The velocity distribution of agitated liquid in a cylindrical mixing vessel was measured, using a set of pitot tubes, the composition of which is shown in Fig. 2. Some of the experimental results are shown in Figs. 4, 5, 6 and 8.
As a result, the flow patterns in an agitated vessel were made clear and it was ascertained that, besides the primary circulation flow around the agitator axis, there was a secondary circulation of liquid caused by the discharging flow from the tip of the impeller as shown in Figs. 7 and 9. Integrating the measured velocity distributions, the discharging flow rates of the impellers were determined and the discharging performances of various impellers (Cf. Table 1) were compared, relative to the power consumption. Dimensionless factor, Nq1, defined by the author, was called the coefficient of discharge. The ratio, N_P/N_q1 in Table 3, shows the relative power required for performing unit quantity of dis-charge.
Furthermore, the power consumed in the neighbourhood of the impeller (NPimp), that is in the cylindrical domain (Cf. hatched region in Fig. 9), was calculated and compared with that consumed in the outer region of the vessel as shown in Table 4.
From these considerations it is concluded that the improvement in the discharging capacity can be accomplished, to a certain extent, by a properly designed impeller, though we shall have to resort to some other devices for more fundamental improvements.

抄録全体を表示

In the previous report4), we suggested that the grindability or the rate of grinding of particles was closely connected with the size distribution pattern of the ground material, which ranged from the bulk crushing to the surface grinding. We assumed, also, that the size distribution of the ground material was affected by the strength of grinding force.
This paper is intended to verify the above-mentioned assumption by means of the ball-mill grinding, which is the most general operation pattern comprizing factors we can easily vary.
Silica balls with four different diameters (and naturally of different weights) and sintercorund balls with one diameter were respectively employed at various rotational speeds of the mill. To investigate the results of the experiment, we assumed the characteristic number of β, which indicates a characteristic of the size distribution relative to the ratio of cumulative weight percent of the coarse part (Y_(r)to that of the fine part (Y_(f)). (Cf. Fig. 4).
When the ball weight or the speed of rotation increased, (or the grinding force increased), β also increased as shown in Eq. 4-2 and Figs. 8 and 9. The values of k_β and k_x were closely related to the particle diameters, viz., when the fed particles were small, the effect of grinding force was small.
From this we may conclude that in fine grinding, the weight of a ball does not count much.
It was observed that in general the grinding rate of particle (X) was determined by the grindability of particle (Γ) and the mechanical efficiency of grinding (η_m), which is the probability of ball impact or that of other grinding forces. Therefore, in fine grinding, Γ becomes negligible, and η_m essential.
Upon this, we assumed that η_m (or impact number per one rotation of the mill) was constant irrespective of the speed of rotation of the mill, and obtained by experiment the relative values of η_m as shown in Fig. 10.
Furthermore, according to H.E. Rose's suggestion9), from the supposed contact area of a couple of balls and probability of impact of balls in the mill., we derived relative efficiency of ball impact, as shown in Eq. 5-13, and ascertained the experimental results and Eq. 5-13, as shown in Fig. 17.

抄録全体を表示

Variations of the concentration of the continuous phase in the pulsed plate column were measured in the experiments on the extraction of acetic acid from methylisobutyl ketone by means of water.
In this column, the longitudinal mixing was very intensive. The variation of the concentration was approximated on the assumption that the backmixing diffusivity was constant throughout the column.
Capacity coefficient eliminated from the effect of the longitudinal mixing and the backmixing diffusivity were calculated in the present work.
It was concluded that the decrease of extraction efficiency in the lower range of operating condition was caused by the longitudinal mixing.

抄録全体を表示

Owing to the stagewise contact which takes place in bubble-cap plate columns, performance data for suchequipments are usually expressed in terms of plate efficiency. A considerable number of investigations on plate efficiencies ever reported have failed to give a coherent picture, because of a great many unknown factors usually involved in them.
The present work has been performed for the purpose of obtaining an expression for predicting the liquid phase mass transfer rate for bubble-cap plate columns. The data presented are derived from the experiment on the absorption of pure CO₂ by city water in a 10 in. I.D. steel column containing one plate and a varying number of caps, from 1 to 7, with 2 and 3 coming in between for the runs, respectively. The bubble-cap is 40mm in diameter, 40mm in height and contains 16 (10.5mm by 3.6mm) rectangular slots per piece. The general layout of the equipment is shown in Fig. 1. When correlating the data thus obtained, a dimensionless equation has been obtained from the basic differential equation (for liquid phase mass transfer at isothermal absorption in a tower-type absorber), as follows: NHL=f(ReL, Se, NX, NZ)
where: NHL=HL/(μμ²/ρρ2g)1/3, ReL=ρuX/μ, Sc=μ/ρDL
NX=X/(μμ²/ρρ²g)1/3, and NX=Z/(μμ²/ρρ²g)^1/3
The liquid film, H.T.U., HL, is calculated from the observed values of NL and the length of the path of liquid flow across the plate. In calculating the true Reynolds number of liquid, it is necessary to have on hand the data on the wetted perimeter or the contact area between gas and liquid flowing on the plate, but since it is impossible to obtain reliable data of such kinds, the approximate Reynolds number of liquid are employed. The one used in this work is based on the depth of clear liquid on the plate and the minimum space between the caps at the center of the plate without gas flow. Because of this approximation, the Reynolds number of gas must also be taken into consideration as one of the necessary operating variables. In this case the Reynolds number of gas employed is calculated from the slot gas velocity and the hydraulic mean diameter of the slot.
A general correlation applicable to the bubble-cap plate columns is presented as within a range of 20% (cf. Fig. 6), but the effect of NZ is not satisfactorily established by this only and further investigation will have to be conducted. The data obtained by Walter and Sherwood3)and Gerster et al.1) are plotted. The data of Gerster et al. show satisfactory agreement with the authors' and the effect of liquid viscosity on HL observed by Walter and sherwood agrees with the effect predicted frotn the present investigation.

抄録全体を表示

In a cylindrical packed tube heated under the uniform wall temperature, the partial differential equation for the temperature in the steady state is given by Eq. (1).
However, when no chemical reaction takes place, Eq. (1)reduces to Eq. (1') with the assumption thatC_p G is uniform over the cross section and ke, r and ke, z are uniform within the bed.
The solutions can be obtained by applying the finite Hankel transform with respect to. In case of semi-infinite cylinder, Eq. (2) representing the boundary condition at the bed inlet and Eq. (3) representing the boundary condition at the inside wall surface, are applied to Eq. (1'). The solution is given by Eqs. (6) and (7), where ξi is the root of Eq. (4). On the other hand, in case of finite cylinder, Eq. (10) which stands for the bed outlet condition should be employed together with Eqs. (2) and (3), in solving Eq. (1'). The solution is shown by Eqs. (6) and (11).
The temperature at the center axis of the packed tube t_c and the mixed-mean temperature over the cross section tm are represented by Eqs. (8) and (9), respectively, which are modified forms of Eq. (6).
However the solution for the temperature distribution in the finite cylinder will be valid for any value of δ, Eq. (11) is nearly equal to Eq. (7) when δ is less than 30 or so.
The value of δ is smaller than 10 for the usual packed beds14), consequently, it may be concluded that Eqs. (6) and (7) are applicable enough, regarding the packed bed as the semi-infinite cylinder.
In case the flow-rate is relatively high, k becomes small, and then Eqs. (6) and (7) are nearly equal to Eq. (13) obtained by Maeda et al4) and Marshall et al1), in which equation the axial heat conduction is neglected. But it should be pointed that Eqs. (6) and (7) serve as a good solution for the low flow-rate range, because k is not so small in this range.
When chemical reaction approximated by q=q0·exp[-αl] takes place, the temperature distribution is reprerented by Eq. (14), assuming the packed bed as the semi-infinite cylinder.

抄録全体を表示