating tine
parallel to the interface and I; is the
variable vertical to the interface. In Eq. 96                                                            Heating line
two different solutions are included                                 (c)
depending on the sign ± in the
                                                                               -2f1
exponential term. Both of them satisfy                                 legend
the wave equation Eq. 77 but the
                                                                          r2 "' air thermal diffusion length (mm)
smoeluantiionnglfe2ssmianytsheeemspatcoebSe>ph0y, sbieccaalluyse it        p =variable parallel to interface
is amplified in the direction C-                                           ~=variable vertical to interface
   To clarify this point it is helpful to
study the amplitude of the field T2 for a
simple system made of air and a low
diffusivity material (stainless steel with
36 percent nickel). The air is in the half
space for S> 0 whereas the stainless steel
is for C< 0. The heating source is an
oblique line inside the material; the line
begins from the origin 0 and forms an
angle e, with I;= 0.
   The amplitude of the field T(p,l;),
calculated by a numerical simulation, is
reported in the contour plots of Fig. 24 for
two different values of the angle 81 chosen
to induce homogeneous (Fig. 24a) or
heterogeneous (Figs. 24b and 24c)
refracted thermal waves. In particular
in Fig. 24a the incidence angle is
8t = 20 degrees< ellm = 30 degrees.
Note that plane thermal waves depart
from the heating line inside the material,
as it is pointed out by the arrows. The
wave propagating toward the air-to-steel
interface is partially reflected back and
also refracted in air. Consequently a
thermal interference occurs just below the
surface, where the incident and reflected
wave are superimposed, as is revealed by
the strong distortion of the wave front. In
air the refracted plane thermal wave
                                                                                                                       Heat Transfer 83
changes direction according to Snell's law
                  (82 = 43°), as pointed out by the arrows in
                  Fig. 24a. Of course, because of the finite
                  dimensions of the heating source, the
                  refracted field is too distant to be a planar
                  wave close to the origin 0.
                     These boundary effects vanish ·within a
                  thermal diffusimllength {'2,28 In other
                  words the refracted wave becomes plane
                  at a suitable distance p > 12• A different
                  case is reported in fi.gs. 24b and 24c
                  where 81 = 70 degrees > Sum = 30 degrees.
                  As the incident wave approaches the
                  air-to-steel interface it is reflected back
                  giving rise to the usual interference
                  phenomenon. In air the thermal field is
                  now too far to be a plane wave (see
                  Fig. 24b). However it is still possible to
                  recognize a restricted region, close to the
                  interface, lNhere the amplitude tends to
                  maintain a plane wave front (see the
                  arrows in Fig. 24c)..: In such a region the
                  un~table solution T2 + takes place instead
                  ofT2 -. As a consequence, the amplitude
                 increases greatly with the height l;. The
                  physical reason is that in this zone the
                  main heat flux comes from the higher air
                  layers rather than from the inside material
                  (see the arrows ill Fig. 24c). Unfortunately
                  this heterogeneous wave could be
                  observed far from the origin (p > 6£2)
                  where the wave is too weak to he
                  detected.
84 Infrared and Thermal Testing
leferences
 1. Almond, D.P. and P.M. Patel.              11. Rosencwaig, A. and A. Gersho.              'i
    PIJototllermal Scifnce mid Techniques.         "Thcrmai.\>Vave Imaging." Scie/lce.
     Dordrccht, Netherlands: Kluwer
     Academic Publishers (formerly                 Vol. 218. \"/ashington, DC: American
                                                   Association for the Advancement of
     London, United Kingdom: Chapman
     & Hall) (]996).                               Science (]982): p 223-228.
                                              12. Mandelis, A. "Green's Function in
 2. Maclachlan Spicc_r, J.\\1• Active
                                                   Thermal \'Vave Physics: Cartesian
     71Iermogmpl1}' fUr Mmmfacturins and           Coordinate Representations." joumol
     Process Control. SPIE short course           of Applied l'll)'sics. Vol. 78, No. 2.
    notes [Orlando, FL]. Bellingham, WA:           College Park, MD: American Institute
                                                   of Physics (1995): p 647-655.
     International Society for Optical        13. U Voti, R. La Tewlca di Deflessiunc
     Engineering (April 1995): p 3-4.             Fototermica: Principia ed Applicazioni.
                                                   Ph.D. thesis. Rome, Italy: UnivcrsWi
 3. Carslaw, B.S. and j.C. jaeger.                 degli Studi di Roma 11 La Sapienza."
    Conduction of Heat in Solids. London,          (1996).
     United Kingdom: Oxford University
     Press, Clarendon Press (1959).           14. Bertolotti, M, G.L. Liakhou, R. LiVoti,
                                                   S. Paoloni and C. Sibilia. 11Therma1
 4. Maclachlan Spicer, j.\·V. Active
                                                   ·wave Reflection and Refraction:
     Thermography fOr Mmmfacturing a11d            Theoretical and Experimental
    Process Colllrul. SPIE short course           Evidence." }oumal ofApplied Pll)•sics.
    notes [Orlando, FLJ. Bellingham, WA:           Vol. 85, No.7. College Park, MD:
     International Society for Optical             American Institute of Physics (April
     Engineering (April 1995): p 4-7.              1999): p 3540-3545.
 5. Holmon, J.P. Experimental Methods for     15. Bertolotti, M., R. LiVoti, C. Sibilia
    Engineers. New York, NY: McGraw-Hill           and G .L. Liakhou. 11 Analysis of
     Book Company (1984): p 1-60.
                                                   Defects in Multilayers through
 6. Vavilov, V. "Infrared Techniques for           Photothermal Deflection Technique."
                                                  Specification, Production, and Tcstins o(
     Materials Analysis and Nondestructive        Optical Components and Systems
     Testing." Jufrarcd Met!Iodology and           [Glasgow, United Kingdom]. SPIE
                                                   Vol. 2775. Bellingham, WA:
     Tec/Jnology. Langhorne, PA: Gordon
     and llreach Science Publishers (1994):        International Society for Optical
    p 230-309.                                     Engineering (1996): p 370-379.
 7. Vavilov, V., E. Grinzato, P.G. Bison,     16. Patel, P.lvl., D.P. Almond and
                                                   H. Reiter. 11Thermal Wave Detection
     S. lvfarinetti and M. Bales. "Inversion       and Characterisation of Sub-Surface
     for Hidden Corrosion                         Defects." Applied Physics B. Vol. 43.
                                                   Berlin, Germany: Springer Verlag
     Characterization: Theory and                  (1987): p 9-15.
    Applications.'' lntematimwlfoumal o(
    Jleat m1d Mass Tmns(i'J: Vol. 39.         17. Thomas, R.L., J,j. Pouch, \•V.H. VVong,
     Oxford, United Kingdom: Elsevier
     Science, Pergamon Press (1996):               L.D. Favro, P.K. Kuo and
     p 355-371.                                    A. Rosencwaig. "Subsurface Haw
 8. Lord Kelvin. 11The Reduction of                Detection in Metals by Photoacoustic
                                                  Jvficroscopy." }uumal of Applied
     Observations of Underground                  Physics. Vol. 51, No.2. College Park,
    Temperature." Tramactions of' tile Royal       MD: American Institute of Physics
    Society o(EdinlJIIISII. Vol. 22.               (1980): p 1152-1156.
     Edinburgh, United Kingdom: Royal         lB. jackson, W.B., N.M. Amer,
     ~ociety of l'dinhurgh (1861): p 405.
 9. AngstrOm, A.J. "New Method of                  A.C. Boccara and D. Fournier.
     Determining the Thermal                       "Photothermal Deflection
    Conductivity of Bodies." Pllilosupllical       Spectroscopy and Detection." Applied
     jVJagaziue. Vol. 25. London, United          Optics. Vol. 20, No. 8. \•Vashington,
                                                   DC: Optical Society of America
     Kingdom: Taylor & Francis Limited             (1981): p 1333-1344.
    (186:l): p 130.
10. Bennett, C.A., Jr. and R.R. Patty.
     "Thermal \i\/ave Interferometry: A
     Potential Application of the
    Photoacoustic Effect." Applied Optics.
     Vol. 21. ·washington, DC: Optical
     Society of America (1982): p 49.
                                              Heat Transfer 85