Working Party on Mixing of the European Federation of Chemical Engineering (EFCE-WPM)

VIII. News

New publication : contribution of the WP on Mixing to Chemical Engineering Research and Design - EFCE 50th anniversary issue
EUROPEAN FEDERATION OF CHEMICAL ENGINEERING WORKING PARTY ON MIXING Development, State of the Artand Future of the Scientific Field
J. Bertrand, M. Poux and J. Aubin

Activities of the Awards Committee within the EFCE - WPM.
At the WP Meeting held during the 10th European Mixing Conference in Delft on 4 July 2000 it was decided to inaugurate a price for Young Research Workers. Awards Committee within the EFCE-WPM received three nominations having very high scientific level and recommended the EFCE - WPM to grant this award to Dr. Giorgio MICALE who was nominated by Prof. Alberto Brucato - University of Palermo, Italy. This proposal was approved by WPM delegates. Next awards of the EFCE - WPM will be granted on the occasion of the 11th European Conference on Mixing in 2003.

At the Meeting held during the 11th European Mixing Conference in Bamberg (Germany) on 14 October 2003 it was decided to inaugurate a price for Young Research Workers. Awards Committee within the EFCE-WPM received two nominations and recommended the EFCE-WPM to grant this award to Dr. Joëlle AUBIN who was nominated by Dr. J. Bertrand - Polytechnique of Toulouse, France. Next Awards of the EFCE-WPM will be granted on the occasion of the 12th European Conference on Mixing in 2006.

Next Events
Fluid Mixing VIII International Conference will take place in London (UK) from 10 to 12th april 2006. It is organized by the IchemE an dthe King's College of London. See further details on www.fluidmixing8.org

The main next international event of the EFCE-WPM will be the 12 th European Conference on Mixing which will be held in Bologna (Italy) from 27 to 30 june 2006. All the informations are avalaible on the web site : http://www.aidic.it/mixing2006/

New EFCE-WPM Chairman.
Dr. J. Bertrand from Polytechnique of Toulouse, France was elected as a Chairman of the EFCE-WPM from January 1st, 2004. Prof. P. Ditl was approved as an Honorary Chairman of the EFCE-WPM and Dr. I. Foøt as an Honorary Secretary of the EFCE-WPM for life.



International Symposium on Mixing in
Industrial Processes - ISMIP4
14-16 May 2001 - Toulouse (France)

COOPERATIVE TEST: EXPERIMENTAL STANDARDS FOR MEASUREMENTS OF VARIOUS PARAMETERS IN STIRRED TANKS


M. KRAUME (1); P. ZEHNER (2)


Abstract. Stirred tanks are used for several operations in industrial practice. Numerous scientific papers have been presented in literature dealing with experimental results on these applications. Comparisons and valuations of these data often fail because geometric parameters, experimental conditions, and measurement techniques differ notably. Therefore, it can be observed that correlations derived on the basis of these experiments often show large discrepancies. In a cooperative test of nine German working groups different experiments were carried out in stirred tanks under completely standardised conditions. Thereby, common measurement techniques were examined in their reliability.

Résumé. Dans l'industrie, les réacteurs agités sont utilisés pour effectuer plusieurs types d'opérations. Il existe un grand nombre de documents scientifiques traitant des résultats expérimentaux liés a leur application. Il est souvent quasi impossible de comparer et de juger de la qualité de ces données car les parametres géométriques, les conditions expérimentales et les techniques de mesure different considérablement. On constate donc souvent de grosses divergences entre les corrélations bâties a partir de ces expériences. Dans une étude menée en coopération par neuf différents groupes de recherche allemands, diverses expériences ont été menées en réacteur agité dans des conditions completement standardisées. De cette façon, la fiabilité de certaines techniques de mesure habituelles a été examinée.

Key-words. Standardisation, power input, homogenisation, solids suspension, gas dispersion


1 INTRODUCTION

In industrial practice, stirred tanks are used for a variety of operations such as homogenisation of miscible liquids, dispersion of gas, mixing of immiscible liquids, and suspension of solid particles. Over the last decades numerous scientific papers dealing with experimental results on these applications have been presented. However, comparisons and valuations of these data often fail because geometric parameters, experimental conditions, and measurement techniques differ notably. In addition use of various tank sizes makes it more difficult to compare results. Therefore, it can be observed that correlations derived on the basis of these experiments often exhibit large discrepancies, e.g. for solids suspension1, 2. This statement is very important for practical engineering work as those correlations are used for design and scale-up of stirred tanks.

For that reason members of the German GVC-VDI working group on mixing carried out a standardisation of tank and stirrer geometry as well as measurement techniques. On this basis reliable experimental results were achieved and are now available for comparison. The data can be useful in particular

To build up a broad data basis standardised experiments were carried out by nine members of the German working party on mixing representing chemical industry, mixing companies, and research institutes (s. Tab.1).

Chemical companies
Bayer AG, Wuppertal Henzler
Bayer AG, Leverkusen Judat
BASF AG, Ludwigshafen Zehner/Haverkemper
Mixing companies
Ekato, Schopfheim Krebs
Stelzer, Warburg Kückelmann
Research institutes
DIL, Quakenbrück Knoch
TU Berlin, Institut für Verfahrenstechnik Kraume
Universität Dortmund, FB Chemietechnik Langer
FH Sachsen-Anhalt Liepe/Sperling

Tab. 1: Participants of the cooperative test.

In order to minimise potential errors and deviations, simple experiments as well as common stirrers and tanks were chosen. Still, for parts of the experiments distinctly different results were achieved, the most important of which are presented and discussed in this paper.


2 MATERIALS AND METHODS

2.1 Standard Stirred Tank and Media

Figure 1 shows the main features of the selected stirred tank. Vessel, baffles and stirrers were procured or built by each experimenter individually. To ensure comparability of results, main dimensions were provided with tolerances. A compromise was made when selecting the diameter of D = 0.4 m: On the one hand technical relevance of results increases with vessel volume, on the other it causes higher expenditure and costs.


Figure 1: Dimensions of standardised stirred tank and stirrers.
Rushton and pitched blade turbines were selected as they are suited for most applications involving liquids of low viscosities. The aim of this test was not to utilise optimised stirrer types and dimensions but stirrers of highest possible finishing accuracy which is given for plain-shaped turbines rather than for propellers.

All experiments were performed using water, air and glass beads. The latter (fractions of two different sizes) originate from the same production batch.


2.2 Measuring Programme and Techniques

Performed measurements included

No restraints were made for the power input measuring method. Instead, existing and largely different devices were used, such as strain gauges, shaft-mounted torquemeters, or even turntables.

Two different methods were applied for measuring mixing time for homogenisation. Firstly, decolourisation of an iodine-starch solution after addition of sodiumthiosulfate was used. This way the mixing process could be visually observed throughout the whole vessel and zones of insufficient mixing could be identified. Mixing time was determined when last streaks disappeared. Besides, electrical conductivity was measured by means of an exactly positioned probe. A certain amount of NaCl-solution was added and the conductivity signal recorded. From the concentration-time-curve the time required for a certain mixing quality (set to 95% for both methods) was derived. In order to rule out deviations due to varying adding locations both solutions were introduced close to the axis. For each set of operating parameters four decolourisation and ten conductivity experiments were suggested to enhance accuracy of statistic averages.

Suspension experiments were performed using glass beads of two different sizes (fraction 1: 0.15 -0.25 mm, fraction 2: 0.8 - 1.2 mm). Both the 1-s- and the 90% suspended slurry height criterion were employed. The 1-s-criterion is met when no particle remains stationary at the bottom of the vessel for more than 1 s, while the 90% suspended slurry height criterion requires particles to be suspended up to 90% of the liquid level. Application of both criteria is based on visual observation.

In gas dispersion experiments in addition to power input, gas hold-up and volumetric oxygen transfer coefficient were measured. For determination of the hold-up a U-tube was fitted to the tank. With increasing hold-up, liquid level in the vessel rises and so does level in the U-tube, where fluctuations are considerably less frequent than in the vessel itself. Hold-up is then calculated from level variation.

The oxygen transfer coefficient was determined by means of the dynamical method. Firstly, by introducing nitrogen the tank content was stripped of all oxygen. It was then sparged with air and the increase in oxygen concentration was monitored by a probe and recorded. When evaluating data the decreasing oxygen concentration of air as well as the inertia of the probe had to be taken into account. Only concentrations between 20% and 90% saturation were considered in order to exclude erroneous start-up and end effects.

Operating conditions were prescribed for experiments, always resulting in turbulent conditions (Reynolds numbers > 104). Data was collected and centrally evaluated and plotted.

3 RESULTS

All results are presented anonymously, i.e. without giving names of experimenters.


3.1 Power Input

Figure 2 shows the measured torques over stirrer speed for both stirrer types. On comparison of values it becomes apparent that especially below 0.1 Nm widespread scattering and systematic differences occur. Deviations of that kind are always to be expected when measured values only amount to 10% or less of the possible maximum of the gauge. Independent of this effect, above 0.1 Nm systematic errors arise, too.

Figure 2: Dependence of torque on stirrer speed.

These discrepancies become even more obvious when comparing power numbers Po, which are shown in Figure 3 as a function of stirrer speed N. Below approx. 100 min-1 practically no agreement was found. As this was expected (see above), these values were discarded for their lack of accuracy by all experimenters.
Figure 3: Dependence of power number on stirrer speed.


It has to be noted that values measured by different authors deviate considerably while variations within one run of measurements are usually smaller. Especially for the pitched blade turbine these deviations are intolerable. Therefore, stirrers from labs 1 to 5 were again investigated by experimenter 6 using his own tank and measuring device. For stirrer speeds above 100 min-1 averaged power numbers and their maximum deviations are shown in Figure 4. Here, values from all experimenters as well as results from author 6 are given. Obviously not only measuring methods alone lead to dissimilarities in results. Deviations in the results of author 6 have decreased to an extent where they can only be explained by slightly differing stirrer dimensions. When gauged by experimenter 6 stirrer diameters were found to differ from the required 125 mm by ±1 mm. Also, the assumed blade thickness of 2 mm was smaller for author 1 (1.7 mm) and was distinctly exceeded by author 5 (3 mm). It is a well known fact3, 4 that Po decreases with increasing blade thickness which is in agreement with measurements. The blade angle, too, often turned out to be smaller than expected.




Figure 4: Averaged power numbers and their maximum deviations.

To sum up torque measurements, it can be stated that torques below 0.1 Nm yielded considerable deviations. Systematic errors above 0.1 Nm resulted from differing stirrer dimensions on the one hand, and from differing measuring devices on the other.

Deviations in power numbers were more pronounced for pitched blade turbines than for Rushton turbines, where they differed by 15% and 10%, respectively.

3.2 Homogenisation

With regards to mixing times, averaged results (decolourisation: 4 measurements, conductivity: 10) for both stirrers are plotted as dimensionless products N tMIX, the dimensionless mixing time, over stirrer speed in Figure 5 and 6. Surprisingly good agreement of values from different authors was found for both methods. In the following Figures 7 and 8 will therefore only be distinguished by the respective method.

Although a scattering of 10 to 20% could be observed, both methods resulted in reasonably similar mixing times. It is assumed that this is due to mixing taking place evenly throughout the vessel, thus enabling the locally limited conductivity measurement to correctly represent the homogenisation process in the vessel as a whole.

Different accuracies of results have to be noted for the two stirrer types. Scattering of results was wider for the pitched blade turbine. In this case, also a small distinction between the two measuring methods prevailed, the reason of which could not be satisfactorily explained.

The authors strictly keeping to the required adding location was of major importance for the overall good agreement. In one exemplary investigation the influence of a 150 mm shift from the axis


Figure 5: Mixing time characteristic of Rushton turbine.

was observed to cause a 30% increase in mixing times. Inaccurate adjustment of excess concentration was identified as another influential source of errors. Since dosage of amounts is never absolutely correct, this excess concentration is bound to be faulty. The true excess concentration, however, can be easily determined by titration and related to a mixing quality of 95%5.



Figure 6: Mixing time characteristic of pitched blade turbine.


3.3 Suspension

The measured stirrer speeds for solids supension exhibit a largely varying degree of agreement. Determining the state of suspension by use of the 1-s-criterion yields similar results for the pitched blade turbine (compare Figure 7). According to Figure 8, however, judging the point where the 1-s-criterion is reached becomes difficult when small particles are to be suspended by use of a Rushton turbine. The region just below the stirrer is visually inaccessible, so a clear decision whether these relatively small particles simply perform a sliding movement or whether they indeed get lifted upwards within 1 s becomes impossible. Obviously individual interpretations of the 1-s-criterion differ as results from each author are quite consistent in themselves. A uniform judgement seems to be easier for larger particles since deviations are distinctly reduced.




Figure 7: Critical stirrer speed for complete suspension of 1 mm glass beads with the pitched blade turbine.

When employing the 90% suspended slurry height criterion smaller disagreement between measurements was found for the pitched blade turbine than for the Rushton turbine. Results differ especially when the 90% suspended slurry height is reached before the 1-s-criterion which is the case for small particle concentrations. Under these circumstances, some particles indeed rise to a height equivalent to 90% of the liquid level while a reasonably large fraction of solids still remains at the bottom. Therefore, the critical stirrer speeds are almost independent of particle concentration, which is especially striking for particle sizes of 1 mm.

Tab. 2 summarises discrepancies between all measured stirrer speeds. As the corresponding power inputs will differ even stronger (increase proportional to N3) this is a quite sobering result. Still, these data make the large disagreement between literature correlations and scale-up rules more comprehensible.



Figure 8: Critical stirrer speed for complete suspension of 0.2 mm glass beads with the Rushton turbine.



Tab. 2: Deviations of critical stirrer speeds for solids suspension.


3.4 Gas Dispersion

In contrast to single-phase measurements differences between power inputs are more pronounced, as shown in Figure 9 for the pitched blade turbine. Since stirrer speed as well as gas flow rate were varied, for simplified graphical presentation only a comparison of data with those of author 6 is used for plotting results. One possible reason for the discrepancies observed might be an inaccurate measurement of gas flow rate as in for constant stirrer speed it is practically the only quantity on which torque depends.

Figure 9: Comparison of gas holdup measurements for the pitched blade turbine.


The selected method for measuring gas hold-up turned out to be insufficiently exact. In spite of the U-tube's dampening effect the liquid level fluctuates considerably - if not as much as inside the vessel - thus making observer independent level measurement hardly possible. As a result, drastically differing gas hold-ups were determined, as shown in Figure 10 for the pitched-blade turbine. Especially hold-ups



Figure 10: Comparison of torque measurements with the gassed pitched blade turbine.
of less than 1% should be regarded with care. At higher hold-ups dissimilarities get reduced but still leave deviations of 20% and more. Again, systematic errors can be observed. The measuring method is not suited for obtaining reproducible results, a fact, however, which is of minor importance in technical applications. On the one hand, calculation of liquid content is affected only to a small extent by gas hold-ups below 10 %, above which determination seems to become sufficiently accurate. On the other hand the main aspect of gas/liquid-systems is mass transfer, characterised by the mass transfer coefficient.

Results from mass transfer measurements are summarised in Figure 11. In this plot suggested by Henzler6 a dimensionless mass transfer coefficient is shown as a function of a dimensionless power input. With the exception of results from author 4 all data appear to be well bundled, especially considering that the power input, a parameter strongly subjected to errors, is used on the abscissa. Plotting the mass transfer coefficient versus stirrer speed for the respective superficial gas velocities does not yield any better agreement.
Figure 11: Dimensionless mass transfer coefficient as function of dimensionless power input.




4 CONCLUSIONS

On the basis of identical experimental situations:

an impression on the accuracy of measurements was gained with the presented cooperative tests. In fact for part of the experiments results are widely scattered. It can be assumed that literature data commonly contain similar deviations. Differences mainly arise from the following reasons:


This again emphasises that even experiments carried out by experienced experimenters are subject to uncertainties and therefore have to be critically judged, especially in a situation where the actual daily work leaves little room for careful performance of measurements.


LITERATURE


[1] EKATO Rühr- und Mischtechnik GmbH (2000): EKATO Handbook of mixing technology, Schopfheim.

[2] Kraume, M.; Zehner, P. (1988): Suspendieren im Rührbehälter - Vergleich unterschiedlicher Berechnungsgleichungen, Chem.-Ing.-Tech. 60, No. 11, pp. 822/829

[3] Bujalski, W.; Nienow, A.W.; Chatwin, S.; Cooke, M. (1987): The dependency on scale of power numbers of Rushton disc turbines, Chem. Eng. Sci. 42, No. 2, pp. 317/326

[4] Rutherford, K.; Mahmoudi, S.M.S.; Lee, K.C.; Yianneskis, M. (1996): The influence of Rushton impeller blade and disc thickness on the mixing characteristics of stirred vessels, Trans. IChemE, Vol. 74, Part A, pp. 369/378

[5] Henzler, H. (1978): Untersuchungen zum Homogenisieren von Flüssigkeiten oder Gasen, VDI-Forschungsheft 587, Düsseldorf

[6] Henzler, H. (1982): Verfahrenstechnische Auslegungsunterlagen für Rührbehälter als Fermenter, Chem.-Ing.-Tech. 54, No. 5, pp. 461/476

back to main EFCE - WPM index