In your example, I'm assuming the loads applied in the dynamic relaxation phase are being held constant in the transient phase and there is no additional loading. The energy ratio plot may be an inconsequential energy bookkeeping error.  Unfortunately, there is not a way to output glstat data during the dynamic relaxation phase, so we really don't know the energy ratio at the end of dynamic relaxation.  Between time=0 and t=10 of the transient phase, are you seeing a state of equilibrium, e.g., no variation of element stresses or nodal displacements?  If not, that indicates that a state of equilibrium was not obtained during the dynamic relaxation phase and some adjustments to *CONTROL_DYNAMIC_RELAXATION are needed.  You can postprocess relax and d3drlf to get a clearer sense of the response during the dynamic relaxation phase.
You are navigating away from the AIS Discovery experience