1. It should be reqarded as a necessary hut not sufficient condition in t h e valIda*ion of m y three-dimsnsional methcd, that whenever the metho& is used eo predict two-dimensional flows the results should be tne same. as for an eouivalent two-dimensicnal metho 3 cortaininq the same turbulence model. The prqsent method can he used to calculate two-dimensional flows i n either of two ways: The rroqram can be run i n the infinite swept winq mode w i t h zero sweep, o'r th- attachment line module can he used with i CI. Both of these modes Of operation have been used to calculate a wide variety of the nominally twodinecsional ?xperimr)ntal flows used as test C.YS?S a t th? 1968 Stanford Conference? on turbulent boundary layer comptation. In all cases t h e deqree of aqieenSnt with experiment was nubstant.ia 1ly the same a s that displayed a t ehs conference by proqrams based on effective viscosity models sinilar to the one used i n the present analysis.
2. €Or a l l of the wins bcundary layer calculations presented i n thin section, t h e outer inviscid flow boundary conrlitions were computed by a threedinnnsional potential flow analysis proqram usins th 5 panel-type int Iuence coefficient rethod deV9LOpd by Joonson and Rubbert I. In a *.vpicnl winq-bcdy calculation, the surface 8 of the uim s n j body are paneled w i t h linearly varyinq source panels, and the uinq wake and m e bound liftinq system insi.lc the win" .are modeled 'by quadratically VaKyinJ 4vlhlet panels. Fiqure 13 shows ?he source panel system used to compute the potenri.al flow about the winq-body of the Poiinq 727-200 airplane. For clarity, t h e doublet panel liftinq system and wake are not shcwn. i j has nct proved to he a disadvantage, as t h e exp?nse involved i n constructing the grid is trivial compared to the costs of tho Fotsntial flow and boundary layer analyses. A typical boundary layer qrid constructed for t h e upper surface of the 727-200 wing a t CL = .PO3 is shown i n Fiqure 1.
3. Results oE boundary layer calculations for the 727-200 winq at a Reynolds number typical of full-scale fliqht are shown i n Figures 19, 15, and 16. The calculations were made in the qrid of Figure 1, with 40 stations alonq the span, 50 stations along the root chord, and 40 paints throuqh the layer a t each surface station. The boundary layer calculations consumed a total of 910 CPU seconds on a CDc 6610 computer. When the calculations were repeated with 25 p i n t s throuqh the layer, the results were practically `identical with those shown for 40 pcints, and CPlJ time was reducsd to 600 seconds. With reqard to these computinq times it should be mentioned that the present program was written with clarity and ease of debuqqinq, rather than efficiency, as the main qoals. With Acre efficient coding the analysis undoubtedly can te wade to run considerably faster.
4. outer fLa,etrdhes for 727-206 win 3 raper Wfaca with MMta qan almlpis at mot.