Using the Amorpous cell module I built a cell with 300 molecules of C8H18 and 10  molecules of C20H42, with Compass forcefield, forcefield assigned, charge group electrostatic and atom based VdW, then with forcite I've done a geometry optimization.After the geometry optimization, I have done equilibrium by forcite（Anneal  NVT and NPT），I have made sure that the cell was fully relaxed. Then , I conducted a 100ps Forcite shear job with the shear rate of 0.1 （1/ps） with COMPASS forcefield and time step of 1fs with the NVT ensemble at the temperature of 298K.Since the shear direction does not affect the results， so I used the software default value (B (BC)), the results showed that the viscosity was 0.09cp.
Then I conducted a 200ps NVT dynamics simulation at the temperature of 298K , the time step was 1fs and the frame were outputted every 100 steps ,after the dynamics simulation , I analyzed the  stress autocorrelation function, and the results showed the viscosity of the mixture was 14.35cp.
However , at the room temperature and atmospheric pressure， the viscosity of the mixture was about 5cp，
As you can see, the viscosity deviation between experiment and simulation is too huge.How could I get a better simulation results？

Thank you for your time and attention！

Really thanks for your help.