FEA of Welds

[DuncanAnderson]

.

Inv Pro will place welds into a weldment and undertake FEM analysis of assemblies

I little bit of research has shown that FEA of welds is a complex subject

But what about a Weld contraint within the Analysis environment ?

Analysis of welds is a complex topic; the singularities produced at the edge of the weld, the change in material properties in the HAZ, the induced stresses caused by thermal expansion and contraction, the change in material properties caused by the arc energy input are some or the undefined variables.

A lot of variables and too many for those with no knowledge of the effects of welding or FEM - finite element method.

However, there must be enough research data publically available that can provide some default values.

Would a Shrinkwrapped simplification together with a substitution help ?

[blair]

Welds and FEA in Inventor see to cause problems. There have been a number of posts with issues on this forum. One of the problems with Weldments in FEA is interpretation of the model within the FEA program. In most models both pieces of the model contact each other. The software will treat these contacts by default at bonded. This may be fine in a weldment with 100% penetration. Even with advanced programs such as Sim-Mech the HAZ is not taken into account or is the change in materials that have been heat-treated prior to welding. As will all FEA programs, the model setup and interpretation is still at the discretion of the operator. A solid engineering background is still needed to understand the model, software and actual "real-world" issues of the problem. Nothing still beats the design it, engineer it, build it and finally break-it methodology. The digital model method is great way to reduce the number of test models built in the process but one still needs to have real-life actual model verification.

[DuncanAnderson]

.

I want to start a discussion that will hopefully lead to the start of a solution

There are a lot of problems, even for the serious FEA/multi-physics applications. Adesk seem to have thrown their hands up in the air and said it is beyond them, so they will not start for fear of comments along the lines that it is a half baked solution.

However, the journey of a thousand miles starts with taking the first step, not my saying

I can't do anything about the competence of Inv users, other than persuading them to get good training.

I can reduce the number of test units that are broken. Researchers have already broken a lot of test samples of welding and presumably published the peer reviewed results.

There are research papers into how materials are affected by heat as well as stresses induced by heating a cooling.

If the FEA environment could recognise a weld as either the weldment created in Inv's weldment environment or a part created as a weld, additional algorithms could be applied in this area.

Initially it will be a 'ball park' analysis.

However, with more use and refinement as well as research these algorithms will become better.

Ultimately this knowledge will benefit Adesk's Simulation and Multi-Physics applications as well.

[JDMather]

---

@Anonymous wrote:

Ultimately this knowledge will benefit Adesk's Simulation and Multi-Physics applications as well. 

---

I would not expect any significant development occuring beyond "casual user" implimentation in Inventor for several reasons, not the least of which they have other products for the "dedicated user".

[DuncanAnderson]

.

Ultimately I think Adesk would have a simplified, automated FEM analysis of welds in Inv Pro and a more serious, user controlled version within the various Adesk Simulation applications, but all based upon the same knowledge and feedback from users.

[blair]

This would be very specialized software with a great deal of development required. It would be nice but Minot holding my breath waiting for it. It's not in Sim-Mech so it won't be in Inventor for anytime soon.

[DuncanAnderson]

.

I think Adesk could make a start with a new type of FEA Constraint, a Weld Constraint.

This could included, when applied factors for

• type of weld
• HAZ
• weld/filler material
• arc energy input
• any pre-post weld heat treatment
• quality of welder

If you look at the whole list it appears over-whelming

However, if you look at each individual bullet you see it it possible to have default factors, like the 'k-factor' in sheetmetal flattening, which users could modify as more information becomes apparent from their welding shops practices.

Adesk could also help by releasing improved tables of factors as more research is reviewed.

The Weld Institue is a good source of information that could help and I'm sure would be interested in helping.

Is Adesk a member of The Weld Institue ? This also applies to welding plastics.

Over a period of time it would be possible to build up a database of factors that together with research by The Weld Institue could be used to simplify the FEM analysis of welds.