If you are having problems getting your solution started, it will often be due to a poor initial guess. For many simple modeled flows, such as incompressible laminar, or incompressible turbulent with the Zero Equation model, a zero initial condition for all variables is sufficient for robust convergence.

Usually, however, better initial convergence rates can be achieved if the pressure and temperature initial values are set close to the boundary condition values, and if the velocity components are some fraction of an expected average value (that is, under estimating initial velocities is better than over estimating). For turbulent flow simulations using the k-e model, some initial velocity scale is strongly recommended.

For more difficult flows, such as compressible flow in a complex geometry, with the 2nd order discretization option (blend factor of 1.0), and with multiple pressure specified openings, you may find that the linear equation solver fails no matter what initial values are selected.

For difficult flows, you should first obtained a solution using more robust solver settings and models. For details, see Monitoring and Obtaining Convergence.

An incompressible case in which the reference pressure is zero is likely to contain negative pressures. This is not a problem for incompressible flow because only the pressure difference is of importance. For a compressible case, the absolute pressure is of importance and negative pressures are non-physical.