about large minus absolute pressure

richard

in my case, i simulate a tunnel with a 30 degrees downstream slope.there is a free surface in the tunnel with water flowing through it. the boundary conditions are: two inlets with velocity inlet for water and pressure inlet for air on the upstream, and the gauge total pressure is set to 0; and one outlet with pressure outlet, and the gauge pressure is set to 0.i initialized the porblem with water full of the tunnel and the x-orientation velocity is set to 23m/s,the water inlet velocity.then i start the simulation, and soon after the start, there is a very high pressure of about 5*10^6Pa existing near the inlet, and the solution is converged and the iterate goes on.the time step is 0.0001s and about 5s later, the absolute pressure becomes minus and decrease further more with iteration process. the absolute pressure limit of above 0 doesn't work! does someone meet such situation? how to explain such things i meet? thanks in advance.