It's rewritten a bit and now the high value is constantly in the first row, independent from the number of evaluations in a session. The results from the differential equation look similiar to each other, but the fef3 varies wildly: For nut =4 there is fef3 = 0.0001, the other nut's have fef3's in the order of 1000. For the first solution the vars2[[4]] is 111.059 and the term with fef3 from rmsd explains your trouble in my eyes:

In[3]:= (111.059/0.0001)^2
Out[3]= 1.23341*10^12

Please find the notebook attached.

Attachments: