The high demands on low fuel consumption and low emissions on the combustion engines of both today, and the future, is highly dependent on advanced control systems in order to fulfill these demands. The control systems and strategies are based on models which describe the physical system. The more accuratly the models describe the real world system, the more accurate the control will be, leading to better fuel economy and lower emissions. This master's thesis investigates and improves the mass flow model used for a compressible restriction, such as over the throttle valve, EGR valve, or the wastegate valve, for example. The standard model is evaluated and an improvement is proposed which does not assume isentropic flow. This seems to explain the deviation from the isentropic Psi-function shown in earlier research such as (Andersson:2005). Furthermore a throttle valve is analyzed in ANSYS in order to show the generation of entropy. The presence of pressure pulsations in a combustion engine is also evaluated, especially how they effect the otherwise assumed steady flow model. It is tested if a mean value pressure is sufficient or if one needs to take the pulsations in to account, and the result shows that a mean pressure is sufficient, at least for the throttle when typical intake manifold pulsations is present. A dynamic flow model is also derived which can be useful for pressure ratios close to one. The dynamic flow model is based on the standard equation but with an extra dynamic term, however it is not implemented and tested due to complexity and time limitation. The proposed new non-isentropic flow model has proven promising and can hopefully lead to lower emissions and better fuel economy.