Active – ThermographyPrinciples
Pulse Thermography
Steap Heating Thermography
Lock-In Thermography
Pulse-Phase Thermography
Derivative technics
Active – Thermography
Principles
Pulse Thermography
Steap Heating Thermography
Lock-In Thermography
Pulse-Phase Thermography
Derivative technics
Reflexion
object
IR Sensor
ThermalanalysisThermalanalysis
voidMetallic part
voidvoidMetallic partMetallic part
Active – Thermography
Principles
Pulse Thermography
Steap Heating Thermography
Lock-In Thermography
Pulse-Phase Thermography
Derivative technics
Basically, PT consists of briefly heatingthe specimen and then recordingits temperature decay curve.
16,5
16,7
16,9
17,1
17,3
17,5
17,7
0 5 10 15 20 25 30 35 40N° échantillon
Poids (g)
avec colle
sans colle
fausses alarmes
Piece(plastic lid)
IR camera
Infrared cameras - ElectroPhysics3-14µm - 320*240 pixelsPyroelectric BST Uncooled FPAobjective : 50 mm - 3-5µmdistance lid - camera : 50 cm
Halogen Lamp
halogene Lamp1000 WDistance Lid - Lampe : 12cm
Acquisition system
Air
Glue
Thermal front
20
40
60
80
100
120
Niveau de gris
2 s ; zone avec colle
2 s ; zone sans colle
3 s ; zone avec colle
3 s ; zone sans colle
4 s ; zone avec colle
4 s ; zone sans colle
5 s ; zone avec colle
5 s ; zone sans colle
6 s ; zone avec colle
6 s ; zone sans colle
0
2
4
6
8
10
12
14
16
18
20
0 10000 20000 30000 40000 50000 60000 70000 80000 90000Temps d'attente en ms
Contraste de différence
Temps de chauffe = 2 secondes
Temps de chauffe = 3 secondes
Temps de chauffe = 4 secondes
Temps de chauffe = 5 secondes
Temps de chauffe = 6 secondes
9
( )( )
{ }
( )
{ }
,,, ,
: ; :
y Sx A
Ng y tNg x tC A S t
Card x A Card y S
A zone avec colle B zone avec colle
∈∈= −∈ ∈
∑∑
0
2
4
6
8
10
12
14
16
0 10000 20000 30000 40000 50000 60000 70000 80000 90000Temps d'attente en ms
Contraste de différence
Jour 1 17:20:00
Jour 1 19:30:00
Jour 2 09:25:00
Jour 2 17:39:00
Heating time = 5s
� Differences between the mean contrastand the best contrast for each trial lessthan 3,8%� Differences for the best contrast for various « waiting time » beforeacquisition: 500ms
10
Thermal imagesWithout glue
With glue spots
���� Thermal prints can identify presence of glue
70
90
110
130
150
170
190
210
230
13 33 53 73 93 113 133 153 173 193Position
Niveau de gris
12
Raw (unprocessed) thermal image
40
60
80
100
120
140
160
180
200
7 27 47 67 87 107 127 147 167 187Position
Niveau de gris
Mean image subtracted
Unprocessed thermal images
Examples
Processed images
Without glueWith glue
14
[ ]répartition suivant les valeurs 0,255
de proportion i
i
h
p
∈
Original image histogram Thresheld image histogram M=4
[ ]répartition selon les niveaux de gris ,
1, ,de proportion
j
j
z
j M q∈
Multithresolding algorithm (Tsai)
mask m×n
Pixel of interest{ } { }voisins 0,5. voisins
Card jaunes Card
site conservé
>⇒
X. Maldague
Active – ThermographyPrinciples
Pulse Thermography
Steap Heating Thermography
Lock-In Thermography
Pulse-Phase Thermography
Derivative technics
The sample is continuously heated, at low power. Variations of surface temperature with time are related to specimen features as in pulse thermography. This technique of step heating (SH) is sometimes referred to as time-resolved infrared radiometry or TRIR.
Steap Heating Thermography
Active – ThermographyPrinciples
Pulse Thermography
Steap Heating Thermography
Lock-In Thermography
Pulse-Phase Thermography
Derivative technics
Lock-in principle: The principle oflock-in thermographyconsists of introducing periodicallymodulated heat into an object and
monitoring only the periodic surface temperature modulation phase-referred to the modulated heat supply. Hence, if thesurface temperature is measured via an
infrared (IR) thermocamera, lock-inthermography means that the information ofeach pixel of the image is processed as if it
were fed into a Lock in amplifier.
Background theory:T
TTaT
c ∂∂=∆=∆
ρλ
Solution (of the form):( ) ( ) ( )( ) fwithetT
tTzTtzTti ..2
.,.. πωω ==
=
Excitation: tωsin.0Ε=Ε
For one dimension: t
tzT
z
tzTa
∂∂=
∂∂ ),(),(
2
2
)(.),(
2
2
zTa
iz
tzT ω=∂
∂
ikzeTzT 0)( =Which solution is: k?
)(..)(2 zTa
izTkω=−
aikω−=2
2)1(2 −=− iitricka
ik2
).1(ω−=
)2
cos(.),(
...),(
2.
0
2.
2.
0
za
teTtzT
or
eeeTtzT
az
tiaiz
az
ωωω
ωωω
−=
=
−
−−
Thermal diffusion length (initial amplitude divided b y « e »)ω
µ a2=
Thermal wave propagation speed
Thermal wave wavelength
First method: Synchronize the camera (not easy/linkto the camera frame rate) with the excitation source and acquired four points (every T/4) per period andper pixel.
( ) ( )224
213 SSSSA −+−=
24
13
SS
SSarctg
−−=ϕ
Second method (most used now):acquire the signal and process it in the Fourier domain
Benefits of Lockin Thermography
1. Non contactOn large structureOn small objects down to 1mm
2. Full fieldComplex structures
3. Easy to useReal componentsSine loadRandom loading
4. Fast to set upblack paint on metal. No paint on composites
5. Wide spectral rangeup to 20kHz
6. Energy measurementPlastic and dissipated energy
NDT Lockin Thermography• SET UP
LampsAmplifier
Function generator
PC
Camera
Trig out (rear)
Analogue input (rear)
Output (front panel)
Lockin (rear)
USB or camlink
USB or Camlink
specimen
Example : Stress analysis fracture mechanics
Load Frequency 10 Hz.
The lines in data graph show time history of the image’s in the B&W image.
Note that camera is calibrated in absolute temperature, and shows small amplitude temperature swings of 100 mK and 45mK.
Thermal image movie (Colors)
Thermal image (B&W)
xx
Map of amplitude in °CThermal image
Frame rate : 140Hz Full-frame
Frame accumulation: 1000 frames
Processing Time : 7 seconds
Lockin Frequency : 10 Hz
Map of phase in degrees
Map of amplitude in MPaLocal value of stress region 1 (MPa)
Library of material (Coefficient of thermoelasticity)
Coefficient of thermo-elasticity allows to transform temperature change in MPa
Km = α/(ρ.Cp)
α = linear expansion (K-1)
ρ = density (Kg/m3)
Cp= specific heat J/kg/K
Frame rate : 140Hz Full-frame
Frame accumulation: 1000 frames
Processing Time : 7 seconds
Acquisition : every 30s
Lockin Frequency : 20 Hz
Thermal profile T=30sT=1min00sT=1min30sT=2min00sT=2min30sT=3min00sT=3min30sT=4min00sT=4min30sT=5min00sT=5min00sT=5min30sT=6min00sT=6min30sT=7min00sT=7min30sT=8min00sT=8min30sT=9min00sT=9min30sT=10min00sT=10min30sT=11min00sT=11min30sT=12min00sT=12min30sT=13min00s
Crack propagation
Lockin Thermography test
Click on the film to see film:• Frequency 0.1Hz• Recording: 1 period
Results - LIT
Average image during the test
Phase image
Magnitude image
Active – ThermographyPrinciples
Pulse Thermography
Steap Heating Thermography
Lock-In Thermography
Pulse-Phase Thermography
Derivative technics
NDT Phase Pulse Thermography• SET UP
LampsAmplifier
Function generator
PC
Camera
Trig out (rear)
Analogue input (rear)
Output (front panel)
Recorder trigger in on LPT
USB or camlink
USB or Camlink
specimen
Phase Pulse Thermography test
Click on the film to see film:• Frequency 0.1Hz• Recording: 1 period
Result - PPT
Steel plate thickness
Phase imageMagnitude image
Frequency : 0.020 Frequency : 0.023 Frequency : 0.025 Frequency : 0.028 Frequency : 0.031 Frequency : 0.033 Frequency : 0.036 Frequency : 0.039 Frequency : 0.041 Frequency : 0.044 Frequency : 0.047 Frequency : 0.049 Frequency : 0.052 Frequency : 0.055 Frequency : 0.057 Frequency : 0.060 Frequency : 0.063 Frequency : 0.065 Frequency : 0.068 Frequency : 0.071Frequency : 0.073 Frequency : 0.076 Frequency : 0.079 Frequency : 0.081 Frequency : 0.084 Frequency : 0.087
Result analysis - PPT
Phase image frequency=0.03Hz
Shunt detection in Solar cells
Solar cell during the test
Standard thermal image shows the test facilities for shunt detection. The set up of the complete system lasts 15 minutes.The test consists of applying a voltage to the solar cells. In case of shunt the current goes directly from the grid line to the base and heats up locally the solar cell. Lock in Thermography allows detecting the heating up by detecting of the periodic heating.
The equivalent circuit of local photocurrent generation near a strong ohmic shunt in a solar cell.
Function generator
Data acquisition
Excitation was set at the following values: Frequency 1 Hz. The voltage is 2.5V +/-2.5V and the shape of the signal was square.
Double click on the images to see motion.
The lines in data graph show time history of the image’s in the B&W image.
Note that camera is calibrated in absolute temperature, and shows small amplitude temperature swings from 100 mK to 25mK in real time. The yellow line doesn’t not show any change of temperature due to limitation of the real time thermal resolution of the camera (20mK).
Thermal image movie (Colours)
Thermal image (B&W) Timing graph of areas of interest 1, 2, 3 and 4
Solar cell quality control
Map of amplitude at 1 Hz
Map of phase at 1Hz
Map of amplitude at 10 Hz
Map of phase at 10Hz
At higher frequency one can detect accurately the defect on the surface.
VIbrothermography (VT)
Detector technology road map
1998 2000 2002 2004 20061998 2000 2002 2004 2006
30 µm25µm
20µm
320 x 256
30µm
InSb - HgCdTe
15µm
640 x 512
15µm
InSb
640 x 512
20µm
InSb
640 x 512
25µm
InSb
640 x 512
15µm
HgCdTe
2nd generation
3nd generation
AltairThermography
JADE family320x256
InSb & MCT
Altair LIStress and NDT
EMERALD 640x512Insb 25µm
1998 2000 2006
EMERALD 640x512Insb 20µmMCT 15µmQWIP
Silver family320 x 256640x512
InSb 30µm & 15 µm
Titanium family320 x 256640x512
InSb et MCT 30µm & 15 µm
Electronics, Electricity and microelectronics.
Mechanics
Plastic industry
Automotive industry
Aeronautics
Medical diagnosis
Biology
Habitat industry, architectural surfaces and material evaluation, road inspection
Food industry:
predictive maintenance in the newspaper industry petroleum facility
Chemistry industry : inspection of tank and vessels.
Arts : paintings, cultural heritage conservation, frescoes.
Textile, paper industry, pharmacetics, wood industry , cosmetics
Areas of active thermography
Depth information, size information?
Integration (finite elements) of the Heat equation for simplified models (already verycumbersome) and comparison.
Supposed to know a lot of information related to the thermal behavior of the samples(thermal conductivity, convection coefficient, emissivity), materials supposed to behomogenous……..
Classification : learning from samples and decision
Two technics have been used
Gaussian based classification (density probality function)
Neural networks approach.