1. [7-9].By using a general z,y to E(z),v(z,y) coordinate transformation, a complex similarity transformation is avoided and a solution-adaptive grid is easily employed.
2. Figure5comparesthepresentvjscousresults toNavier-Stokesresults.6 Thedashed linesareNavier-Stokes results while the solid lines are the current results. On the main airfoil the present results compare much better with experimental data than the Navier-Stokes results. However, this is believed to be due to some shortcoming of the Navier-Stokescode, since the present code should not be expected to produce a better solution, just a less expensive one. The trailing-edgeregion Cp distributions are in fair agreement with each other but slightly lower than experiment. The Navier-Stokes solution generally comparesmorefavorablywith experimentaldata than the present computationsover most of the shrouds. However, in theupper inlet regiononthelowersurfaceof theupper shroud, the current results are in better agreement with the experimental data than the Navier-Stokes results. In the lower inlet region (deked similar to the upper inlet area) the present results are in better agreement with experimentaldata,becausetheNavier-Stokescomputation failed to pick up the peak Cp at the leadingedge. Both the present and Navier-Stokes results differ from the experimental data at the trailing edge of the main airfoil and on the lower surface of the upper shroud near the leading edge. This indicates that the boundary-layer assumptionsarenot responsible,butpossiblytheinfluence of high curvature on the turbulence is not being properly modeled. Thepresentresults areroughlyequivalentto the Navier-Stokesresults interms ofaccuracybut areobtained in roughly 1-2% of the CPU time requiredby the Navier-Stokesalgorithm.
3. Additional insight into the nature of this flow is obtained from the boundary-layer velocity-vectorplots of Figs. 6a-6c. The length and orientation of the arrows at each grid point (x,y) are proportional to the velocity vector. Figure 6aisaclose-upviewof thethroatregion. In the lower inlet a thick, separated boundary-layer exist on thelowersurfaceof themain airfoil. This thick boundarylayer (and to some extent the relatively thin separated boundary-layer on the upper surfaceof the lowershroud) has accelerated the inviscid flow in this narrow channel and increasesthe leadingedgesuction peak on the upper surfaceofthelowershroud. Forpurposesofillustratingthe separation region,the velocityvectorsat the trailingedge of the main airfoilaremagnified by a factor of four while the geometry is drawn to scale in Fig. 6b. Notice from Fig. 6b that the separation region on the lower surface extends to more surfacegrid pointsand is thickerrelative to the upper surface. Theseparation at the trailing edges of the main airfoilwas alsoobservedby the Navier-Stokes computations. The developmentofthe boundary layeron both surfacesof the upper shroudareclearlyillustrated.
4. ThedeHavillandAircraft of Canada, Limited, "AnalysisofResults from Test of an AsymmetricHigh Speed Augmentor-Wing Model (WTCB) in the NAE Twc-Dimensional High Reynolds Number Blowdown Tunmi,* DHC-DIR 75-2, August 1975.