More recent computational-based methods with realistic modeling of fan housing geometries have been developed where the flow in the bladed region is modeled by a distribution of bound and trailing-vortices and by body-force fields [47] and [48]. Some model only the through-flow region with the boundaries of regions B and C specified [13], while others attempt to model the whole flow field including resolution of the eccentric vortex [47] and [48]. These methods can be classified as actuator techniques and in principal encompass simpler methods based on the actuator disk and basic potential vortex approximations like those described above. Actuator methods have been used widely in the gas turbine industry to represent both stationary and rotating blade rows in multistage turbomachines (e.g. streamline curvature and through-flow methods [54] and [55]). The key feature of these simplified models is that the discrete number of blades is replaced by a distributed bound-vortex or body-force field, and their function is to reproduce the effects generated by the physical blades, which is to add or remove angular momentum (rcθ) to or from the fluid passing through the blade row. If the blades are rotating, their function is also to impart a change of total pressure and total temperature to the fluid, in addition to changing the fluid's angular momentum. In either case (stationary or rotating blades), the method assumes that the flow is “well behaved” in that the fluid closely follows the blade shape in the relative frame. In other words, the blades provide effective flow guidance and hence flow deviation is small, and a simple cascade model can be used to estimate the performance of the blades. The use of the actuator method appears to be the most promising steady-flow prediction method for cross-flow fan applications. However, it is much more challenging to develop such methods for cross-flow fans than for conventional turbomachines in that there are three distinct regions through the impeller, with region A acting as a turbomachine flow-path where a cascade model can be implemented, while regions B and C are much more challenging to model.
For aero-propulsion applications, compressibility effects can become significant and the most appropriate simplified steady-flow methods are CFD-based methods with a shock-capturing capability. One such method is the formulation based on solving the compressible-flow Navier–Stokes equations with added source terms in the bladed region to model the presence of the blades as a distributed body-force field [56]. This formulation can be implemented into existing CFD solvers [47] and [48]. To illustrate the concept, we now describe a simplified method based on a non-axisymmetric actuator-duct approximation using the body-force field approach [47] and [48]. For simplicity the flow is assumed to be 2D and incompressible.
Let View the MathML source be the unit vector normal to the relative mean streamlines in the impeller region, or the streamlines that the average flow follows as seen in the relative frame. Note that the streamlines can have different shape, depending on their radial and angular positions in the impeller region. The flow-tangency condition can be described as