Each electron has an associated transmission probability, put in another common nomenclature is that for every channel you'll have a transmission probability between 0 and 1. So if you have 2 channels you can get between 0 and 2 and so forth.
So your simulations seems perfectly fine.
For bulk systems you can see the number of channels in the bandstructure. Basically each energy crosses a set number of bands, for bulk systems the transmission equals the number of bands crossed at that energy.
Only when introducing defects will you see fractional transmissions that can still maximally be the number of bands for the electrodes at the given energy (remember the electrons originate from the electrodes).
When applying a bias the electrodes are still in equilibrium since one only shifts the electronic structure. This is the basis for NEGF calculations in TranSiesta, OpenMX, ATK etc.. I.e. the eigenspectrum of an electrode at $\mu$ gets all eigenvalues shifted according to $\epsilon(\mu) = \epsilon(0) + \mu$.
So the same procedure about examining the bandstructures from the electrodes applies, simply examine energies while taking into account the chemical potentials.