I would go back and determine why I needed to use soft springs to make the linear analysis work. From what you describe the model needs to be fully constrained and shouldn't need the soft spring option.
Non-linear doesn't have soft spring because it doesn't make sense. Non-linear can handle large deflections.
Thanks for your comments.
This is my model outline. I have cut off half of the model and apply symmetry constrains on all the center surfaces.
For geometry constrains: I have fully fix the 1st constrain point. For the 2rd constrain point, I set its vertical displacement to be zero. which I hope it will make the whole model to bend due to the thermal force. Is that correct?
Under such constrains set, software complains that the strain in the first time step is too small so it cannot converge. It is propobaly make sense to me because my wire is sitting on a HTTC substrateand I just included the thermal force(25 to 90 degrees) and gravity in the model.
Under such case, is that mean I have to initalize a tiny displacement on wire during certain time step to make it converge?
moel ouline.bmp 1.8 MB
From what you show I would use sliding constraint on the side face, place two constraints at location on mutually perpendicular to each other and the side face and then place another "vertical" constraint a location two. Then run a linear analysis without soft springs.
If it will not run, consider checking that everything is bonded properly as this is an assembly analysis.
You can check for unbonded components by putting a fixed constraint or a small force on them one by one or looking to see if there is a separation when you use soft springs.