It is widely known that hydrogen is extremely harmful to the safety of a metal construction. It influences the mechanical properties of base metal or a joint or, directly results in fracture during welding due to cracking, so called hydrogen induced cracking НІС or hydrogen assisted cracking НАС. While the time to cool from 800 °С and 500 °С for steels determines the hardening, the time to cool to 100 °С determines hydrogen retained in the metal. Slower cooling times allow more hydrogen to escape to the air.
HAZ cold cracking, sometimes called underbead cracking, Figure
6- 1, is a well known phenomenon in welding technology.
Figure 6-1: Hydrogen underbead (assisted) cracking in the HAZ of a single-pass, shielded-metal arc bead-on plate weld deposited on a C-Mn microalloy steel, from 
Cracks are caused essentially by residual tensile stresses that form as a result of thermal shrinkage when a weld cools. The residual stress level often exceeds the yield strength of the material which in turn causes plastic flow. When the plastic capacity of the joint region is exhausted cracks can form. The situation is complicated by the simultaneous formation of hardened microstructures of limited plastic capacity. Moreover, it is further complicated by the presence of hydrogen. It has long been recognized that necessary though not sufficient conditions for НІС are:
a) Sufficiently high hydrogen concentration,
b) Sufficiently sensitive microstructure,
c) Sufficiently high tensile stress.
However, there is disagreement on the values of numbers to define sufficiently.
Hydrogen forms in the arc and is “pumped” into the HAZ through the weld metal. In this model the effect of hydrogen is assumed to dominate the mechanism. Hydrogen diffuses into the HAZ at high temperatures but escapes as the weld cools.
The hypothesis is that a critical level of hydrogen exists in the HAZ below which cold cracks do not form. This critical level
depends on composition, joint shape, workpiece size and shape, yield level of the base material and ambient temperature.
Figure 6-2: Preheat to avoid cracking computations; submerged arc welding: voltage 25 V, welding speed 5mm/s, ambient temperature 10°C; local preheat conditions: heater width 100mm, heater strength O. IW/mm, from [22 and 23].
In practice preheat is used to slow down the cooling and provide an opportunity for the HAZ hydrogen to drop below the critical level [22 and 28]. At temperatures below roughly 100 to 200 °С, the diffusion of hydrogen can be dominated by diffusion along the gradient of the hydrostatic stress toward high hydrostatic (tensile) regions. Below this temperature diffusion is so slow that any hydrogen left can be considered to be trapped. At the same time cracks form at temperatures below 100 °С and therefore the hydrogen environment is fixed before cracking occurs. A critical time is defined in the Yurioka model  which is the cooling time to 100°C, i. e., the time for hydrogen to drop to the critical level. If the real cooling time is greater than the critical cooling time then cracking is avoided.
The real cooling time of the weld can be calculated with the help of a long time low temperature heat flow model. Such computations were used to generate the data presented in Figure 6-2. This Figure is submitted to show just how useful the system can be in practice.
The preheat level needed to avoid underbead cracking for (A) uniform preheat (B) preheat with electrical strip heaters is shown.
The cracking mechanism is very complex and it will be some time before it is well enough understood to model it precisely. Nevertheless this model is an important development and the computations form a useful basis for engineering judgment in practice.
Pipeline designers are aware that hydrogen induced cracking can cause failures and they are anxious to specify designs that minimize the risk of failure. Late in the life of a pipeline, НІС is often associated with corrosion. Early in the life of a pipeline or a repair, НІС is often associated with welds. Specifically, our objective is to develop criteria for the design of weld procedures that minimize the risk of failure due to НІС.
It is desirable to weld natural gas pipelines under pressure both for making new connections to the pipeline, called hot-taps, and to repair damage to the pipeline. Welding on the pressurized pipeline avoids the costs of shutting down the pipeline and depressurizing. In thinner walled pipes, there is a risk of bum-through when welding while the pipeline is pressurized. Should bum-through occur, the safety of the welding personnel are at risk and the pipe will have to be shut down to repair the bum-through.
Kiefner [1 and 2] studied the problem in some detail. He recognized that the risk of bum-through was greatest in thin wall pipes. He also proposed a model to predict the risk of bum-through. The model used a rather simple 2D finite difference method with forward Euler time integration for the thermal analysis. If the maximum internal wall temperature exceeded 982 С (1800 °F), then the risk of bum-through was considered high. Kiefner included the effect to flowing the gas or liquid on the thermal analysis through the convection coefficient. However, he did not consider the effect of stress and deformation.
Displacement, strain and stress are computed by solving the conservation of linear momentum and mass equations. If the pressure and temperature are high enough, the wall thin enough and the time at temperature long enough, a groove can form under the weld by visco-plastic flow, i. e., creep. Such a groove changes the wall thickness and the distance from the bottom of the weld pool to the inner wall of the pipe and the temperature on the inner wall of the pipe. This will change the carburized layer thickness and composition. This coupling or interaction between thermal and stress analysis is an important nonlinear effect that has usually been neglected in computational weld mechanics.
Burn-through is primarily a high temperature creep phenomena and requires modeling viscous flow. Solving the energy equation on the current deformed geometry captures an important nonlinear coupling between stress analysis and energy analysis.