The three-dimensional unsteady cavitating propeller flow is modeled in non-linear theory. A low-order boundary element (panel) method [22, 8] is employed. The panels are placed on both sides of the blade surface, as shown on the left part of Fig. 2. The method was first developed for fully wetted unsteady propeller flows [14, 27, 11, 23]. The names of the codes are PSF-10 for steady and PUF-10 for unsteady flow. It was then extended in the case of partially and supercavitating 2-D hydrofoils (PCPAN and SCPAN are the corresponding codes) and then in the case of cavitating 3-D hydrofoils (MXPAN3D). The method predicts more accurate cavity shapes and pressure distributions on the blade (especially at the leading edge, tip and blade/hub intersection) than PUF-3A.
A comparison of the predicted cavity shapes and volumes
from THPUF-3A and PROPCAV is given
in Fig. 6 for propeller DTMB 4990.
The propeller geometry and the used axial wake inflow are given in
Fig. 5. Notice that the predicted
cavity shapes and the maximum cavity volume are quite similar.
However, the predicted cavity from PROPCAV grows ``later'' and collapses
``earlier'' than that predicted from THPUF-3A,
as it can be seen in Fig. 6. This means that the cavity
volume velocity, thus the propeller induced hull pressures, will
be larger in the case of PROPCAV. It should also be noted that
THPUF-3A with a typical 
grid takes about 3 mins
and PROPCAV with a typical 
grid takes about 60 mins to run
on a DEC Alpha 600(5/266).
Figure 6: Cavity patterns and cavity volume predicted
by THPUF-3A and PROPCAV.
Figure 5: Geometry for propeller DTMB 4990 and contour plot of
the used axial wake inflow.