Author’s Accepted Manuscript

Experimental and Computational Study of a High

Speed Pin Mixer via PEPT, Visualization and CFD

Anne K. Konz, Erich Windhab

www.elsevier.com/locate/ces

PII: S0009-2509(16)30429-8
DOI: http://dx.doi.org/10.1016/j.ces.2016.08.007
Reference: CES13104
To appear in: Chemical Engineering Science
Received date: 19 January 2016
Revised date: 15 June 2016
Accepted date: 5 August 2016
Cite this article as: Anne K. Konz and Erich Windhab, Experimental and
Computational Study of a High Speed Pin Mixer via PEPT, Visualization and
C F D , Chemical Engineering Science,
http://dx.doi.org/10.1016/j.ces.2016.08.007
This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal pertain.
Experimental and Computational Study of a High

Speed Pin Mixer via PEPT, Visualization and CFD

Anne K. Konza*, Erich Windhabb1

a
Laboratory of Food Process Engineering, Institute of Food, Nutrition and Health, ETH

Zurich, Schmelzbergstrasse 9, CH-8092 Zurich, Switzerland

b
Laboratory of Food Process Engineering, Institute of Food, Nutrition and Health, ETH Zurich, LFO E18,
Schmelzbergstrasse 9, CH-8092 Zurich, Switzerland.

*
Corresponding author. Phone (mobile): +41 77 418 47 17. E-mail address: [email protected]. Present
address: Oleanderstrasse 14, CH-8050 Zurich, Switzerland

Abstract

Continuous high speed pin mixers have been successfully implemented industrially to mix high fractions of

powders into highly viscous non-Newtonian liquids. However, they have not been described in the literature

yet. This work presents a multi-method approach, consisting of conventional torque and throughput

measurements to calculate average residence times and dimensionless Reynolds/power numbers for

suspension mixing, Positron Emission Particle Tracking (PEPT) to gain information on particle trajectories

and velocities in the investigated opaque model system, high speed film visualization and CFD simulations.

By the combination of these methods, valuable information on the influence of variable process parameters,

particularly on the influence of pin configuration, pin shape and inclination of semi-cylindrical pins on

mixing mechanisms and efficiency, could be gathered for a broad range of viscosities and powder fractions in

Newtonian and non-Newtonian fluids. Flow patterns such as a “Split-and-Recombine” convection pattern at

the pins, axial mixing mechanisms and beneficial process conditions to reduce particle sedimentation caused

by centrifugal forces in the mixer were retrieved from the results. The PEPT methodology was successfully

implemented for a high-shear continuous mixing process of suspensions for the first time. As the combined

methodology shows reasonable and reliable results that were also applicable at industrial scale, the work

1
Phone: +41 44 632 53 48, Fax: +41 44 632 11 55, E-mail address:
[email protected].
1
provides sound basis for further research development of measurement techniques for other mixing

processes, especially in opaque systems.

Keywords: Pin mixer, PEPT, CFD, methodology, solid-liquid mixing, opaque systems

1 Introduction

Continuous mixing processes have progressively replaced common used batch processes in

several industrial fields, such as the pharmaceutical, chemical, biotechnology and

construction industries. This development is driven by several combined advantages of

continuous processes, such as being compact with short mixing paths at high throughput

rates, saving resources for interim storage, consuming less space and other resources and

allowing to establish continuous process chains (Patel et al., 2012; Portillo et al., 2009;

Weinekötter and Gericke, 2000).

For both batch and continuous mixers, thorough understanding of their influence on the

operation conditions and further on the process mechanisms that determine product

properties and quality is required to successfully adapt the parameters to each new mixing

task and for scale-up.

Parameters typically relevant in continuous mixing processes are: Mixer geometry, (if

variable) configuration of the rotor, rotation speed, material throughput rate and the

feeding and properties of mixing components. The effects of the parameters mentioned

previously have been studied by several authors (e.g. Florian et al. (2014)). In particular,

the effect of the impeller geometry as a variable suitable for optimization has been

emphasized by Gao et al. (2011a).

However, information on the influence of parameters is often lacking due to the large

variety of mixers. Furthermore, a transfer of empirical knowledge is seldom possible since

2
systematic design methodologies and control strategies are limited in their development

(Gao et al., 2011b).

Kingston and Heindel (2014) however give rise to concern about the complexity in the

design of experiments and the high amount of resources needed to collect, analyze and

interpret data if too many parameters are incorporated and investigated simultaneously.

Previous experience is therefore beneficial to identify the most influential parameters for

effective research on important interactions. As a further valuable approach, the gain in

computational power allows the flow mechanisms to be accessed via simulations and to

predict results to restrict efforts for large series of experiments. However, they may be

limited to estimations depending on the complexity of the mixing system.

For continuous mixers, agitation is often provided by blades/impellers mounted on the

driving shaft, causing convection as a primary source of mixing (Portillo et al., 2008).

Values for convective, shear and deagglomeration forces can be derived from the mixer

geometry and the rotation speed. Power measurements or direct monitoring of the torque

exerted by the rotor during steady-state mixing operation allow to calculate the specific

energy impacting on the processed materials. Their throughput rate and mass hold-up in

the mixer enable calculation of the average residence time and also efficiency evaluations

for configurations that result in comparable product quality.

The Metzner-Otto concept (Metzner and Otto, 1957) provides an approach for the

dimensionless characterization of flow behavior, leading to a representative shear rate that

can be utilized for scale-up and comparison of mixer types. This approach is also suitable

for non-Newtonian fluid systems, but however sometimes limited in applicability.

Complex mixing mechanisms or geometrical parameters changing over time cannot always

be included in the model equations. In the centrifugal high speed mixer presented in this

work, the non-constant immersion depth of the pins into the suspension and air-

entrainment impaired a full Metzner-Otto characterization.

3
Noninvasive methods provide information on particle or fluid element trajectories, velocity

fields, residence time distributions, flow mechanisms in different regions of the mixer

geometry and on mixing quality by analyzing the auto-correlation of particle movement.

Optical tracers/tracer fluids can be applied in a transparent assembly and with transparent

fluid systems, such as with PIV. Opaque systems require the use of different tracing

principles such as LDV, ERT and also the PEPT methodology used in this work.

Invasive methods allow to determine the influence of operation parameters on defined

properties of the mixture connected to product quality and to calculate residence time

distributions by collecting and analyzing samples with or without marker substances.

However, the high rotation speed and low wall-clearance of the mixer characterized in this

work hindered on-line sampling of the materials and was therefore not put into focus.

Previous work on mixers with adjustable rotor extensions mainly relates to powder mixers

with paddles with variable configurations of paddles and their inclination towards the axial

flow direction, and how the several process parameters influence mixing efficiency. One or

more often a set of methods is used to reach this goal, with some of the studies given in the

following on mixer types comparable to the one investigated in this work.

Yabushita et al. (1989) investigated the influence of pin sections in a barrel screw extruder

for a rubber compound. PEPT was used by Laurent and Bridgwater (2002a) to describe the

influence of different agitator designs in horizontal powder flow and by Barigou et al.

(2003) to gather information on concentric pipe flow regimes in solid-liquid food

suspensions at volume fractions up to 55 %. Eesa and Barigou (2008) compared the

horizontal laminar flow profiles of non-Newtonian suspensions gained by PEPT with CFD

simulations and found a very good agreement - a similar approach was implemented in our

work.

Pitched blade turbines were studied by Pianko-Oprych et al. (2009) who compared results

of PEPT and PIV measurements for suspensions (5 % w/w) and by Liu and Barigou

4
(2013). They modeled the velocity field and phase distribution of dense monodisperse

solid-liquid suspensions and compared them with PEPT experiments, finding a high degree

of agreement for solid concentrations of up to 20 % (w/w).

Pakzad et al. (2013) compared CFD and ERT results for non-Newtonian batch reactor

mixing with different types of impellers.

To our knowledge, only one multi-method approach for a centrifugal mixer using PEPT

has been presented by Portillo and coworkers (2008, 2009; 2010), but in that case

pharmaceutical powder paddle mixers at much lower rotation speeds were operated. They

used PEPT and NIR respectively to investigate particle trajectories at different operation

settings to assess the mixing quality of samples and found appropriate paddle inclinations

to increase backflow and enhance mixing.

The multi-method approach presented here is used to investigate the influence of operating

parameters on mixing, to our knowledge describing a continuous high speed pin mixer for

the first time. Information on the characterization of this mixer type is unavailable in the

literature, although it has been established to mix solids into highly viscous fluids for

industry-relevant production scales and compositions. As well, the PEPT methodology has

not been described yet for its application on continuous solid-liquid mixers, particularly at

the high rotation speeds given (Barigou et al., 2009; Guida et al., 2010,2011; Edwards et

al., 2009; Perez-Mohedano et al., 2015). It was utilized to reveal flow mechanisms in the

opaque mixer-suspension system. High speed filming for visualization and the comparison

of experimental results with CFD calculations were further methods applied in our

approach.

The influence of variations of pin configuration and particularly pin shape (and inclination

for semi-cylindrical pins) on mixing enhancement and on the local flow around each pin

was characterized. Since pin configurations with semi-cylindrical pins were found to

5
improve mixing and reduce particle sedimentation at the mixer wall, a specific solution

approach for problems arising from the centrifugal demixing tendencies is additionally

addressed in an excursus. The applicability and comparability of the used methods for

certain research aspects, particularly opaque, industrial fluids that are not otherwise

amenable to PIV or LDV measurements, was evaluated. This offers important information

to the research of mixing.

6
2 Experiments

2.1 Pin mixer design

2.1.1 General

Fig. 1 a Schematic of the driving shaft with designed pin arrangement, inlet and outlet positions and

designation of axial zones. b The high speed pin mixer during operation: Housing and driving shaft

assembled with an arrangement of 100 % cylindrical pins (configuration 1, Table 2), powder inlet from the

back, fluid inlet from the front (shifted inlet position not used in the experiments), temperature sensors at the

top, view of the Perspex visualization window. c Photograph of the driving shaft with cylindrical pins

mounted (50 % in the front zone, 100 % in the back zone)

Fig. 1 shows the continuous high speed pin mixer used in this study which was designed

and manufactured as a downscaled pilot apparatus, being geometrically similar to a pin

mixer utilized for the industrial production of powder-filled polymers. For optical
7
explorations of the flow mechanisms near the mixer wall, the pilot mixer is additionally

equipped with a Perspex window providing view along the whole mixer axis (Fig. 1b). The

mixer housing comprises inner dimensions of 321 mm in length and 90 mm in diameter and

contains cooling channels for maintaining constant temperature conditions during

experiments.

The fluid inlet positioned at an axial position (left to right) of 21 mm and entering laterally

from the front is fed by an eccentric screw pump with a volumetric flow rate freely

adjustable up to 1000 L/h. The powder inlet positioned at an axial position of 71 mm and

entering laterally from the opposite back side of the mixer is fed by a single-screw powder

dosage unit. Via a funnel, the mixer outlet can be connected to the fluid pump to enable

circulation of fluids and suspensions.

On the driving shaft with a diameter of 45 mm, up to 128 pins can be assembled on two

spirals (27.31° inclination) with the second spiral solely covering the posterior 192 mm of

the mixer axis (Fig. 1a, c). The motor drives the shaft and attached pins with a freely

adjustable rotation speed of up to 3000 rpm which (with 1 mm wall clearance at 3000 rpm)

generates a pin tip speed of 13.82 m/s and shear rates of up to 13820 1/s.

2.1.2 Parameters/Configurations

To characterize the flow behavior of solid-liquid systems under the high shear conditions

in the pin mixer, the following parameters were varied: mixer rotation speed (0-3000 rpm),

fluid or suspension volumetric flow rate (0-1000 L/h), powder mass flow rate (affecting the

powder fraction) and the pin configuration. The inlet positions of powder and fluid

described previously were used, having advantageous effects.

Two pin shapes (Fig. 2) were mainly used in the experiments: A pin with cylindrical shape

with a related pin-wall gap width of 1 mm, and a semi-cylindrical pin with a related gap

width of 1 mm. The inclination of the semi-cylindrical pin was adjusted to inclinations of

8
-45° or +45° (Fig. 2 c), influencing the axial flow behavior.

Fig. 2 a Schematic view of the cylindrical pin (top: 2D drawing, bottom: 3D CAD representation).

b Schematic view of the semi-cylindrical pin (top: 2D drawing, bottom: 3D CAD representation).

c Inclination angles of the semi-cylindrical pins relative to axial flow direction

Furthermore, the pin fraction was varied and set to either 25 %, 50 % or 100 %. A

differentiated pin assembly for the front zone with only one pin spiral and the back zone

with two pin spirals was only investigated for experiments on the reduction of particle

sedimentation (section 4.2).

2.2 Preparation of model solid-liquid compositions with diversified properties

2.2.1 Composition of the model fluid

Two types of fluid were used: Pure glucose syrup (DE 60, BC*Sweet 01257, density:

1400 kg/m3; Blattmann Schweiz AG, Waedenswil, Switzerland; further on abbreviated as

9
“GS”) was chosen to investigate Newtonian flow behavior in the mixer in PEPT

experiments and in high speed filming visualization experiments.

A non-Newtonian, shear-thinning model fluid (“MF”) was developed to emulate mixing

properties of highly viscous polymers mixed with filling powders relevant for industrial

applications of the high speed mixer. The model fluid consisted of 71.4 % (w/w) glucose

syrup and a 2 % (w/w) aqueous solution of Carbopol solved in water (28.1 % (w/w), Zurich

tap water for usage in high amounts); Carbopol© 0.57 % (Ultrez 20; Lubrizol, Brussels,

Belgium, resulting pH in the optimal application range of 4-5).

2.2.2 Composition of solid-liquid combinations

Several powder fraction levels of bentonite (montmorrilonite clay mineral, described by

Luckham and Rossi (1999)) (Wyoming Greendbond 453; Albert Isliker und Co. AG,

Zurich, Switzerland; laser diffraction particle size X90,3 = 53 μm with a large fraction of

agglomerates; bulk density ca. 800 kg/m3) were combined with the MF either separately by

powder dosage or in a raw suspension premix for continuous circulation of the resulting

suspensions (Table 1).

Table 1 Composition and application of the combinations of several bentonite fractions with the model fluid

Name Constituents Powder Use in experiments

(w/w) addition
“MF+5%-Bent.” 95 % MF premix Parameter studies (3.1)
5 % bentonite
“MF+6%-Bent.” 94 % MF premix PEPT experiments (3.2),

10
 6 % bentonite CFD simulations (3.4)
“MF+30%-Bent.” 70 % MF dosed PEPT experiments (3.2),
30 % bentonite separately CFD simulations (3.4)
variable MF + stepwise dosed Sedimentation excursus
increased separately (4.2)
bentonite fraction

A powder fraction of 5 % (w/w, density: 1200 kg/m3) was chosen for the mixer’s parameter

studies. The “MF+6%-Bent.” premix represented the composition with the highest

viscosity still allowing continuous circulation through the mixer. The “MF+30%-Bent.”

composition was developed to mimic an industrially produced powder-filled polymer

composition (“IC”) produced with the upscaled version of the high speed pin mixer (Fig.

3b). The addition of a large amount of fine powders (i.e. bentonite) into the model fluid

lead to resistance to the flow, thus increase in apparent viscosity at a given shear rate.

The PEPT experimental conditions using pure GS and the MF+6%-Bent. and MF+30%-

Bent. solid-liquid systems were reproduced by CFD simulations as well. The flow behavior

of the IC was used for CFD simulations relating to the influence of semi-cylindrical pin

inclinations.

11
2.2.3 Rheology

Fig. 3 a Viscosity data of GS, MF, MF+5%-Bent. composition, MF+6%-Bent. composition, and MF+30%-

Bent. composition at 30 °C; Power law fit constants: GS: K=5.426, n=0.97; MF: K=0.289, n=0.557; MF+5%-

Bent.: K=15.482, n=0.212; MF+6%-Bent.: K=15.570, n=0.140; MF+30%-Bent.: K=647.0, n=0.241.

b Viscosity functions of the MF+30%-Bent. composition and the industrial composition at 20 °C

Viscosity data of the fluids and compositions investigated in the experiments at 30 °C and

their constants fit to the Power law equation η = Kγn-1 are shown in Fig. 3a. A comparison

of MF+30%-Bent. with the IC at 20 °C is shown in Fig. 3b.

The viscosity measurements were performed with a stress-controlled Physica MCR 300

rheometer (Physica Messtechnik GmbH, Stuttgart, Germany). A concentric cylinder

geometry (CC27, Searle type) was used except for the measurements with 30 % powder

12
fraction (PP25 plate-plate geometry). A preshear period of 1 min at 1000 s-1 and then a

shear rate profile was applied (increasing from 0.01 to 1000 s-1, then decreasing from 1000

to 0.01 s-1 and utilizing the resulting viscosity values from the backward measurement as

no hysteresis was observed).

2.3 Experimental methods

2.3.1 Parameter studies on the influence of configuration, rotation speed and

throughput rate on net torque and average residence time with the MF+5%-Bent.

composition

The mixer was equipped with a torquemeter and connected to a computer running a

LabView program customized to monitor and record the torque directly at the mixer

driving shaft (temporal resolution: 70 ms, continuous averaging over 50 values to reduce

fluctuations).

Pin configurations 1, 2, 3, 4 and 5 were used respectively, listed in Table 2. At steady-flow

circulation conditions at 30(+-3)°C, three 60-second measurement runs were conducted

and the suspension’s mass hold-up in the mixer gathered and determined gravimetrically.

The net torque during circulation of MF+5%-Bent., average residence times and average

wall layer thickness for each configuration, rotation speed and throughput rate were

calculated accordingly. The parameters considered in this study and their respective

importance are shown in Table 3.

Table 2 Configurations used and details for the assembly in the front and back zone (Fig. 1a)

C. No.a Front & back (main mixing) zone

Pin fraction Gap widthb Pin shapec Inclination angle
% mm °
1 100 1 C -d

13
2 50 1 C -d
3 25 1 C -d
4 25 1 SC +45
5 25 1 SC -45
a b c
= Configuration Number; : between pin and wall; : C = cylindrical, SC = semi-cylindrical;
d
: - = no inclination (cylindrical pins)

Table 3 Parameters varied in the study and their respective effects

Parameter Range Effect
Rotating rate 0 - 3000 rpm Affect the energy input, and slurry
homogeneity
Volumetric flow rate 0 - 1000 L/h Affect residence time, slurry homogeneity
and throughput
Powder mass flow rate N/A Affect residence time, mixing homogeneity
and throughput
Pin configuration N/A Affect power consumption and mixing
homogeneity

2.3.2 PEPT

PEPT experiments were conducted at the Positron Imaging Centre, University of

Birmingham, UK. In this study, 18F-marked resin particles (Fan et al., 2006) with a

diameter of 100 m (density: 1100 kg/m3) were used and tested to be neutrally buoyant in

all investigated fluids and suspensions (similar density). Table 4 provides the combinations

of composition, configuration, passage mode (circulation or single passage) and measured

temperatures of the experiments.

14
Table 4 Configurations (Table 2), operation modes and measured temperatures for the three model

compositions investigated in the PEPT experiments

Composition C.a Throughput rate Passage mode Temperature

GS 1 100, 400, 700, 1000 L/h circulation 39° C
MF+6%-Bent. 1,2 400, 700, 1000 L/h circulation 24-25° C
MF+30%-Bent. 2 120 kg/h single 27-33° C
a
= Configuration

The rotation speed was set to either 500, 1000, 1750 or 2500 rpm.

The calculation settings for the trajectory calculation algorithm (Parker et al., 1993) were

set to fopt=20 (optimization parameter), Nevent=75 (decay events per generated data

point) and Nslice=5 (overlap parameter of data frames).

Data were further processed with Excel and MATLAB.

2.3.3 High speed film visualization

High speed films with a frame rate of 30 000 frames/s were recorded with a Memrecam

fx RX-6 (NAC Image Technology, Simi Valley, USA) through the window on the upper

side of the pilot scale pin mixer. The rig also consisted of an intensive 1000 W lighting

assembly.

2.3.4 Simulations

The following conditions were used for the simulations: The geometry of the pilot pin
mixer was replicated with a mesh size of 0.6-2.6 x 106 hexahedral cells. The gap between
the pin and the wall has a resolution of 9 cells over the height. Density and viscosity data
as previously presented and a reference pressure of 1 bar were used and laminar and non-
buoyant behavior assumed. The simulations were performed in a steady-state mode
(relative frame), with a counter-rotating outer wall. All rotor walls and the pins have a no-
slip condition. Using a step function at the inlet, the gas/liquid fraction was imposed
depending on the radius and according to the operating point mass flow. In the outer part
of the inlet, Bentonite/MF was introduced, in the center air.

15
The governing equations used in Ansys CFX are the Free Surface Model for the interphase
transfer (VOF-Model) and the incompressible Navier-Stokes equations for solving the flow
conditions. The Shear Stress Transport Model (SST) was used to describe the turbulence.
The computation grids are show in the Fig. 4.

Fig. 4 a Computational grids for the rotating impeller

3 Results

16
3.1 Parameter studies

3.1.1 Net torque with 5 % (w/w) bentonite suspension and calculated average

residence times

Fig. 5 a Net torque for configurations 1-5, combinations of three rotation speeds with two mass throughput
rates (MF+5%-Bent. suspension). b Average residence time, calculated from the mass hold-up gathered from
the mixer. Configurations: 1: 100 % cylindrical pins, 2: 50 % cylindrical pins, 3: 25 % cylindrical pins,
4: 25 % semi-cylindrical pins with +45° inclination, 5: 25 % semi-cylindrical pins with -45° inclination; all
with 1 mm pin-wall gap width

Fig. 5a shows the net torque values during circulation of the non-Newtonian MF-5%-Bent.

premix. A definite dependence of the net torque values on the rotation speed was not

found. Despite the low pin fraction of 25 %, the net torque values of configuration 5 mostly

exceed those of configuration 1-4. The difference in the pins’ shape in configurations 3, 4

and 5 shows a decreasing effect on the net torque values for semi-cylindrical pins with

+45° inclination relative to the axial flow direction (configuration 4) and an increasing
17
effect for semi-cylindrical pins with -45° inclination (configuration 5). A pin inclination of

-45° is expected to induce axially backwards oriented motion of fluid compartments,

thereby leading to a higher drag on the pins and a higher net torque, with the opposite

effect for a pin inclination of +45°.

The average residence times as calculated from the respective mass hold-up gathered from

the mixer are depicted in Fig. 5b. (An exceedingly high torque and emergency stop averted

the measurement at the combination of 3000 rpm and 100 kg/h for configuration 5.)

The suspension mass retained in the mixer (values proportional to the average residence

times) strongly depends upon the mass throughput rate. This is also reflected by the

observed suspension consistency, showing a higher viscosity and mixing grade for the

lower mass throughput rate whereas the higher mass throughput rate resulted in a very

fluid consistency with lumps of powder poorly mixed into fluid. The retained mass values

range from 75 to 1250 g. The average suspension layer thickness at the mixer wall as

calculated from the mass retention values ranges from about 1 to 11 mm, suggesting that a

sufficient immersion depth of the pin tips into the suspension mass is provided at a pin-

wall gap of 1 mm.

The obtained average residence time values range from 1 to 45 s, suggesting a low overall

mixing intensity for a mass throughput rate of 250 kg/h and a close correlation of lower pin

fraction and prolonged average residence time.

The strong effect of the pin shape on the time available for mixing is confirmed, especially

at an inclination of -45° by intensifying axially backwards oriented suspension flow. A

high net torque working on the suspension combined with a high average residence time as

given in configuration 5 indicates best prerequisites for intensive mixing forces exerted on

the suspension, particularly at low mass throughput rates such as 100 kg/h.

18
3.1.2 Reynolds and Power numbers

During the preparations for the parameter studies with the MF+5%-Bent. suspension,

torque measurements were recorded to develop a description of the mixer following the

Metzner-Otto concept (Metzner and Otto, 1957) by calculation of a representative shear

rate and a mixer constant, using glucose syrup as a Newtonian fluid. Several deviations

from the included general assumptions hindered this full description, mainly local

temperature deviations by friction between pin and wall and air entrained which influenced

the overall fluid density. Both factors lead to changes in viscosity that could not be

accounted for in the rheological viscosity measurements. Additionally, the wavy fluid

surface caused the pin-dipping length and thereby the effective rotor diameter to vary

constantly.

19
Fig. 6 a Reynolds-Power number plot for configurations 1-5 (Table 2), combinations of three rotation speeds

at 100 kg/h throughput rate (MF+5%-Bent. suspension). b Reynolds-Power number plot for configurations

1-5 (Table 2), combinations of three rotation speeds at 250 kg/h throughput rate (MF+5%-Bent. suspension)

However, the torque data for the MF+5%-Bent. suspension (section 3.1.1, viscosity

function in Fig. 3a) were used to calculate Reynolds and Power numbers for all

investigated parameter combinations, shown in Fig. 6a for 100 kg/h throughput rate and in

Fig. 6b for 250 kg/h throughput rate.

The estimated Reynolds numbers indicate there was mixing in a turbulent regime. The

Power numbers are in a very low range of 10-2 to 10-5. Their span is narrower for a rotation

speed of 1000 rpm for both throughput rates. For rotation speeds of 2000 rpm and

particularly 3000 rpm, the relatively low values for configuration 4 with an assembly of

25 % +45°-inclined pins are assumed to be caused by the low filling degree and low

residence time in the mixer due to increased downstream velocity. The Power number
20
values for configuration 5 with an assembly of 25 % -45°-inclined pins, which effectuates

a high filling degree and residence time by a higher recirculation, are similar to

configuration 1 with 100 % cylindrical pins. This indicates a similar or even higher energy

impact of configuration 5 despite a 4-fold lower compared to configuration 1.

3.2 PEPT

Fig. 7 3D trajectory of the tracer particle transported in GS. The color scale indicates the residence time with

the tracer particle entering the mixer from the pump at the right lower end at 9.1 x 105 ms and leaving the

mixer at 9.8 x 105 ms, resulting in a residence time of 7 s; conditions: configuration 1, rotation speed:

2500 rpm, throughput rate: 1000 L/h

Velocity data, residence times, occupancy data and 2D as well as 3D trajectories were

derived from the PEPT data and were further evaluated. Fig. 7 shows a 3D trajectory of the

tracer particle transported in GS. It is evident that a spiral movement near the mixer wall is

produced. Portillo et al. (2010) obtained similar dense-spaced particle trajectories in a

continuous powder mixer equipped with blades for low rotation speeds. The latter

obviously had the same effect of a relatively low tracer particle velocity like GS as

transporting fluid in our PEPT measurements.

21
All data were evaluated respective to their median values to avoid a strong influence of

outliers, caused by the tracer particle being stuck during its movement through the mixer.

For the mixing of 30 % (w/w) bentonite fraction into MF, as the particle had to be

recollected and added manually, only 10 trajectories could be received that only allowed

for estimated representative values.

3.2.1 Axial velocity frequency distributions and median axial velocities

Fig. 8 Axial velocity distributions for the tracer particle movement a in GS for pin configuration 1,

combinations of three rotation speeds with three fluid throughput rates; b in MF+6%-Bent. suspension for pin

configuration 1, for combinations of two rotation speeds with three suspension throughput rates. Both

quantified as given by the fraction of data points from cumulative trajectory information

22
Figs. 8 a and b give axial velocity distributions for the Newtonian GS fluid and the non-

Newtonian MF+6%-Bent. suspension, respectively, for several combinations of throughput

rates and rotation speeds. Most plausibly caused by the pins’ action to split the mass in

front of the pins, a high share of axial backward movement is measured and the velocity

values are low. As expected, the higher the throughput rate and/or the rotation speed, the

higher the velocities. The axial velocity frequency distribution for the Newtonian GS is

narrower and suggests that the tracer particle moved slowly and steadily very close to the

wall.

3.2.2 Total (x-y-z components combined) velocities and their radial distribution

The radial distribution of total velocities between mixer wall and driving shaft, collected

from all recorded trajectories, is shown in Fig. 9a for GS and in Fig. 8b for the MF+6%-

Bent. composition. The median velocity depending on the radial position indicates lower

values of median velocity (ca. 0.13 m/s) for the Newtonian GS (viscosity: 4.1 Pa s)

particularly near the mixer wall, whereas the values for the non-Newtonian MF+6%-Bent.

composition (viscosity: 5 x 10-3 Pa s) are twice as high as for GS. The faster and more

discontinuous axial and total transport found for the bentonite suspension with its lower

viscosity is reflected in the more uniform and higher velocities also near the mixer wall.

23
Fig. 9 Total velocity of the tracer particle movement vs. its distance from the mixer wall for pin

configuration 1, rotation speed 2500 rpm, fluid throughput rate 1000 L/h a in GS; b in MF+6%-Bent.

suspension

3.2.3 Median residence times

Selected results of median residence times from the PEPT experiments are included in

Table 4 (section 3.5) and compared to results from the parameter studies and simulation

calculations. They range about 10-fold higher than the simulation values for the MF+6%-

Bent. and MF+30%-Bent suspensions and even about 100-fold higher for GS which is

mainly attributed to the comparatively large tracer particle size of 100 μm. This reason for

the deviation of values and other possible causes are discussed in section 3.5.

24
3.3 High speed film visualization

Fig. 10 a “Split-and-Recombine” flow pattern visualization captured from a high speed video sequence

(20 000 frames/s, GS at 30 °C, rotation speed: 2000 rpm). b Schematic showing two subsequent pins, the

velocity vectors and the fluid flow (from right to left side), split behind the first pin, partly recombining

behind it and partly being already hit by the next pin

Fig. 10 demonstrates the “Split-and-Recombine” as visualized by high speed filming in GS

flow (a) and a schematic, delineating the flow mechanism in (b): The mass is split into two

portions in front of each pin (moving from the left to the right side of the frame) and

recombines behind the pin. This phenomenon is particularly obvious at lower viscosities

and is likely attributed to repeated convective mixing. For processing GS, the Reynolds

numbers for the local flow around a cylindrical pin have been calculated according to

Schlichting and Gersten (2006). They lie in the range between 7.7 for a rotation speed of
25
1000 rpm (pin tip speed 4.61 m/s) and 23.2 for a rotation speed of 3000 rpm (pin tip speed

13.82 m/s). For Reynolds numbers between about 3 and 30, a laminar flow around a

cylindrical pin is postulated (Schlichting and Gersten, 2006) with small eddies forming

behind the pin.

Fig. 13 (section 3.4) also shows visualizations from photographs in comparison with

simulation results for GS.

3.4 CFD simulations

Fig. 11 Suspension surface radius contours (orange contours close to the wall, green and blue contours at

inner radius) for the MF+30%-Bent. composition at a rotation speed of 2500 rpm and a mass throughput rate

of 300 kg/h at 60 °C

The simulation visualization shown in Fig. 11 also reveals an impression of the “Split-and-

Recombine” mechanism as well as the spiral bulge forming at higher mass throughput

rates and viscosities of the mixture. Similar “cross-sectional spiraling motion” as a result of

the combination of axial transport and blade (or in our case pin) rotations has been

observed by Portillo et al. (2010) in their work with continuous powder-only mixers.

26
Fig. 12 Volume fraction of the IC at a rotation speed of 1750 rpm and a mass throughput rate of 1000 kg/h at

60 °C for three pin inclination angles: a -45°-inclined semi-cylindrical pins, b cylindrical pins,

c +45°-inclined semi-cylindrical pins; below: explanation of orientation in the mixer

The simulation visualization on IC suspension volume fractions in the mixer depicted in

Fig. 12 shows the dependence of the mixer’s filling degree on the pin shape and

inclination, implying correlated residence times. The suspension volume occupies a major

portion of the mixer volume for the semi-cylindrical pin assembly with an inclination angle

of -45° and only a minor portion with an inclination angle of +45°, with the filling degree

for the cylindrical pin assembly in between.

The simulation showed no clear trend for the turbulence in the mixed mass volume,

particularly below the pins, but additional simulations (not shown) revealed a stronger

degree of turbulence for a pin inclination angle of –45°.

27
Fig. 13 Photographs through the mixer window (left side) and simulation visualizations (right side) for GS at

30 °C, both for a fluid throughput rate of 100 L/h and a rotation speed of 1000 rpm (top) and 2500 rpm

(bottom), respectively

The simulation visualization for GS at 2500 rpm shown in Fig. 13 reflects the experimental

findings reliably concerning aspects such as the appearance of the fluid surface and the

thinner fluid layer where pin traces remain recognizable. Since no additional tracer particle

is involved, this indicates that the values calculated by the CFD simulations predict the

flow behavior in the mixer satisfactorily.

3.5 Exemplary comparison of results for the residence time, measured in PEPT

experiments and parameter studies and calculated by simulation

Table 5 Comparison (congeneric compositions and throughput rates) of residence time values from PEPT

measurements, CFD calculations and one parameter study, all for a rotation speed of 2500 rpm

Composition C.a PEPT experiments Simulations Parameter studies

GS 1 119.77 sb 0.7782 sc
throughput rate 1000 L/h 1000 L/h
MF+6%-Bent. 1 6.42 sb 0.5569 sc
throughput rate 1000 L/h 1000 L/h
b
MF+6%-Bent. 2 11.81 s 0.7784 sc

28
 MF+5%-Bent. 1.25 sc
throughput rate 400 L/h 400 L/h 250 kg/h ≈ 350 L/
h
MF+30%-Bent. 2 7.04 sb 0.5982 sc
throughput rate 120 kg/h 1350 kg/h
a b c
= Configuration, = median value, = average value

Table 5 gives an overview over selected residence time values as measured by PEPT,

calculated in simulations and one value calculated from the parameter studies. For the

reasons given above, median values were calculated for the PEPT experiments while the

values from the parameter study and the simulation calculations are average values. Apart

from that, the simulations were calculated using the same conditions as in the PEPT

experiments, however based on a configuration with two pin spirals over the full mixer

length in contrast to the experimental pin configuration (Fig. 1a) with only one spiral in the

front zone.

The value from the parameter studies for the MF+5%-Bent. composition corresponds well

to the one calculated from simulation for the MF+6%-Bent. composition.

The average residence time values calculated from simulations all cover the same range of

around 0.5-1 s, regardless of composition and throughput rate.

The values received from the PEPT experiments range 10-fold higher. In the case of the

MF+30%-Bent. composition used, a comparison is clearly not suitable due to a largely

differing throughput rate. But for the other compositions as well, particularly for GS, the

PEPT experimental results reveal a very high median residence time which is attributed to

the tracer particle’s almost exclusive movement near the mixer wall (also due to its

comparatively large size) that is not reflected by simulation (1). This behavior has also

been underlined by Florian et al. (2014) for centrifugal regimes.

Furthermore, simulations were based on a premixed suspension, not taking the behavior of

a single particle into account (2).

29
Pianko-Oprych et al. (2009) mentioned another aspect to be considered when comparing

experimental results to those of simulation calculations. Due to the high rotation speed

required to enable mixing of high powder fractions into the very viscous liquid, a

significant amount of air was entrained into the suspension, also influencing the

experimental results as in fact a 3-phase-system was measured. In contrast, simulations

could not take air entrainment into account as they were calculated with the viscosity

values taken from rheological measurements where no air was entrained. This difference in

the preconditions is assumed to induce a part of the differences found in the comparison

above (3).

We expect these three arguments to be the main reasons why experimental and simulation

results were not closely comparable and that this difference reflects the chances in and

limitations to either investigational approach. Their implications have to be considered

according to the intention of the application of either approach in research.

Nevertheless, most of the result ranges for both the chosen Newtonian as well as the non-

Newtonian fluid are in good accordance to a rough estimation of the residence time that

takes throughput rate and wall layer thickness into account where a range of 2-10 s is

received.

4 Discussion

4.1 Flow mechanisms and their detection by PEPT, high speed film visualization

and simulation – “Split-and-Recombine”

The exemplary high speed film visualization in Fig. 9 and the simulation results in Fig. 10

and Fig. 11 show the flow mechanism around the pin to be initiated by a splitting of the

mass in front of the pin for both Newtonian as well as non-Newtonian fluids. Depending

on process conditions and at lower viscosities in particular, a part of the mass flows

30
together behind the pin and is split again by the following pin. We describe this flow

pattern as “Split-and-Recombine” as it enables repeated convective recombination of fluid

elements. These findings were also validated for the Newtonian fluid GS by the PEPT

experiments: The measured axial velocities show the degree of axially backwards oriented

movement to be only slightly lower than the one of forwards oriented movement, which

suggests a back-and-forth transport.

In numerical results, Chen et al. (2011) found that pins disrupt and partially disorganize

particle trajectories in mixing sections, split material flow and change particle moving

directions. They stated that pin mixing sections increase the material residence time and

thereby also in mixing performance and efficiency. The same effect of pins dividing a

temporary channel in two different channels, thereby separating flows, was observed by

Yabushita et al. (1989) in their investigation of barrel extruders equipped with additional

pins.

For suspensions with higher viscosities, simulations display the major portion of the fluid

mass to be transported in a spiral bulge. This additional mechanism accounts for the higher

axial velocities in these compositions and was also reflected by the PEPT results

(section 3.3.1) and the lower residence times in the parameter studies.

Regarding the detection of flow mechanisms in continuous pin mixers, the visualization

methods through the mixer window and PEPT only reflect the situation near the mixer wall

as explained in section 3.6. However, for other types of mixers PEPT can provide reliable

results also in the inner parts of a mixer, but the tracer particle size still needs to be chosen

carefully. A major advantage of PEPT is its access to opaque systems while not

influencing the mixing process and allowing for elaborate evaluations beyond the scope

presented here.

Our simulations gave reasonable predictions of flow mechanisms, velocities and residence

times in the mixer.

31
4.2 Influence of pin shape and inclination angle on the residence time and mixing

intensity as well as on particle sedimentation at the mixer wall

As reported in the results of the parameter studies and of the conducted simulations, the pin

shape and the inclination angle of the semi-cylindrical pins have a greater influence on all

investigated aspects than the pin fraction.

This finding is consistent with the observations of Chen et al. (2011) who extracted from

their simulations that the mixing performance is not necessarily enhanced by an increased

number of pins.

Opposing effects of +45° and -45° inclination angle of semi-cylindrical pins were found:

+45°-inclined pins were substantiated to cause enhanced axial flow, lower residence times

with an associated lower filling degree of the mixer and due to the combination of these

impacts a lower mechanical input on the suspension. -45°-inclined pins promote backflow

with associated higher residence times. In addition to the positive impact of prolonged and

also backwards-oriented recirculation in the mixer, simulations indicate stronger

turbulence around the pins and in the pin-wall gap. Additionally, the turbulence around the

pin tip could be particularly significant although the responsible micro-scale mechanisms

could not be identified yet.

The effect of inclined blades in continuous powder mixers, leading the flow partially

backward, on mixing performance was also investigated by Gao et al. (2011a) and in

several works of Portillo and coworkers (2008, 2009; 2010). Gao et al. (2011a) observed

more effective mixing for alternate blade patterns with some of the blades pushing the

powder backwards and ascribed this to convective recirculation, enhancing the mixing rate.

Gao et al. (2011a) and Gao et al. (2011b) found an increase in fill level to increase the

mean residence time and lower fill levels to shorten residence time and slow down axial

mixing. Lee et al. (2012) reported that decreased screw speed and kneading blocks

32
arranged in an increased angle in mixing zones increased residence time in granulation

processes.

Portillo et al. (2010) described increased impeller rotation speeds to result in longer total

path length. However, they found that an increase in flow rate at lower rotation speed

increased the residence time and decreased the axial dispersion while an increased flow

rate at higher rotation speed still decreased the residence time, but had a complex effect on

axial dispersion.

Laurent and Bridgwater (2002b) reported effects of fill level and increased blade angle to

promote axial dispersion forces in a horizontal mixer that they investigated using PEPT.

Florian et al. (2014) summarized that in continuous mixers relying on high-shear

convection-blades, the residence time distribution and connected mixing homogeneity was

directly affected by the parameters discussed in literature: blade rotation rate, blade design,

blade configuration, flow rate, and material properties. These parameters were also the

ones that could affect mixing performance in the continuous high speed pin mixer

investigated in our work.

Excursus: Experiments on the influence of process parameters on demixing and

sedimentation at the mixer wall and confirmed impact of flow behavior around the

pin and the backwards-oriented flow created by the -45°-inclined semi-cylindrical

pins

Optimizing conditions for initial powder wetting by the highly viscous liquid is decisive to

prevent demixing by centrifugal forces in the high speed pin mixer. Fig. 14 gives an

impression of the deposition forms of a separated powder particle layer at the mixer wall at

high powder fractions and its streaky appearance as effectuated by the pins grooving

through the layer.

33
Fig. 14 Deposition of powder as a soaked particle layer adhering to the mixer wall. a Mixer window with

deposits along the full axis with marker lines indicating the pin tip diameter. b Stripes of deposits forming as

“lines” in pin tip diameter width and dotted “lines” in half the width. c Close-up of the deposit pattern.

d Demonstration of the layer’s spatial structure and thickness with an incidental imprint of a pin on a

demounted stub of a provided unused powder inlet.

In a case study connected to this work, several configurations with combinations of

different pin fractions in the front and back mixing zone and cylindrical and semi-

cylindrical pin shapes were tested regarding their potential to prevent particle

sedimentation at the mixer wall. Increasing powder fractions were dosed to the MF flow at

several combinations of rotation speeds and suspension throughput rates.

A configuration consisting of 50 % -45°-inclined semi-cylindrical pins in the front zone

and 100 % cylindrical pins in the back main mixing zone was found to allow the highest

powder fractions of up to 50 % to be added. Only then, particles started to adhere to the

mixer wall as evaluated through the mixer window. The superiority of this configuration is

supposed to result from a high degree of backwards-oriented flow in the front zone where

powder and fluid get into contact and correlated higher residence time in this zone. The

implementation of this configuration in the industrial-scale version of the high speed pin

mixer using the IC evinced the transferability of the configuration’s advantages. The

connoted reduction in agglomerates impairing product quality and the reduced mixer

34
downtimes required for cleaning are supporting the usefulness of the multi-method

approach chosen in this work.

4.3 Implications for the selection of process conditions for optimal mixing results at

different viscosities

For the continuous high speed pin mixer in this work and the non-Newtonian solid-liquid

compositions to be mixed, the major parameter influencing both the process and the

resulting product was the pin configuration. The application of specific pin shapes and

inclinations for the different requirements of each mixing zone is especially influential. A

strong mechanical impact on the respective composition and minimized particle deposits at

the mixer wall are necessary for optimal mixing results.

5 Conclusion

The combination of experimental and computational methods presented in this work

provides a valuable toolbox for the engineering and optimization of continuous high speed

mixers with complex configuration and parameter options, particularly for the mixing of

opaque compounds. The respective advantages and limitations of each method have been

explored and underlined, but the demonstrated combination of approaches gives access to

the investigation of selected aspects that are important for a successful implementation of

continuous mixing processes. The applicability of the presented methods is independent of

the materials and apparatus used. Nevertheless, markers to determine the quality of mixing

still need to be selected separately corresponding to the materials and the desired

requirements.

35
Acknowledgements

The financial support of the Swiss “Kommission für Technologie und Innovation” KTI with integrated

financial support from an industrial partner is gratefully acknowledged. The authors would like to thank

Prof. E. Casartelli, N. Ledergerber and P. Stern from the Fluid Dynamics Group of the Lucerne University of

Applied Science and Arts for providing high performance CFD simulations and visualizations in this project,

based on the KTI collaboration. The authors further want to thank Dr. T. Leadbeater and Dr. J. Gargiuli for

their support during the PEPT experiments at the Positron Imaging Centre at the University of Birmingham

that was substantial for the successful application of the PEPT methodology under new technical conditions,

and Dr. Steven Wang for his support during the submission process.

References
Barigou, M., Fairhurst, P., Fryer, P., Pain, J.-P., 2003. Concentric flow regime of solid–liquid food
suspensions: theory and experiment. Chemical Engineering Science 58, 1671-1686.
Barigou, M., Chiti, F., Pianko-Oprych, P., Guida, A., Adams, L., Fan, X., Parker, D.J. & Nienow, A.W.,
2009, Using positron emission particle tracking (PEPT) to study mixing in stirred vessels: validation and
tackling unsolved problems in opaque systems, Journal of Chemical Engineering of Japan, 42(11), 839-84
Chen, J., Dai, P., Yao, H., Chan, T., 2011. Numerical analysis of mixing performance of mixing section in
pin-barrel single-screw extruder. Journal of Polymer Engineering 31, 53-62.
Eesa, M., Barigou, M., 2008. Horizontal laminar flow of coarse nearly-neutrally buoyant particles in non-
Newtonian conveying fluids: CFD and PEPT experiments compared. International Journal of Multiphase
Flow 34, 997-1007.
Edwards, I., Axon, S.A., Barigou, M., Stitt, E.H., 2009, Combined use of PEPT and ERT in the study of
aluminium hydroxide precipitation, Industrial & Engineering Chemistry Research, 48(2), 1019–1028.
Fan, X., Parker, D., Smith, M., 2006. Enhancing 18 F uptake in a single particle for positron emission particle
tracking through modification of solid surface chemistry. Nuclear Instruments and Methods in Physics
Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 558, 542-546.
Florian, M., Velázquez, C., Méndez, R., 2014. New continuous tumble mixer characterization. Powder
Technology 256, 188-195.
Gao, Y., Ierapetritou, M., Muzzio, F., 2011a. Investigation on the effect of blade patterns on continuous solid
mixing performance. The Canadian Journal of Chemical Engineering 89, 969-984.
Gao, Y., Vanarase, A., Muzzio, F., Ierapetritou, M., 2011b. Characterizing continuous powder mixing using
residence time distribution. Chemical Engineering Science 66, 417-425.
Guida, A., Nienow, A.W., Barigou, M., 2010, PEPT measurements of solid-liquid flow field and spatial
phase distribution in concentrated monodisperse stirred suspensions, Chemical Engineering Science, 65(6),
1905-1914.
Guida, A., Nienow, A.W., Barigou, M., 2011, Mixing of dense binary suspensions: multi-component
hydrodynamics and spatial phase distribution by PEPT, AIChE J, 57(9), 2302-2315.
Kingston, T.A., Heindel, T.J., 2014. Optical visualization and composition analysis to quantify continuous
granular mixing processes. Powder Technology 262, 257-264.
Laurent, B., Bridgwater, J., 2002a. Influence of agitator design on powder flow. Chemical Engineering
Science 57, 3781-3793.
Laurent, B., Bridgwater, J., 2002b. Performance of single and six-bladed powder mixers. Chemical
Engineering Science 57, 1695-1709.
Lee, K.T., Ingram, A., Rowson, N.A., 2012. Twin screw wet granulation: The study of a continuous twin
screw granulator using Positron Emission Particle Tracking (PEPT) technique. European Journal of
Pharmaceutics and Biopharmaceutics 81, 666-673.
Liu, L., Barigou, M., 2013. Numerical modelling of velocity field and phase distribution in dense
monodisperse solid–liquid suspensions under different regimes of agitation: CFD and PEPT experiments.
Chemical Engineering Science 101, 837-850.

36
Luckham, P.F., Rossi, S., 1999. The colloidal and rheological properties of bentonite suspensions. Advances
in colloid and interface science 82, 43-92.
Metzner, A., Otto, R., 1957. Agitation of non‐ Newtonian fluids. AIChE Journal 3, 3-10.
Pakzad, L., Ein-Mozaffari, F., Upreti, S.R., Lohi, A., 2013. Evaluation of the mixing of non-Newtonian
biopolymer solutions in the reactors equipped with the coaxial mixers through tomography and CFD.
Chemical Engineering Journal 215, 279-296.
Parker, D., Broadbent, C., Fowles, P., Hawkesworth, M., McNeil, P., 1993. Positron emission particle
tracking-a technique for studying flow within engineering equipment. Nuclear Instruments and Methods in
Physics Research Section A: Accelerators, Spectrometers, Detectors and Associated Equipment 326, 592-
607.
Patel, D., Ein‐ Mozaffari, F., Mehrvar, M., 2012. Effect of impeller type on continuous‐ flow mixing of
non‐ Newtonian fluids in stirred vessels through dynamic tests. The Canadian Journal of Chemical
Engineering 90, 290-298.
Perez-Mohedano, R., Letzelter, N., Amador, C., VanderRoest, C.T., Bakalis, S., 2015. Positron Emission
Particle Tracking (PEPT) for the analysis of motion in a domestic dishwasher, Chemical Engineering Journal,
259(1), 724-736.
Pianko-Oprych, P., Nienow, A., Barigou, M., 2009. Positron emission particle tracking (PEPT) compared to
particle image velocimetry (PIV) for studying the flow generated by a pitched-blade turbine in single phase
and multi-phase systems. Chemical Engineering Science 64, 4955-4968.
Portillo, P.M., Ierapetritou, M.G., Muzzio, F.J., 2008. Characterization of continuous convective powder
mixing processes. Powder Technology 182, 368-378.
Portillo, P.M., Ierapetritou, M.G., Muzzio, F.J., 2009. Effects of rotation rate, mixing angle, and cohesion in
two continuous powder mixers—a statistical approach. Powder Technology 194, 217-227.
Portillo, P.M., Vanarase, A.U., Ingram, A., Seville, J.K., Ierapetritou, M.G., Muzzio, F.J., 2010. Investigation
of the effect of impeller rotation rate, powder flow rate, and cohesion on powder flow behavior in a
continuous blender using PEPT. Chemical Engineering Science 65, 5658-5668.
Schlichting, H., Gersten, K., 2006. Grenzschicht-Theorie. Springer-Verlag, Berlin, Heidelberg.
Weinekötter, R., Gericke, H., 2000. Mixing of Solids, Mixing of Solids. Kluwer Academic Publishers,
Dordrecht, Boston, London, p. 109.
Yabushita, Y., Brzoskowski, R., White, J., Najakima, N., 1989. Flow of Rubber Compound in a Pin Barrel
Screw Extruder. International Polymer Processing 4, 219-224.

Highlights

 A novel fast-rotating pin mixer for handling of viscous dense suspensions is described

and characterized.

 PEPT technique is applied in a fast-rotating system to measure 3D particle trajectories

in opaque flows.

 Information attained from different experimental approaches allows further

optimization of the up-scaled mixer.

 CFD and experimental results confirm the specific flow mechanisms of Newtonian and

non-Newtonian fluids in the novel mixer.

37