The best way of modeling a weld heat source depends on many factors.

The first factor to consider is how accurately we want to model the heat source. Few if any welding processes used in industry are controlled more accurately than 5% and many are controlled less accurately. Our knowledge of the values of material properties such as thermal conductivity, specific heat, latent heats, Young's modulus and Poisson's ratio, etc. are rarely known with accuracy greater than 5%. This immediately restricts the accuracy that can be achieved by a model to not more than say 5 to 25 %.

The second factor is our objective in modeling. What do we want to use the weld heat source model for? If it is to be used to predict hot cracking, then it will have to be accurate near the weld pool and point heat sources models will be of little use. If the purpose of the weld heat source is to predict distortion and residual stress in low alloy steel structures, then accurate temperatures below about 600 to 800°C could be most important. Temperatures above this range have much less effect on distortion and residual stress.

The third factor is what information is available for use in a weld heat source model. Perhaps the simplest models use only the weld power input and the shape and size of the weld pool. Important data can be obtained from a macrograph of a weld cross-section. A useful estimate of the front weld pool is a circular arc with diameter equal to the width. The length of the back half of the weld pool is often of the order of twice the width of the weld pool. These rather rough estimates of the weld pool size and shape are often useful when more accurate data describing the weld heat source are not available. These data are relatively easy and cheap to obtain. These data tell us little or nothing about the physics inside the weld pool and in the weld arc.

Models that include the magneto-hydrodynamics of the arc, fluid flow in the weld pool, can involve unstable phenomena such as turbulence that lead to mathematical problems that can be extremely difficult to solve if they can be solved at all. Except in low power welds where the instabilities tend not to arise, the value of such models has been primarily in improving our understanding of the importance of various terms such as Lorentz force, buoyancy force, surface tension forces, etc. In the opinion of these authors, these models have not yet been useful for predicting the behavior of high power production welds.

In these authors’ opinion, the most accurate models of weld heat sources available today are those developed by Sudnik [24, 25, 26 and 27] and his colleagues for specific welding processes. They parameterize the welding process and the parameters are chosen to capture the most important physics of the weld pool. In addition the parameters are carefully correlated with experiment for a given welding process. Although the time and money needed to develop these models for each welding process can be considerable, they have the very important advantage of being the most accurate predictive weld heat source models available. Once developed, they should last for the life of the welding process. In this sense, developing such a weld heat source model is a one-time expense.