Effects of liquid viscosity on flow patterns were experimentally studied and so far as flow patterns of paddle type impeller were concerned, more concrete explanations were given on the correlation between N_P and Re, which is considered to be an over-all reflection of flow pattern. In the second place, the relation between flow patterns and scale-up was studicd, followed by the study concerning those of geometrically dissimilar impellers.
Experimental conditions were as follows:
Experiment 1: (effect of liquid viscosity)
Tank diameter: 28cm
Liquid (water and glycerine solution) viscosity: 1-108c.p.
Liquid depth: 13cm; Paddle (L=12cm, W=1.6cm)
Experiment 2: (scale-up)
Tank diameter: 15.5-50.0cm
Liquid (water) depth: approximately equal to tank dia. Geometrical ratios relating to paddles, tanks (including baffle plates) and so on were the same as those in Ex. 1.
Experiment 3: (geometrical dissimilarity)
Tank diameter: 28cm
Liquid (water) depth: 26cm
Paddles: dimensions employed are shown in Table 1.
The procedure of measuring liquid velocity reported here is the same as that which was reported in the previous paper.¹⁾
The effect of liquid viscosity on flow patterns is apparent as shown in Fig. 6 (c), in which the pattern is no longer independent of rotational speed of the impeller. This figure corresponds to that of the experimental region (c) described in Fig. 5. Flow patterns expected from the curves in Fig. 5 are checked in Figs. 6 and 7.
Figs. 8 and 9 show flow patterns in geometrically similar mixing systems. Each figure indicates that the pattern remains unchanged, irrespective of dimensions of apparatuses and rotational speed of impellers, so far as the turbulent state is concerned, provided the depression of liquid surface when unbaffled is not taken into consideration.
Flow patterns, particularly tangential velocity distributions in Fig. 4, indicate marked differences among geometrically dissimilar impellers, but a rather simple correlation between C^* and Re^* was observed as presented in Fig. 10, covering a considerably wide range of experimental conditions.
The procedure through which the correlation was made is also described in the previous paper¹⁾.