Gas hold-up and liquid circulation in internal loop reactors containing highly viscous newtonian and non-newtonian liquids

Abstract

When highly viscous liquids are used in bubble columns, the bubble population is bimodal; most of the gas passes through as large bubbles, but there is a substantial hold-up of small bubbles whose contribution to the gas flow is small. This paper describes flows of this nature for internal loop reactors (ILRs): an ILR is simply a bubble column containing an internal draught tube to promote liquid circulation. The gas flow enters at the bottom of the draught tube, which contains large and small bubbles: the large bubbles disengage at the top; the small bubbles recirculate with the downflowing liquid outside the draught tube. Two ILRs were used for measurements of gas hold-up, large bubble rise velocity and liquid circulation velocity: (i) a column of diameter 0.15 m containing a draught tube of diameter 0.09 m; (ii) a column of square cross section, 0.6 × 0.6 m, containing a square draught tube, 0.3 × 0.3 m. The rising velocity of single injected bubbles was also measured in clear liquid with no distributor gas flow in a 0.08 m diameter cylindrical column. The results show that the large bubble rising velocity is, to a first approximation, governed by equations appropriate for inviscid motion, notwithstanding the high viscosities of the liquids used: evidently inertial effects are dominant. The small bubbles receive only about 0.3–0.5% of the gas flow. The balance between this input and the rate of coalescence determines the small bubble hold-up. The mean residence time of small bubbles may well be comparable with the “half-life” for oxygen transfer from these bubbles. Hence the common assumption made in mass transfer calculations, that the oxygen concentration in these bubbles is always in equilibrium with the liquid, may not be valid. The liquid circulation velocities in the ILRs were measured by neutrally buoyant flow followers, and calculated from a pressure balance round the circulation loop: the expression used to calculate the pressure loss in the riser had to be modified to take account of liquid carried up by each large bubble; apparently this liquid flow did not contribute to the pressure loss.