NPH anisotropic , enthalpy can't conserve

Hi all,
I am trying to do isoenthalpic-isobaric simulation for copper. When I use 'aniso' to control the stress tensor, the enthalpy of the system can't conserve. Actually, the enthalpy keeps decreasing, and this happens to both solid and liquid copper. I tried different potentials(Cu_u3.eam, Cu_u6.eam and one sutton-chen potential built by myself) to test the enthalpy change, and run the simulation for 15ns to give enough time for equilibrium, however all got decreasing results.
I pasted my input command as below. I used the read_restart to get the initial solid configuration and velocity(about 1030K). The box is cuboid, longer in z direction, and contains 2592 atoms. The restart file is attached.
Could anyone help me to solve this problem? If you need more informations, please let me know. Thanks.

Regards,
Yue

restart1050_nph_1 (224 KB)

Aidan may have a comment/idea.

Steve

Just as the instantaneous temperature fluctuates under npt, the
instantaneous enthalpy (ke+pe+p*v) will fluctuate under nph. It should
settle down to a constant average
value once the system is equilibrated. If the enthalpy is decreasing, this
is because one or more of ke, pe, p, and sign(p)*v is decreasing. Which
one is it? In solids, it is possible for relaxation processes to occur
very slowly. In liquids, you should not use aniso, as there is no
restoring force to oppose shape changes. You should start out with a
simpler system e.g. liquid, with nph keyword iso.

Aidan