Welding engineers have recognized the need for accurate temperature measurements to provide better process control and ensure welds of a higher quality. It is important to make sure that welds occur at a desired temperature and cool appropriately. By monitoring thermal gradients, it is possible to determine several characteristics of the weld, all of which are pertinent to the quality of the weld. Among these characteristics are weld bead width, penetration depth, weld position control, and cooling rates. Some manufacturers do not have adequate methods for monitoring thermal changes, and this inadequacy limits their process control.  Due to the demand for tight tolerances and precise components, it is becoming increasingly necessary to examine methods for real-time temperature monitoring in welding processes. It has been shown that the temperature history of a material undergoing a welding process is one aspect that requires further investigation.  In the past there has been significant research looking at both predictive heat transfer models and real-time infrared sensing and the benefits each can provide in welding applications.
Early Weld Modeling and Temperature Measurement
J.M. Barry and his colleagues  improved an existing mathematical model for the welding of a steel plate held subject to several requirements. Previous models provided reasonable predictions when a) the temperature was below 1000°F, b) the location of interest was more than 1.5 plate thicknesses away from the weld, and c) the plate was fully penetrated by the weld.
In Barry’s set up they took temperature measurements near single pass submerged arc bead-on-plate welds on steel plates with thicknesses of ¼, ½, and 1 inch. Several methods were used to measure the surface temperature. These methods included temperature sensitive lacquers, radiation pyrometry, and thermocouples that were welded directly to the plate. Table 2-1 shows the conditions under which the welding occurred.
It was found that the temperature sensitive lacquers were useful in determining the location of where peak temperatures occurred, but they were not able to tell the magnitude of the peak temperatures. Efforts involving radiation pyrometry were not completely successful because of large variations in the emissivity even after the surface had been treated in an attempt to control the optical properties of the target plate. Radiation pyrometry was useful, however, in showing that the thermocouple did in fact provide an accurate measurement of the surface temperature.
Chromel-alumel, also known as K-type, thermocouples with a 0.004 in. diameter provided the only useful method for collecting data, and they were welded to the top surface and edges of the target plate. Generally, thermocouples are not ideal for arc welding processes because of the rapid thermal cycles and the high thermal conductivity of the base metal. It has been shown that thermocouples are subject to significant error when arc welding aluminum, but with steel as the base metal, whose thermal conductivity is lower by one order of magnitude, they provide acceptable measurements. Several previous investigations even examined temperature internal to the base metal by drilling holes and inserting the thermocouples. The main problem with this approach is that by altering the geometry of the work piece, the temperature profile is also affected. As a result, the temperatures measured by the thermocouples are not representative of realistic base metal temperatures.
Two useful temperature relationships are presented using point source considerations. Equation 3 models the peak top surface temperature at a point some distance away from a partial penetration weld . The equation is presented below and it is subject to the following assumptions: constant thermal properties, constant torch speed, straight torch travel, no heat transfer from the surface of the metal to the surroundings, edges are far enough away as to not influence the temperature profile, and steady state conditions.
Where: Tp is the peak temperature [°F], T0 is the initial uniform temperature of plate [°F], ρ is the density of the base metal [lb/ft3], Cp is the specific heat of the base metal [Btu/lb °F], Y is the distance from the weld [ft], t is the thickness of base metal [ft], V is the arc travel speed [ft/hr], Z is the fraction of total arc heat delivered to the base metal [%], q is the arc power [Btu/hr], and Tm is the melting temperature [°F].
A significant source of error in this equation is a direct result of the thermal gradient that exists between where the weld occurs and the point at some distance away from the weld where the temperature is to be determined. Because the thermal properties of steel vary considerably with changes in temperature, approximations of these quantities must be made which in turn introduce error into the model.
Equation 4 models the peak temperature on the bottom side of the base metal. This relationship has been shown to be accurate in instances where the bead width to base metal thickness ratio is less than 1.4.
Where: TpB = centerline peak temperature of the base metal bottom face [°F], and b is the effective width of the heat source [ft].
The above equation is an improvement over earlier methods of calculating peak bottom surface temperatures, but it is still rough. It does provide answers that are of the correct magnitude when b is taken as the width of the weld bead. 
Using a bead width to base metal thickness ratio between 0.8 and 1.2, Figure 2-1 shows experimental data plotted along curves calculated by Equation 4. The experimental data points fall mostly within the predicted range. This implies that a strip heating source model may be useful for some applications.
Both of these calculations have limited utility, but are adequate predictors in the absence of more advanced methods or techniques. Each model is subject to its own restrictions, and neither one exactly mirrors a real physical process. In the scope of this project, a more important consideration is that neither method offers real-time feedback of temperatures, temperature changes or temperature gradients.
The following paper is described in detail because it outlines important initial considerations for the development of an experimental setup where a TIG welding process can be monitored with an infrared sensor. Among these initial considerations are where to focus the infrared sensor on the work piece, what type of infrared sensor is best suited for the application, and how to determine an appropriate value of the emissivity of the work piece.
In 1963, Ramsey et al.  carried out some of the earliest research on using infrared temperature measurement as applied to welding processes. Their main objective was to determine if in fact infrared radiation could be used reliably to determine the base metal surface temperature profile and from that information further correlate those patterns with changes in welding parameters such as penetration depth and bead width. If such a signal could be accurately acquired, then new methodologies for welding control, manipulation of welding current and welding speed, could be made available.
First, Ramsey et al. deemed it wise to determine the temperature profile of the metal around the weld pool so that the best location for temperature measurement could be located. In welding processes involving thin plates, heat flows parallel to the plane of the base metal and is essentially two dimensional. Figure 2-2 illustrates the welding coordinate systems for this model.
2 LITERATURE REVIEW
2.1 Early Weld Modeling and Temperature Measurement
2.2 IR Temperature Measurement of Base Metals
2.3 Applications of Infrared Signals to Control TIG Welding Processes
2.3.1 Joint Tracking
2.3.2 Bead Width and Weld Penetration
2.3.3 Sensing of Cooling Rates
3 EXPERIMENTAL SETUP
4 RESULTS AND DISCUSSION
4.1 Preliminary Thermocouple Analysis
4.2 Infrared Sensor Results
4.2.1 Viewing Angle
4.2.2 Shield Testing
4.2.3 Variability Between Similar Trials
4.2.4 Multiple Runs on a Single Target Plate
4.3 Thermocouple Results
5 CONCLUSION AND RECOMMENDATIONS
&GET THE COMPLETE PROJECT
Non-contact Temperature Measurement of Stainless Steel in a TIG Welding Process