From now on keep the number of k-points and the wrap around errors fixed (i.e. try to use always the same ratio between the value which avoids all wrap around errors and the actual FFT-mesh). Test how many bulk and vacuum layers are necessary to get a reasonable surface energy, and a reasonable converged force on the first (and possibly second) slab layer.
Mind: Do not change more than one parameter from one calculation to the next calculation. It is almost impossible to compare two calculations which differ in the number of k-points and in the size of the supercell. Be very careful about the FFT-meshes: If you increase the size of the supercell without increasing the size of the FFT mesh the results do not improve. Actually results get even worse in this case because the wrap around error increases.