Thanks @jousefm, I have had a look at your simulation. I have done what you suggested and given the load a time function. I have also reversed the direction of the load (I want the links in tension) so I can see the stress in the flange of the nut. I have left the contacts as physical, as you said this should be OK. A fourth run failed.
Next I tried reversing the master-slave relationship of the contacts. A 5th run failed.
Can you see what is wrong?
Meanwhile I will try a simulation with bonded contacts.
I have two conceptual difficulties (for now).
Firstly, the SimScale documentation and tutorials I have found so far do not give much guidance on how to choose between bonded and physical contact. How do I chose?
Secondly, as I understand it, the ‘time’ function of the load actually means that an FEA run needs to iteratively increment the load. I think I have an idea why it might need to, but an explanantion would help.