Are you modelling the weld process?  If you do this, you would establish the unloaded residual stresses.  You could use the resulting stress state as the base for a subsequent analysis step for the duty loads.  Alternatively, if you are assuming linear elastic properties, you could superimpose the residual stress results with the duty load case stresses.

To be honest, I haven't tried this with ANSYS, but I have a Msc project student working on a similar problem using ABAQUS, and he is modelling this as two steps.

Dear Dr. Mcmillan,

Thank you for the nice reply. 

I am not modelling welding process, I know the welding residual stress values and idealized distributions. My purpose is to applly welding residual stresses as additonal stress to the working loads however, the analysis would not be linear elastic, I consider materail and goemetrical nonlinearities. 

For instance, in case of fillet welding, tensile residual stress which equals to yield stress of the material occurs edges of the plate. If I apply yield strain as initial strain to the corresponding zones, does the compressive stress automatically occur ? I am not sure this approach is ok. 

Best Regards.

Murat

I think the welding stresses can be simulated by calculating the associated thermal stresses and thereby coupling Thermal analysis and as you have told that you know the residual stress values and the distributions, a simulation of thermal loads may be possible.

Dear Nayak,

Thank you for your kind reply. 

I know the magnitude and distribution of residual stress,so I do not need to perform thermel analysis. Welding residual stress has been calculated by simplified formulation from the literature. The purpose of my study is to consider residual stress as additional load for the ultimate strength calculation. 

Best regards.

Murat

I think that considering the residual stresses in FE modeling might not solve the problem of calculating the weld stregth for static loading, especially if you want to apply design codes. For instance, the design formulae in "Eurocode 3" neglect residual stress (I think that they are implicitly taken into account in the failure criterion). On the other hand, if your FE model has the weld bead modelled explicitly, it would have stress concentration at weld toe/root (even with elasto-plastic model). Which design criterion would then you adopt to compute the ultimate weld strength?

Dear Dr. Benasciutti,

Thank you for your valuable opinions,

I think the problem is not only about welding residual stress, it may be any kind of residual stress effect. So, I wonder if I apply initial strain to the corresponding zone where I calculated the tensile residual stress by simple formulas, would the compressive residual stress automatically develop on the remaining part of the plate?

Best regards.

I think that the determination of ultimate static strength of welded joints by 3D finite element simulations represents a great challenge, irrespective of considering residual stress or not, or initial strain in the joint. This is due to the peculiarities of welded joint (geometry, material, design criteria, etc.). Suitable methods have to be used, which are different than non-welded components.

Dear Dr. Benasciutti

Thanks  for your valuble contributions. I'll consider your suggestions.

Best regards.

Among the real constants of many element types, there is the pre-stress value to be assigned to the element.

YOU CAN USE INISTATE COMMAND FOR THIS

 I have a similar need as Murat. The distribution of residue stress has been measured by contour profile method. I try to incorporate the measured residue stress profile into the finite element modeling. What is the best way to do it? Can I define pre-stress value one by one for every element?

Hi Minhua,

I don't know how the distribution of your residual stress but you can consider the residual stress by imposing initial strain. If you impose the  tensile initial strain to the certain elements, corresponding compressive strain will automatically develop in the remaininig elements to meet the equilibrium condition. 

Bests. 

Hi Murat,

That is a good point. Thanks.

Minhua

Hi Murat!

 I want to apply residual stress but I don`t know how? Actually I want to add residual stress and I know the magnitude and the distribution (Paik model). Have u found the way?

Hello Dear Zohreh,

If you have the magnitude and distribution of the welding residual stress that is tensile residual stress is usually equal to yield stress of the material for ordinary steel. You can introduce tensile  initial strain, which is equal to yield strain, to the model where tensile residual stress is developed. Then, owing to self equilibrium, compressive strain will be automatically developed. I am a Mechanical APDL user, and I employ "inistate" command to do such operation.

Bests. 

 Hello Dear  Murat!

Thank you for you answering. you sayed :" You can introduce tensile  initial strain, which is equal to yield strain, to the model where tensile residual stress is developed. Then, owing to self equilibrium, compressive strain will be automatically developed."

But I don't know how can I do that. please help me? I really need your help.

Hello again,

Until now, I have not carried out detailed collapse analysis regarding welding residual stress but in my recenet calculations, I observed some problems related with imposing "initial strain". In this case, additional deformation occurred, and I coulld not get desired welding residuall stress level if the plate is relatively thin. On the other hand, I have  directiy imposed "initial stress" in the associated elements. In this case, additional deformation wa not generated but when the magnitude of WRS exceeds a certain level, effect of WRS becomes same. 

To conclude, please try both "initial strain" and "initial stress"  techniques. Then, choose the most appropriate one for your model.  You can use "inistate" command in ANSYS APDL environment. 

Hello dear murat!

Thank you for the quick reply.

I am sorry I did not cleery understand what you mean by "I know I can use inistate to export stress. But I don't know how can I impose the stress". Anyway, please see the attached file. I hope it may work for your question. 

Hi Murat,

In the file u attached, u have included 'Pstres' command while the 'nlgeom' is 'on'. But I think u shouldn't include 'pstres' command when u are doing non-linear geometric analysis, instead u should use 'sstif,on' which is 'on' by default when 'nlgeom' is 'on'.