In most cases it is not practical to use FEA to calculate weld stress directly. It is generally a better practice to try and model the weld stiffness and use the stress in the parent material as a basis for evaluating weld stress.
If you think about the real world geometry of your typical weld, it is full of cracks, stress concentrations, internal voids and flaws. Every weld is different. It would be near impossible to include all of these features in a model. If you did make a model with some or all of these features (and many have done so), what would you learn about your design? You would probably learn that there are some very high stress singularities and the weld would fail your basic design criteria, yet many structures have been built successfully with welds.
Traditionally engineers calculate the load carried by the weld and use that for calculating the stress through the weld throat. Knock down factors are applied to account for possible flaws due to the welding process (overhead, shop, field, machine, etc…). The knock down factors act as an extra safety factor to account for all those flaws that are not practical to calculate. Different industries and companies use different factors based on their requirements. Most factors range from 30-70% efficiency.
This same process may be applied to your FEA model, only use the stress in the parent material near the weld to estimate the weld throat stress.
There are exceptions, some welds are ground smooth and are examined for internal flaws. In these cases the weld effectively becomes part of the parent material except for some possible heat damaged zones where strength may be degraded.
Creo Simulate has some unique options available for handling welded joints in shell element models. These are discussed in the following links.
