In general MD simulation, the system is in equilibrium, constitutes the equation of motion of the particle, realizes a minute (constant NVE) ensemble and canonical (constant NVT) ensemble according to purpose. As a result, it is possible to know all the trajectories of the individual particles (molecules, atoms) making up the system, but it is rare to effectively utilize minute information up to that extent, what is of interest in practice are various "Average value" and "fluctuation" of physical quantity. In the equilibrium state, the time average of the flux of the physical quantity is zero, but the value at each instant varies around zero. The correlation time of this fluctuation is closely related to the decay function of the macroflow (Green-Kubo equation), which is a proportional coefficient between the spatial gradient of the target physical quantity and the macroflow, the equilibrium equation MD simulation. This relational expression can be calculated. A simulation of a considerably long time (at least about 10 times the correlation time of variation) is necessary to actually obtain various transport coefficients from the equilibrium variation.

I think after several analysis you can judge the degree of satiation of your MD system but it depends on your system. http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin/gmx-tutorials/index.html