Welding simulation

jacob

Hi, does anyone have experience with welding simulation?
I am using a method where one bead step is heated each time, but I get about 10 000 °C, which does not seam physical.
Is this behavior normal?

mf

jacob

Hello,
first try what is done here please, it is a common mistake because the default value is not 1 (in my opinion it should be =1 in THER_NON_LINE):
https://forum.code-aster.org/public/d/28772-thermo-analysis-going-wrong
Also, how you defined the heat capacity is important, recheck if all is correct (units, values, typos?). Is heat capacity defined as nonlinear curve (@melting temp you should have a steep peak in the curve) then your timeStep is also important, if in doubt lower timeStep. I've done a welding simulation in the past, but it was difficult. If all goes wrong, define T instead of heat input, this should be a lot easier,

Mario.

jacob

Hi, thanks for the reply, but I don't think there is a problem with numerics.

I am talking about physical phenomenon.

To simplify, what I mean by this I did the same in solidworks, and also getting thousands of °C.
Initial temperature is 80 °C, on surronding surfaces the is convection and only the bead segment is heated with 14 700 W for 2.4 s.

What do you mean that there should be steep peak in the curve of rho_cp?
My varies from 3.6e6 to 5e6 and then it is constant.

The guy here also does not have any cut off for cp
https://youtu.be/CRYmOGCl1sI?list=PLvACBM1uN9Epa0PJZkWCYnU4v7FE-o6wh&t=2402

mf

jacob

Perhaps from the latent heat at the melting point? The self-proclaimed "master" wearing a cap reversed has one in row 12 of his excel. And of course it is related to model/numerics/stepping somehow, the numbers are huge! You should also include radiation, these losses dominate >600°C. See this simulation I have done 2 years ago, the material data is quite extensive: https://github.com/emefff/Thermal-Evolution-In-Welding-In-Code-Aster/blob/main/welding_plates_50steps.comm. It adds a hot chunk of weld material (a wedge) in every step. Also choose input power carefully, the losses in a TIG-arc etc are also huge (radiation!). The real power can be much lower, no wonder there is often a factor of 10 (!) in difference in power input between TIG and EB for example, also the heated volume is completely different. Amongst being a means of welding a TIG-arc is also a very powerful UV and visible lamp as well as an ozone generator (+ ionized gas molecules).

Mario.

jacob

when I tried to use his values

*rho

rho_cp_1 = DEFI_FONCTION(INFO=1,
[1,0]<stdout>: INTERPOL='LIN',
[1,0]<stdout>: NOM_PARA='TEMP',
[1,0]<stdout>: NOM_RESU='TOUTRESU',
[1,0]<stdout>: PROL_DROITE='CONSTANT',
[1,0]<stdout>: PROL_GAUCHE='CONSTANT',
[1,0]<stdout>: VALE=(0, 3532500, 75, 3815100, 175, 4074150, 225, 4176200, 275, 4372450, 325, 4505900, 375, 470215, 475, 5196700, 575, 5879650, 675, 6641100, 725, 11241200, 775, 7457500, 1500, 3140000, 1540, 3140000, 1590, 6648950, 1840, 6648950, 1800, 3140000, 2050, 3140000, 5000, 3140000),
[1,0]<stdout>: VERIF='CROISSANT')
[1,0]<stdout>:
[1,0]<stdout>:
[1,0]<stdout>: ╔════════════════════════════════════════════════════════════════════════════════════════════════╗
[1,0]<stdout>: ║ <EXCEPTION> <FONCT0_44> ║
[1,0]<stdout>: ║ ║
[1,0]<stdout>: ║ Les abscisses de la fonction 0000000e ne sont pas strictement croissantes. ║
[1,0]<stdout>: ╚══════════════════════════════════════════════════════════════════════════════════

I had a look at you simulation a while ago but your bead is quite large, when I had only 4 beads the temperature was much lower. But the bead was unrealistically large.

jacob

Thus the calculated Q_max should be applied like this?

mf

Sure, why not? Ramping up power, forces, temperatures etc. is never a bad idea in FEA. However, I don't think it should be necessary to solve your problem. I am pretty sure the reason for your temperature overshoots lies somewhere else. But as of now you haven't shown any details, so: ???

Mario.

jacob

Yes, it's not it.
if I use this

and also include radiation, the temperature is about 3000 °C. The max of cp has the main influence.

mf

jacob

This is more than 10 times the value of liquid steel above 1536°C in J/(kg.K)?? Seems excessive.

What about the power input, how do you set that? If I remember correctly, FLUX_REP is a power/area and SOURCE is a power/volume?? However, the units are correct?

Mario.

jacob

Yes, the units are correct, I used SOURCE so W/m**3. I have got the same results with solidworks as you can see above.
The q is based on real values, but of course I don't have physical properties with respect to temperature of the real weld.

mf

jacob

So everything is alright or is something wrong in both? I can't tell.

Mario.

jacob

That's what I don't know. The initial and boundary conditions are all right, but it just seems to me that 10 000 ° is unrealistically high.

mf

jacob
...... and we will probably never find out if you do not share .comm etc.......

jacob

url=https://forum.code-aster.org/public/assets/files/2026-02-20/1771586391-533149-weld-1-1.7z

mf

jacob
Hello again,

well, it seems your shared files are only a part of the whole thing. However, you do not have any RAYONNEMENT set active here (maybe not shared??). The definition of the material regarding THERMAL seems ok for SI units (m, kg, s, ...). However, your E is for N.mm-system (or you are welding some plastic material), which does not affect the thermal part of course. Other than that I don't see anything wrong.
What could be, and that is not in the shared file, is that the newly attached material is not 'connected' to the mesh in the same timestep. Meaning, your Q heats up the new weld material on its own (isolated from the parts) and does not dissipate the heat immediately.
PARM_THETA/THETA is still not = 1.

Mario.

EDIT: don't know your welding method but heat input per mm seems high according to what I've read: e.g. TIG with low alloy steel is 0.5-1.5kJ/mm see here https://www.ijmerr.com/SpecialIssue/13_(p.89-94).pdf. Efficiency with TIG is only 0.6-0.75 not 1 (as I said it's a great arc lamp and gas ionizer): https://www.colliewelding.com/hicalculators.php. I assumed TIG because among the traditional methods it's the best.

jacob

Hi,
yes, because the problem is about high temperature for one step, so I wanted to keep it as simple as possible. I get the same results in soliworks, so it's not about CA, but about welding simulation in general.
RAYONNEMENT and THETA = 1 have insignificant influence.
In mechanics I don't use SI.
The problem is not about new added material.

I use input values from our welders.

jacob

with radiation I am at 7000

DEBUT(LANG='EN')

V_1 = 1

Reference_Temp = 20

Initial_Temp = 80

Weld_Temp = 2500

eta = 1

q = 2.1e6

u = 7e-3

Q = q * u * eta

L_1 = 0.0162397

t_1 = L_1 / u

MESH_T = LIRE_MAILLAGE(UNITE=2)

MESH_T = MODI_MAILLAGE(reuse=MESH_T,
ECHELLE=0.001,
MAILLAGE=MESH_T)

MESH_T = DEFI_GROUP(reuse=MESH_T,
CREA_GROUP_MA=_F(GROUP_MA=('Group_Of_All_Volumes', ),
NOM='ALL_VOLUMES'),
MAILLAGE=MESH_T)

MESH_T = DEFI_GROUP(reuse=MESH_T,
CREA_GROUP_MA=(_F(NOM='PARTS',
UNION=('BODY', 'BONNET')),
_F(DIFFE=('ALL_VOLUMES', 'PARTS', 'WELD_1_1'),
NOM='WELD_OFF')),
MAILLAGE=MESH_T)

MODEL_T = AFFE_MODELE(AFFE=_F(GROUP_MA=('PARTS', 'WELD_1_1', 'ambient', 'ambient_1_1', 'ambient_0_2', 'ambient_0_3', 'ambient_0_4'),
MODELISATION='3D_DIAG',
PHENOMENE='THERMIQUE'),
MAILLAGE=MESH_T)

DUMMY_T = AFFE_MODELE(AFFE=_F(GROUP_MA=('Group_Of_All_Volumes', ),
MODELISATION='3D',
PHENOMENE='MECANIQUE'),
MAILLAGE=MESH_T)

h = DEFI_CONSTANTE(VALE=10.0)

#comment: h_weld = DEFI_CONSTANTE(VALE=1000.0)

sigma = DEFI_CONSTANTE(VALE=5.67e-08)

epsilon = DEFI_CONSTANTE(VALE=0.2)

epsilon_weld = DEFI_CONSTANTE(VALE=0.65)

T_ambient = DEFI_CONSTANTE(VALE=Reference_Temp)

Lambda = DEFI_FONCTION(NOM_PARA='TEMP',
PROL_DROITE='CONSTANT',
PROL_GAUCHE='CONSTANT',
VALE=(0, 51.9, 100, 51.1, 200, 49, 300, 46.1, 400, 42.7, 500, 39.4, 600, 35.6, 700, 31.8, 800, 26, 1000, 27.2, 1500, 29.7))

rho_cp = DEFI_FONCTION(NOM_PARA='TEMP',
PROL_DROITE='CONSTANT',
PROL_GAUCHE='CONSTANT',
VALE=(0, 3532500.0, 75, 3805623.0, 175, 4049173.125, 225, 4142750.5, 275, 4329212.875, 325, 4452135.333333333, 375, 4635661.0, 475, 5099055.0, 575, 5739587.0, 675, 6456037.5, 725, 10905306.5, 775, 7219821.875, 1500, 2939210.526315789, 1540, 2934305.263157895, 1590, 6200407.77631579, 1840, 6135489.684210527, 1890, 2891384.2105263155, 2050, 2871763.1578947366, 5000, 2510000.0))

steel_source = DEFI_MATERIAU(THER_NL=_F(LAMBDA=Lambda,
RHO_CP=rho_cp))

steel = DEFI_MATERIAU(THER_NL=_F(LAMBDA=Lambda,
RHO_CP=rho_cp))

MATERIAL_T = AFFE_MATERIAU(AFFE=(_F(GROUP_MA='WELD_1_1',
MATER=steel_source),
_F(GROUP_MA='PARTS',
MATER=steel)),
MODELE=MODEL_T)

Convection = AFFE_CHAR_THER_F(ECHANGE=_F(COEF_H=h,
GROUP_MA=('ambient', ),
TEMP_EXT=T_ambient),
MODELE=MODEL_T)

Radiation = AFFE_CHAR_THER_F(MODELE=MODEL_T,
RAYONNEMENT=(_F(EPSILON=epsilon,
GROUP_MA=('ambient', ),
SIGMA=sigma,
TEMP_EXT=T_ambient),
_F(EPSILON=epsilon_weld,
GROUP_MA=('ambient_1_1', ),
SIGMA=sigma,
TEMP_EXT=T_ambient)))

Dummy_DEPL_T = CREA_CHAMP(AFFE=_F(NOM_CMP=('DX', ),
TOUT='OUI',
VALE=(0.0, )),
MODELE=DUMMY_T,
OPERATION='AFFE',
TYPE_CHAM='NOEU_DEPL_R')

I_1 = POST_ELEM(CHAM_GD=Dummy_DEPL_T,
INTEGRALE=_F(GROUP_MA=('WELD_1_1', ),
NOM_CHAM='DEPL',
NOM_CMP=('DX', ),
TYPE_MAILLE='3D'),
MODELE=DUMMY_T)

V_1 = EXTR_TABLE(NOM_PARA='VOL',
TABLE=I_1,
TYPE_RESU='REEL')

IMPR_TABLE(TABLE=I_1,
UNITE=11)

FIN()

applying radiation on other surfaces than ambient_1_1 has insignificant effect.