81 



this result. In fact, uniform wall temperature has 

 produced the largest transition Reynolds numbers to 

 date. The primary difference between the results 

 shown here and those published previously is that 

 the present experimental curve reaches the limit 

 Reynolds number of 42 x 10^ at a lower overheat 

 than before. This change is attributed to the 

 improvement of the exit conditons with the develop- 

 ment of the laminar flow nozzle. 



All of the data of Figure 3 were taken by main- 

 taining laminar flow over the full length of the 

 tube and observing transition at the exit. If the 

 flow velocity is increased further, so that the 

 transition region moves upstream in the test 

 section, the measured transition Reynolds numbers 

 are much lower. In addition, there is a hysteresis 

 effect when transition is allowed to move more than 

 about 1 m upstream from the exit. That is, to 

 restore fully laminar flow over the full tube length 

 the velocity must be reduced to a value lower than 

 that which previously yielded fully laminar flow. 

 This hysteresis may be a phenomenon which is accen- 

 tuated by the flow tube geometry. The free stream 

 in the flow tube is confined by the boundary layer, 

 so that the boundary layer can influence the free 

 stream once it becomes turbulent. This free stream 

 influence could propagate upstream, which has led 

 to conjecture about disturbances from the test 

 section exit affecting the transition Reynolds 

 number. 



To test this hypothesis of downstream disturbances 

 affecting transition Reynolds number, a separate 

 study has been conducted to determine the dependence 

 of transition upon the tube exit geometry. As dis- 

 cussed above , there are three types of exit nozzle 

 available: orifice plates, the smooth contraction, 

 and the plug valve. In addition, the length of 

 unheated tube between the heated test section and 

 the exit can be varied from zero to 3.7 m in incre- 

 ments of 1.22 m. For each configuration, transition 

 can be determined either at the exit itself or at 

 the end of the heated section . Transition at the 

 exit is easily determined by flow visualization, as 

 shown in Figure 4. This photograph of the smooth 

 exit contraction shows laminar flow (4a) and turbu- 

 lent flow {4b) , both at a length Reynolds number 

 of approximately 40 x 10 . In Figure 4a, note 

 the glassy region very near the exit, which soon 

 becomes milky in appearance as the air-water shear 

 layer undergoes transition. The longitudinal streaks 

 in Figure 4a are appraently due to Goertler vortices 

 generated in the concave part of the smooth exit 

 contraction. They are not seen with the plug valve 

 exit, which has no concave region. 



The data of Figure 3 are for one 1.22 m extension 

 section on the end of the heated section, followed 

 by either the smooth contraction or the orifice 

 plate. Transition is measured at the exit in either 

 case . Note that the transition Reynolds numbers 

 with the orifice plate exit are about 20 percent 

 lower than with the smooth contraction, showing a 

 definite effect of the exit condition. Figure 5 

 shows the same comparison with 2 . 44 m of unheated 

 extension tube between the test section and exit. 

 Here we see a much larger difference between results 

 with the orifice and with the smooth contraction. 

 The smooth contraction transition Reynolds numbers 

 are nearly the same as with 1.22 m of extension, 

 tiibe , while the orifice Reynolds numbers have 

 dropped almost by a factor of two. Clearly the 

 effect of the exit condition upon transition Reynolds 



FIGURE 4a. Exit jet from smooth nozzle: 

 boundary layer. 



FIGURE 4b. Exit jet from smooth nozzle: 

 boundary layer. 



turbulent 



