experts help please

[mhn10624]

hi
i am working on modeling tig arc welding of thin aluminum plate (simple but weld). i have done this :
1.writing fortran code -dflux- for moving double ellipsoid heat source
2.create part (a .5 m *.5 m square with thickness .005 m)
3. define material (temperature dependent k,density,Cp,...)
4.difine load,body heat flux
5.predifined condition for temperature (ambient=25)
6.inteaction : convection( -h- dependent on temperature) and radiation - constant
7.step1 : 75 sec(welding time) - heat transfer and step2: 325 sec(cooling time)-heat transfer
8.mesh : standard-heat transfer - linear
9.job : load dflux
10. submmit
heat transfer problem is solved but now what is the problem ?!!
conduction is not sensed by the elements as well as it should!
a node .12 m appart from the weld line should have a maximum temp (200 c) as an experimental work reported.but temperature variation is just 25 c ! also nodes dont sense the temperature of the heat source till it reaches them
a node has constant temp till the time the heat source pass it. if it is on the weld line temp varies good but from .05 m to the end of plate side we have no sensible temp change.is this a basical problem of code ABAQUS or ????
thanks alot

 

[corus]

As with all these problems, the first thing is to check your units, particularly when you have density which could be out by a factor of 10^9.

 

[mhn10624]

i checked units many times
aluminium 1100 :
density = 2710 kg/m3
k= 220 w/m.c (varies with temp)
h = 25 w/m2.c
Cp= order 900 j/kg.c
dimensions : .5 m * .5 m * .005 m
Q= n VI w
it is about 10 days that i am checking units and values !!!
please help
what is happening that i cant find it !!

 

[corus]

You could try further checks by, say, having a constant heat source instead of moving it. In addition find a 1D problem with your materials and heat source and compare your solution with the analytical one. These things usually find the mistake you're making.

Input files or .cae files can be uploaded here, providing they're not too big.

 

[mhn10624]

this is fortran code

 

[mhn10624]

and this is .cae
i will check what you said ....
i will become thankful of you so much if you check these files
best regards sir

 

[mhn10624]

note : in dflux v=24 (voltag) is missed ... it should be added please

 

[corus]

I haven't ran the job yet but noticed that in your load definition you have the magnitude set to zero. Change that to 1.

 

[mhn10624]

i changed it to 1
no difference
load is user defined - dflux -
i am realy confused - when i change k from 222 to 2220 , or changing density from 2710 to 2.7 , heat transfers to the further elements and they sense the moving heat source
but with real data of aluminium 1100 , no conduction to further elements . it is not what happens in reality

 

[corus]

I ran the job and it works fine with temperatures up to 500 C whilst welding. In step 2 the cooling seeems rapid, but that must be due to your heat transfer coefficient. Also, you specify cavity radiation but don't use it? Attached is a picture of the temperatures in mid-step 1.

 

[corus]

As an extra comment: In jobs like this it's better practice to exploit the symmetry that exists and halve your model. You could then have a much quicker run-time, and a smaller odb file. You can reconstruct the full geometry in Viewer by using the mirror command in Odb options.

 

[mhn10624]

yes
excatly this is the problem !
on the weld line and distanses about .05 m from it (transverse direction), temperature is sensed and it is ok . but as you see nodes further from this distance dont sense any temp change.
as an example :
from an experimental work , temp. of point (x=.12,y=0,z=.05) has a maximum temp of 200 c . but this modeling reports max temp about 55c!
heat is not conducted as it is conducted in reality.why?
aluminum has high conductivity.if we have a temp raise in one point, the other points of plate sense it and temp change is sensible.
Abaqus cant model high conductivities?!! there many works done with abaqus for aluminum.so what is the problem with my work.why temperature change is just around weld line?
i am realy confused

 

[corus]

As far as I can see from the results temperatures do increase away from the weld line, and you can see that as the weld moves then there is a residual temperature increase that seems to rapidly decay. Temperatures in front of the weld may not increase due to conductivity as much, but that depends on the speed of welding as well as your material properties. At the end of step 1 (and the welding process) then these residual temperatures completely disappear fairly rapidly. I don't know the source of your cooling heat transfer coefficient but I'd expect values of about 5-6 W/m^2 C or thereabouts, for natural comvection. You include surface radiation, and your stefan boltzmann constant should be 5.67e-8 W/m^2 K^4 (I forget what you used). A reduced heat transfer coefficient will generally give you higher temperatures away from the weld and reduced cooling rate after welding.

Abaqus solves for any material thermal conductivity obviously, however it may be less accurate where there is a 'thermal shock' at a surface. Normally you can see this in the results as they show some instability. In these cases you need a much finer mesh, and a smaller time step. As I said before, use symmetry along the weld line and partition the mesh to be much finer towards the weld. I doubt this would alter your results much, though it may increase the calculated maximum temperature where the thermal shock is most severe from your applied heat flux.

 

[mhn10624]

it is about 10 days that i am checking different possibilities ...
i used finer mesh and submmit
i deactivated radiation and submit
i used less 'h' = about 2.5-9 and submmit
but just temperature along weld line changes
max temperature of nodes in transverse direction dont increase as much as it should
i will have an experimental work too
if abaqus show me max temperature of 55c for point (.12,0,.05),and experiment shows 200c(as the experiment already done by a friend), how can i trust answers?!!!

 

[corus]

I altered your model so that it has symmetry, refined the mesh towards the heat source, and reduced the heat transfer coefficient, but it made little difference to your results. One thing to check is the heat source input as the maximum temperature calculated is lower than the melting point of aluminium. In addition, I'd check the results of the experiment as people can make mistakes. For your reference I've attached a picture of the mesh and an animation of the results. This animation may appear jerky as the frames were only saved every so often.

 

[corus]

Here's the picture of the mesh

 

[mhn10624]

dear corus
i did what you said (as you have done) and as you said there is no difference. i called one of my friends who worked weld modeling by Ansys.
he said he had some problems like this in points near the sides (far from weld line). but not such an error !
it is really unacceptable for aluminium that the points dont sense heat (such a heat that raise the centerline temp. from 25 to 600!)
i am really really confused!
my thesis has stoped for about 3 weeks just for this problem.i dont know what to do ?!!
i dont know anyone who is expert in this field.you and maybe your friends are my last hope.
thanks a lot ...
here is experiment data for point (.12,0,.05)

 

[mhn10624]

question 1 :
make the parameters a,b,cf,cr so small (concentrating the heat source)- the temperature of weld line raise to 2500c.but no change to the temperature of the points near the side or in the middle of semi-plate.
why?
question 2 :
make the input power 10times greater (for example volt*10=240 volt). the temperature on weld line rise to 4000c but the temperature of the points near the side or in the middle of semi-plate just rise to 70-75 (from 55 when we had real source with volt=24 !!)
the problem is in conduction or maybe heat storage terms (density*Cp)?!?!?

 

[corus]

Try googling 'temperature distribution in aluminium plate while welding'. There are a few papers on it that you can compare with. This paper

http://jmst.ntou.edu.tw/marine/11-4/213-220.pdf

gives temperatures of about 200C about 7mm from the source (if I've read it right). Perhaps the experimental data is incorrect. Try repeating the test but this time observe it yourself and maybe include additional points on the surface.

 

[mrgoldthorpe]

Suggestion of one check (I haven't/can't check odb): if this is a 2D model, does SOLID SECTION (old style ABAQUS, sorry) specify thickness=0.005; not defaulted to 1m if you use m units.

Just a thought.