stress analysis of welded and un welded joints 1

Welds are usually not modeled for larger structures (unless they are supposed to be analyzed for fatigue but then we model just the connection). You could run some tests on simplified models to see how the addition of welds changes results in your case.

On Apollo Engineering website there used to be a nice article with a comparison of results when various weld modeling approaches are used in the case of a simple tee joint. Unfortunately, the article disappeared but here are the results they got (in form of deviation from the hot spot stress method):
- single body, no weld - 14,11%
- two bodies, modeled welds, single bonded contact - 14,05%
- single body, welds modeled as chamfers - 8,33%
- single body, welds modeled as fillets - 4,32%
- multi-body, welds modeled as triangle extrusions, bonded contact between welds and plates, frictionless contact between plates - 3,72%

Since you use SolidWorks, check the article "How to Model Weldments for Efficient Finite Element Analysis" on the Mentored Engineer website. There's also a short book titled "Solidworks Simulation for Real Weldments" that might be of interest to you.

The normal practice we follow is to not explicitly model welds in the analysis. If welds are full pen, no need for weld evaluation/modelling. For welds other than full pen, take nodal resultant forces and moments at weld location and evaluate the weld with analytical calculation (based on structural code AISC/EN or relevant code applicable). For more details on calculation part, refer Shigley's Mechanical Engineering Design.

See if you can find "Determination of Weld Loads and Throat Requirements Using Finite Element Analysis with Shell Element Models - A Comparison with Classical Analysis" by M. A. Weaver.

Rick Fischer
Principal Engineer
Argonne National Laboratory