Mixing in high throughput experimentation reactors
The application of High Throughput Experimentation (HTE) stirred vessels in the catalyst and pharmaceutical industries enable the parallel screening of potential formulations. Such tasks only require a fraction of the raw material and experimental time that are needed in conventional lab scale react...
Main Author: | |
---|---|
Published: |
University of Birmingham
2009
|
Subjects: | |
Online Access: | https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.519674 |
id |
ndltd-bl.uk-oai-ethos.bl.uk-519674 |
---|---|
record_format |
oai_dc |
collection |
NDLTD |
sources |
NDLTD |
topic |
660.2832 TP Chemical technology |
spellingShingle |
660.2832 TP Chemical technology Chung, Kenneth Hoi Kan Mixing in high throughput experimentation reactors |
description |
The application of High Throughput Experimentation (HTE) stirred vessels in the catalyst and pharmaceutical industries enable the parallel screening of potential formulations. Such tasks only require a fraction of the raw material and experimental time that are needed in conventional lab scale reactors. However their small sizes (~ 15-250 ml) contribute to a low Reynolds Number, Re, which results in poor mixing efficiencies. Together with unconventional geometries, i.e. lack of baffles and simple impeller designs, their fundamentals are not fully understood. The present study applies Particle Image Velocimetry (2-D PIV) and Planar Laser Induced Fluorescence (PLIF) techniques to a HTE scale stirred vessel (T = 45 mm, V = 72 ml) to determine the mixing behaviour. Three mixing strategies: centreline unbaffled (U), conventional baffled (B) and off-centre eccentric agitator (E) configurations, were investigated using a pitched blade turbine (PBT). Experiments were performed in the high transitional regime (Re \( \approx\) 6000) using distilled water as the working fluid. A uniform power input of P/V = 168Wm\(^{-3}\) was applied. A method based on multiple horizontal and vertical 2-D PIV measurements was used to reconstruct the 3-D flow field in each of these configurations, since the conventional 3-D PIV is unusable at this scale. It was found that the determination of turbulent kinetic energy (TKE) using the isotropic assumption was perfectly valid for (B), but will lead to a considerable underestimation in both (U) and (E). In addition to the three configurations, a square section vessel (S) (T = 41.5 mm, V = 72 ml) and regular vessel with a tilted impeller axis (T) were also studied. With a modified experimental procedure a log variance method for mixing time was applied using PLIF where all the usable pixels in an image were accounted for. Not only was (U) found to be highly inefficient, (B) also registered a slower mixing time due to a small amount of tracer being trapped behind the baffles, which makes the (E), (S) and (T) an even better choice in turbulent mixing. The use of (S) enabled the formation of a more compact HTE unit, also its trailing vortices were able to reach a height of y/H = 0.6, bringing more energy to the upper reaches of vessel, as visualised by angle resolved PIV measurements. The flow number of impeller and the amount of pseudo-turbulence were also determined and they agreed well with literature values. However in the laminar mixing regime using Polypropylene Glycol (PPG) as the Newtonian working fluid (μ = 0.4–0.8 Pa s, P/V = 0.6–5.5 kWm\(^{-3}\), Re = 5–35), at low Re values, (S) only managed a mixing performance comparable to the (U) configuration as the baffling effect of its four corners are less pronounced. However, mixing performance improves in (S) at higher Re values. (B) and (E) gave a comparable mixing performance, suggesting (E) should always be adopted for its viability in both turbulent and laminar mixing regimes. For gas-liquid mixing using air and water (P\(_G\)/V = 168 Wm\(^{-3}\), Q\(_G\) = 0.5–1.0 vvm), an image analysis algorithm was developed which enabled measurement of gas and liquid phase velocities separately. In addition to the PBT, a Rushton Disk Turbine (RDT) was also used (C = D = 0.33T). The power input required for the small mixing vessel to achieve complete gas dispersion was not achievable at the required gas flow rate; hence the experiments are carried out in the flooding regime. This had the advantage of clearly discernible differences between gas and liquid flow pattern for validation purposes. A new image algorithm was written to separate out bubbles in the imaging plane which transforms the in-plane bubbles into tracer particles. The local velocities of the gas phase are then obtained using the conventional cross-correlation technique. The results showed qualitative agreement with experimental observations of global gas phase flow patterns in literature. |
author |
Chung, Kenneth Hoi Kan |
author_facet |
Chung, Kenneth Hoi Kan |
author_sort |
Chung, Kenneth Hoi Kan |
title |
Mixing in high throughput experimentation reactors |
title_short |
Mixing in high throughput experimentation reactors |
title_full |
Mixing in high throughput experimentation reactors |
title_fullStr |
Mixing in high throughput experimentation reactors |
title_full_unstemmed |
Mixing in high throughput experimentation reactors |
title_sort |
mixing in high throughput experimentation reactors |
publisher |
University of Birmingham |
publishDate |
2009 |
url |
https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.519674 |
work_keys_str_mv |
AT chungkennethhoikan mixinginhighthroughputexperimentationreactors |
_version_ |
1719013979864432640 |
spelling |
ndltd-bl.uk-oai-ethos.bl.uk-5196742019-04-03T06:40:53ZMixing in high throughput experimentation reactorsChung, Kenneth Hoi Kan2009The application of High Throughput Experimentation (HTE) stirred vessels in the catalyst and pharmaceutical industries enable the parallel screening of potential formulations. Such tasks only require a fraction of the raw material and experimental time that are needed in conventional lab scale reactors. However their small sizes (~ 15-250 ml) contribute to a low Reynolds Number, Re, which results in poor mixing efficiencies. Together with unconventional geometries, i.e. lack of baffles and simple impeller designs, their fundamentals are not fully understood. The present study applies Particle Image Velocimetry (2-D PIV) and Planar Laser Induced Fluorescence (PLIF) techniques to a HTE scale stirred vessel (T = 45 mm, V = 72 ml) to determine the mixing behaviour. Three mixing strategies: centreline unbaffled (U), conventional baffled (B) and off-centre eccentric agitator (E) configurations, were investigated using a pitched blade turbine (PBT). Experiments were performed in the high transitional regime (Re \( \approx\) 6000) using distilled water as the working fluid. A uniform power input of P/V = 168Wm\(^{-3}\) was applied. A method based on multiple horizontal and vertical 2-D PIV measurements was used to reconstruct the 3-D flow field in each of these configurations, since the conventional 3-D PIV is unusable at this scale. It was found that the determination of turbulent kinetic energy (TKE) using the isotropic assumption was perfectly valid for (B), but will lead to a considerable underestimation in both (U) and (E). In addition to the three configurations, a square section vessel (S) (T = 41.5 mm, V = 72 ml) and regular vessel with a tilted impeller axis (T) were also studied. With a modified experimental procedure a log variance method for mixing time was applied using PLIF where all the usable pixels in an image were accounted for. Not only was (U) found to be highly inefficient, (B) also registered a slower mixing time due to a small amount of tracer being trapped behind the baffles, which makes the (E), (S) and (T) an even better choice in turbulent mixing. The use of (S) enabled the formation of a more compact HTE unit, also its trailing vortices were able to reach a height of y/H = 0.6, bringing more energy to the upper reaches of vessel, as visualised by angle resolved PIV measurements. The flow number of impeller and the amount of pseudo-turbulence were also determined and they agreed well with literature values. However in the laminar mixing regime using Polypropylene Glycol (PPG) as the Newtonian working fluid (μ = 0.4–0.8 Pa s, P/V = 0.6–5.5 kWm\(^{-3}\), Re = 5–35), at low Re values, (S) only managed a mixing performance comparable to the (U) configuration as the baffling effect of its four corners are less pronounced. However, mixing performance improves in (S) at higher Re values. (B) and (E) gave a comparable mixing performance, suggesting (E) should always be adopted for its viability in both turbulent and laminar mixing regimes. For gas-liquid mixing using air and water (P\(_G\)/V = 168 Wm\(^{-3}\), Q\(_G\) = 0.5–1.0 vvm), an image analysis algorithm was developed which enabled measurement of gas and liquid phase velocities separately. In addition to the PBT, a Rushton Disk Turbine (RDT) was also used (C = D = 0.33T). The power input required for the small mixing vessel to achieve complete gas dispersion was not achievable at the required gas flow rate; hence the experiments are carried out in the flooding regime. This had the advantage of clearly discernible differences between gas and liquid flow pattern for validation purposes. A new image algorithm was written to separate out bubbles in the imaging plane which transforms the in-plane bubbles into tracer particles. The local velocities of the gas phase are then obtained using the conventional cross-correlation technique. The results showed qualitative agreement with experimental observations of global gas phase flow patterns in literature.660.2832TP Chemical technologyUniversity of Birminghamhttps://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.519674http://etheses.bham.ac.uk//id/eprint/909/Electronic Thesis or Dissertation |