Philippe HERVE
Laboratoire d’Energétique et d’économie d’Energie
Université Paris X Nanterre
1, chemin Desvallières
92410 Ville d’Avray
France
Isabelle TOSELLO
Commissariat à l’énergie atomique
DEC / ASSEMBLAGE
CEA SACLAY
91191 Gif sur Yvette
France
- Impulse Thermography –
Application to Welding Arc
1.
Introduction
Improving the quality of parts assembled by welding is based upon the control of thermal supply conditions to the material. It is
thus necessary to know the temperature distribution involved on the welding scene.
The scene is complex because it presents high temperature gradients along with solid-to-liquid phase changes and a possible
supply of material.
In the case of arc welding (GTAW), the thermographic problem is particularly hard to solve because of the plasma emission
spectrum [2].
The classical method of assessing the temperature by aiming in the infrared leads to huge errors because of the variations in
surface emissivity.
We offer a method which allows to determine the temperature distribution by means of an active optical method. The precision
relies on a targeting with short wavelengths.
2.
Impulse photo-thermometry principle
The surface at temperature T0 is periodically excited by a light source which thus creates a small variation T0 in the surface
temperature. The variation of this temperature is a relatively complex function depending on  the absorption factor of the
material, a the thermal diffusivity, the form of the impulse and the excitation frequency.
The basis for the calculation of T0 is the equation of heat equation [1] but generally,  and a are unknowns and T0 is a new
unknown.
At wavelengths such that T<3000 m K, the surface luminance can be expressed by means of simple analytical equations
with emissivity , transmission factor .
L  cte   e  C2

;
exp 

L  cte   
 C2
T

T  To

T
Using two wavelengths 1 and 2, we obtain :
 C
ΔL1
ε
 cte 1 1  exp  2
ΔL2
ε2 2
 Τ
 1 1 
  
 λ1 λ2 
The error over the temperature is due to the fact that 1 , 2 ,1 ,2 are not known.
T 
 
T ² 2 1

 ln  1 1 
C2 2  1
  2 2 
The same type of error can be found in bichromatic pyrometers. Thus the interest of operating at the shortest possible
wavelength in order to diminish the unknown term which is the source of error.
It is a fact that the error T0 is proportional to the product of the wavelengths selected for the measurement. At the same time,
the system sensitivity increases rapidly towards short wavelengths [3].
3.
Experimental set-up
The experimental set-up is shown on figure 1.
The two wavelengths selected for the measurement have been chosen in the range over which argon plasma does not
present intense rays. The excitation laser (CO2) emits at =10,6µm.
An optical system (Fig. 2.) provides for the recording of two images at 1 and 2 on a matrix camera (1340x400 pixels) using
liquid nitrogen as coolant (Fig. 4.).
The succession of images are then processed in order to obtain the ratio between the two alternative components at 1 and 2.
Periodic thermal
impulses
Measurement at two
wavelengths of the
radiation emitted by
the welding zone
Laser
exciter
CCD
Camera
1
Reference
signal
Fusion bath
Bilens

Argon plasma
2
0
f
Amplitude of the
modulated
radiation
emitted by the
fusion bath
synchronous
detection
0
Bath T
Fig. 1. Experimental set-up
4.
Melted zone
Fig. 2. Optical system
Tests and results
The method has been tested in an industrial environment where the electromagnetic disturbances due to the welding process
were very important. Fig. 3. shows, on a single pixel, the detector response to a sinusoidal excitation of the surface
temperature.
The measurement precision is of the order of 50K around the material fusion temperature (for instance 2000K)
modulation (V)
Reference signal
Photomultiplier signal
time (s)
Fig. 3. Test with additional source of light, PM filter 632.8nm ; fm=210 Hz
5.
Fig. 4. Matrix camera (1340x400 pixels)
Conclusion and prospects
This method of measurement is able to determine a temperature range with a relative precision of about 2,5%.
In a current development, an extra measurement within the infrared will allow for the knowledge of phase change frontiers.
Our final objective is to control the whole set of thermal supplies in welding.
REFERENCES
1
H. GRIEM « Plasma spectroscopy » Mc GRAW HILL - 1975
2
H.S. CARSLAW et J.C. JAEGER « Operational Methods in Applied Mathematics » 1960.
[3]
P. HERVE A. MOREL “ Thermography improvements using ultraviolet pyrometry. Quantitative
Infrared Thremography” p 26-31 QUIRT 1996