Date of Award

Spring 2013

Document Type

Thesis

Degree Name

Master of Science in Chemical Engineering - (M.S.)

Department

Chemical, Biological and Pharmaceutical Engineering

First Advisor

Piero M. Armenante

Second Advisor

Laurent Simon

Third Advisor

Robert Benedict Barat

Abstract

Glass-lined, stirred reactors and tanks are of significant industrial importance, especially in the pharmaceutical and fine chemical industries. These reactors are manufactured with a “glass-lining,” i.e., a glass layer applied to the agitator, the inside of the reactor/tank and to any of the surfaces in contact with its contents in order to maximize corrosion resistance, facilitate reactor cleaning, and minimize product contamination. Because of glass-lining fabrication issues, a retreat blade impeller with a low impeller clearance off the tank bottom is commonly used in glass-lined reactors. In addition, since wall baffles cannot be easily mounted on the wall of glass-lining reactors, a single baffle, such as a “beavertail” baffle, mounted from the top of the reactor is utilized instead.

Despite its common use in the pharmaceutical industry, some of the most important mixing characteristics of this type of reactor have not been fully studied, such as the power dissipated by the impeller under different baffling conditions and blend time, i.e., the time required by a system to achieve a predetermined level of homogeneity. Therefore, this work was focused on the determination of the impeller power dissipation and the blend time in these reactors as a function of a number of variables commonly varied during the operation of these reactors, including different liquid levels (fill ratios), impeller agitation speed, and baffling configurations.

In this study, a torispherical-bottomed, 61-L, scaled-down model of a commercial reactor (DeDietrich) similar to the type of glass-lined reactors frequently utilized in the pharmaceutical industry is used. The blend time and impeller power dissipation for this system are experimentally obtained as a function of liquid level (i.e., liquid height-to-tank diameter ratio, H/T), baffling configurations, and the agitation rates. Three baffling configurations are considered, i.e., a partially baffled system (where a single beavertail was used), a fully baffled system (i.e., four rectangular baffles) and an unbaffled system. The H/T ratio, corresponding to the ratio of the liquid level to the reactor diameter, is varied between 0.3 and 1. Six different agitation rates between 75 and 200 rpm are considered.

The Power Number, Np, is found to be a function of the liquid level, baffling system, impeller type and impeller Reynolds Number. Larger values of Np are associated with more completely baffled systems. In addition, Np decreased with decreasing H/T ratios.

The blend time to achieve 95% homogeneity of a tracer, 095, is found to be inversely proportional to the agitation rate for the partially and fully baffled systems, although large deviations are present at lower H/T ratios. The blend time is not always inversely proportional to the agitation rate for the unbaffled system. The dimensionless blending time, 095N, is also obtained for all baffling configurations, H/T ratios and agitation rates. 095N is found to be largely independent of the impeller Reynolds Number for the partially and fully baffled systems for H/T ratio between 0.7 and 1. The blend time and dimensionless blending time results indicate that they both are functions of the liquid level, impeller Reynolds Number and baffling configuration.

Share

COinS