About SIA (or defect) insertion, how to fix the symmetry

Dear All

My question about, that I have tried to insert an SIA interstitial into crystal structure like iron,
has maybe 2000 atoms,
what I did:
remove one atom form perfect structure and place two atoms instead which will simulate the SIA dummbel, but my problem is when I try to find how it is different in term of formation energy between <111> SIA dummbel and <110> SIA dummbel, I see the results out of expectation. since it is known that the formation energy for <110> is lower than the formation energy of <111> SIA defects but my result is opposite.

this because the once we do the simulation these interstitials reorganize into a more stable configuration. Therefore the energy that I have assigned to a particular orientation (my initial one) may correspond to a different configuration (the relaxed).

this the script I used to find the formation energy:

If you want the formation energies of unstable SIAs, just do the
minimize. Completely skip the dynamics at 30K as this may lead to
unpredictable results. You better check what is the real configuration
of atoms after minimize.

BTW. why would you first remove one atom? Usually formation energy of
SIA is the energy of the block of n+1 atoms - (n+1) times energy of
perfect atom, which is energy of perfect block of n divided by n.