Too high pressure gradient in an emptying case
 

Hi guys, I'm modeling the emptying of a tank full of methane pressurized up to 300 atm to the atmosphere through a pipe. I used coupled implicit solver with k & epsilon standard model. What is bothering is that almost with all turbulent models I dont reach to a convergent solution for velocities and the residuals oscillate so much around some quantities which is low enough.The solutions are realistic and the mass flow rate converges but the oscillations never fade away. The Mach number is above 1 in some little regions at the end of the tube which means it does not obey the 2nd law of thermodynamics, but i think its negligible. Would you please tell me which solver or turbulent models or parameters I should choose to overcome this situation?!