The characterization of CSTRs requires that power consumption and
flow moved by the impeller should be available at steady state conditions, i.e.
for a fully developed flow field.
The possible alternatives
to obtain a developed flow field for a reactor are:
(i) a steady state approach, or a (ii) a full transient approach.
In the first case, the flow field can be
computed for a fixed position of the impeller with respect to the baffles
[Wechsler et al., 1999].
Calculations are stopped when no more changes in the solution are observed.
Selection of a ``frozen'' position for the impeller is arbitrary and can
significantly affect the calculated flow field, especially for large impeller
or wide baffles for which impeller-baffles interaction is enhanced.
Otherwise, the domain can be divided into two regions, a rotating part 
containing the impeller in which equations are solved in a rotating
frame of reference, and a static part the remaining part
of the vessel , in which the static frame of reference is used.
In the static region, effect of the impeller is azimuthally averaged.
The obtained
flow field is independent from impeller position and time
[Harvey et al., 1995].
In the full transient approach [Brucato et al, 1998, Schafer et al., 1998],
the movement of the impeller is
explicitly simulated and the evolution of the flow field is calculated until
no other variations than those due to blade frequency can be observed.
The domain is divided into a rotating and a static region as before, and
coupling between regions is made at each time for corresponding positions,
maintaining azimuthal periodicity coming from the impeller region at the
interface with the static domain.
The converged solution obtained in this way is a pseudo-steady state
since variations
in time are present due to relative position between vessel and blades.
The simulated flow field is the one instantaneoulsly observable for a real
tank.
For the simulations made in this work, a full transient approach based on the sliding mesh method is adopted. The region containing the impeller is a subdomain in which fluid equations are solved in a rotating frame of reference. This subdomain is connected to the rest of the vessel through a sliding surface. Velocity and pressure data calculated at the sliding surface are used as boundary condition for the static domain. Correspondence between static and rotating regions is modified at each time step, according to the changing relative position of impeller and vessel. Starting condition for each simulation is fluid at rest. From starting time onward, the impeller is moving at constant rotation speed.