It is often assumed that the heat generation due to mechanical work is small and so we can use uncoupled analysis separating the thermal model and mechanical model. Does any one know of a published article which investigates the effect of such assumption, giving a reason why it is valid in terms of physical phenomena?
In my opinion:
For welding, sequential coupling method is usually used to simulate trasient deformation and stress. I think the reasion are: (1) mechanical behavior of weld zone is introduced by heating and espically cooling process, while the heat generated from mechanical behavior is weak and thus ignored; (2) the convergence is difficult to debug because of the high nonlinear characteristic, if the full coupled method is used.
Hi,
To be more specific, uncoupled analysis technique gives reasonable results in case of arc and laser welding. While in friction stir welding the choice of coupled analysis will be significant.
https://pp.bme.hu/ci/article/download/9257/7087/.
Thanks Youmin. I agree to you on both your points. Do you know an example where the heat generated from mechanical work has been quantified?
Hi Joseph,
Interesting question. I don't know of any studies that have looked specifically at this issue although it is mentioned in Lars-Erik Lindgren's review papers on weld modelling (Lindgren 2001 and 2006). However in Friction Stir Welding, all of the heat generation is by frictional and viscous effects so maybe your literature search should start there.
One thing you could do for a plate butt weld is calculate an upper bound for the energy generation due to mechanical effects by considering a plate of rigid-perfectly plastic material. Then you have:
upper bound = UTS x cross-sectional area x failure strain x length
This roughly approximates the maximum heat you could produce mechanically without cycling. Then you can compare this to the heat input from the arc... However the real test of whether visco-plastic heating is significant is whether it is enough to significantly change the local mechanical properties, by further raising the local temperature sufficiently for the stress-strain curve to be reduced significantly.
Youmin's second point about complexity of the material constitutive model is important. Even without considering viscous effects, it is extremely difficult to produce a reliable model of most welding processes, particularly if you are aiming to calculate the resulting residual stress. See the work of Mike Smith and the NeT round-robins for examples. The result is normally strongly dependent on eg. the hardening law used and how any solid-state phase transformations are accounted for... So pragmatically, it would be very difficult to ensure that a model which incorporated all of these effects and had was thermal-mechanically coupled was actually useful, in the sense of being able to predict anything reliably!
I think you can easily estimate this:
The energy of plastic deformation is the area under stress-strain curve.
About 90% of this is converted into heat. So knowing the amount of plastic deformation, you can estimate how much energy is converted into heat and use the heat capacity to get a temp. increase (under adiabatic conditions).
Alternatively, you can directly calculate an adiabatic stress-strain curve from isothermal data and see how strong they deviate.
- Badges
- Science topic
- Similar topics
- Heat