I try to simulate the surface tension of water by using Lammps. First, I build a water box (4*4*4 nm3), and obtain the required data after enough simulation time (about 5 ns for equilibration and another 1 ns for collecting data).

The surface tension of water is calculated by this equation, r = 0.5 * Lz * (Pzz-0.5*(Pxx+Pyy)), but the result I get is far small than experimental value. Many people have reported their simulation results of surface tension, which is exactly fit experimental values, and I just follow the steps in those papers. Can anyone give me some suggestions about this ?