Armed with the experience gained here: https://forum.freecadweb.org/viewtopic.php?f=18&t=30697
I returned to the challenge of modeling post-tensioning of a concrete bridge. I finally got it to work, but not after overcoming a few further challenges, i.e.:
1) I used symmetry to cut the model in half. Due to mesh refinement at the anchoring points, the model had simply become too big to be practical.
2) I had to introduce contact constraints at the vertical faces of the tendon to prevent it shooting off laterally
3) Convergence could only be achieved by a) reducing contact stiffness on the vertical faces (which is fine, because they only serve to keep the tendon in place) and b) reducing the temperature of the tendon (to mimic post tensioning) in 2 steps.
All-in-all it took me a full weekend to get it to work
and I therefore do not consider this a practical approach for modeling post-tensioned concrete. A much more robust way to achieve this would be with simple, old-fashioned, small deformation interface elements ... which are unfortunately not available in CCX.
As mentioned, I use half a model with symmetry conditions applied:
A 50x50mm tendon is placed in 51x51mm duct, which is cut out from the beam. The tendon is anchored to two steel anchor plates, which react against the beam.
As before, the tendon is put under tension by dropping the temperature by 500 degrees. This process is achieved in 2 steps to ensure convergence. The deformation of the bridge under dead and live load before shortening the tendon is 171mm, which far exceeds design guidance (<L/300). However, after applying tension, the deformation reduces to 99mm, which just meets design guidance.
The result plots show that the actual tensile stress achieved in the tendon is approximately 1350 MPa (smaller than allowable) and that the required reinforcement ratio is 6%, which both agree well with the values obtained in the pre-tensioning example at the start of this topic.
Time to move on to other challenges