Powder Technology 335 (2018) 114–123

Contents lists available at ScienceDirect

Powder Technology

journal homepage: www.elsevier.com/locate/powtec

Impact of grinding aids and process parameters on dry stirred

media milling
P. Prziwara ⁎, L.D. Hamilton, S. Breitung-Faes, A. Kwade
Institute for Particle Technology, Technische Universität Braunschweig, Volkmaroder Straße 5, 38104 Braunschweig, Germany

a r t i c l e i n f o a b s t r a c t

Article history: The demand on minerals with increasing product fineness is currently rising in many industrial applications.
Received 24 January 2018 Especially in dry grinding processes, fine powders are difficult to produce and to handle. Particle-particle attrac-
Received in revised form 4 April 2018 tive forces, which become more decisive with decreasing particle size, lead to a higher extent of agglomeration,
Accepted 10 May 2018
material adherences and a more cohesive flow behavior. As a consequence, dry fine grinding processes demand
Available online 21 May 2018
large quantities of energy due to the amount of applied energy that is dissipated into heat. Current approaches
Keywords:
for improving the energy efficiency mainly cover the improvement of a) machine equipment and b) the material
Dry fine grinding behavior. The main focus of this study was to investigate both aspects within one single study: On the one hand,
Grinding aids dry fine grinding of limestone was investigated in a dry operated stirred media mill, which is a promising and
Dry stirred media mill emerging option for dry fine grinding applications. On the other hand, the impact of the particle stabilization
Powder flowability by liquid grinding aids on the grinding performance was evaluated. It was demonstrated that the grinding
mechanism inside the mill depends on both the mill parameters as well as the powder flow behavior. Therefore,
it is of crucial importance to adjust the mill parameters to the applied grinding aid when it comes to dry fine
grinding in media mills.
© 2018 Elsevier B.V. All rights reserved.

1. Introduction stress energy, high stress frequency as well as better energy efficiencies
compared to ball mills [3,4]. Stirred media mills, which are already
The production of fine minerals and mineral products is often linked established in wet fine and ultra fine grinding, have recently been
to high energy consumptions. Especially dry fine grinding processes developed for dry applications in both, vertical (e.g. [5,6]) and horizon-
require high specific energies because most of the applied energy is tal (e.g. [7,8]) designs. First studies underline that dry stirred media
transferred into heat. Moreover, material sticking and adhesion of the milling is very different in comparison to wet stirred milling, i.e. as com-
fine dry particles lower the grinding efficiency significantly. Considering parably moderate stirrer tip speeds [7,9] and lower bead filling ratios
an increasing demand on fine minerals, new methods improving [7,10] are favorable for a high energy utilization. However, due to the
the common production ways are of great importance. The different difficult process control of fine grinding processes in dry stirred media
approaches for increasing the energy efficiency of dry fine grinding pro- mills, this development is still in an early to intermediate stage.
cesses can roughly be divided into three groups: (1) further develop- An enhancement of the grinding process by controlling the behavior
ment of mills and mill equipment, (2) improving classifiers as well as of the ground material is mainly achieved with the aid of various liquid
grinding-classifying-circuits and (3) enhancing the process behavior of chemicals (see e.g. [11]) which are added to the grinding process
the ground material [1]. This investigation focuses on two of these fields in small quantities of 0.01–0.2 wt% in relation to the product mass. In
simultaneously: Comminution by dry operated stirred media mills as an industrial grinding plants, the primary objective of those grinding aids
emerging mill type as well as improving the material processability by is to increase the mill throughput and/or energy efficiency or even to
an appropriate application of liquid grinding aids. reach certain product finenesses [12]. They primarily act by reducing
Nowadays, conventional ball mills are steadily being replaced particle-particle interactions, and thus, by improving dispersability
by more efficient mill types like vertical roller mills or high pressure and flowability of the product powder [13]. A previous study revealed
grinding rolls [2]. Another promising option for efficient dry fine and how different grinding aids and grinding aid concentrations affect vari-
ultra fine grinding is the use of stirred media mills due to their high ous properties of ground limestone like product fineness, agglomeration
tendency, powder flowability, surface energy and adhesions on grinding
⁎ Corresponding author. media [14]. In addition to that, the impact of grinding aid applications
E-mail address: [email protected] (P. Prziwara). on the material transportation and retention time inside the mill as

https://doi.org/10.1016/j.powtec.2018.05.021
0032-5910/© 2018 Elsevier B.V. All rights reserved.
 P. Prziwara et al. / Powder Technology 335 (2018) 114–123 115

well as on air classification efficiency has been demonstrated in various

investigations [1,15–17]. Although grinding aids have been established
in several industrial applications since the 1960's, the impacts of process
and mill parameters and also grinding aid dosage strategies on grinding
aid effectiveness are still not understood satisfactorily. Additionally,
most of the few referring publications deal with conventional ball
mills instead of modern mill designs. For example, Deckers and Stettner
[18] came to the conclusion that the grinding aid efficiency in tumbling
ball mills can depend on the type of grinding media as well as the mill
speed, but is independent of the ratio between product and media
masses. In contrast to that, Dombrowe et al. [15] did not identify any
influence of the process parameters such as mill speed, filling ratios,
temperature and grinding media on the grinding aid efficiency. How-
ever, Mothedath and Ahluwalia [19] determined a further impact
of grinding aids on the stress mechanisms inside a tumbling ball mill
at different process parameters. They found that high grinding aid con-
centrations result in a decrease of the contribution of attrition to the size
reduction. Furthermore, Scheibe et al. [20] investigated different grind-
ing aid dosage strategies – like dropwise dosing or addition by spraying
and evaporation – as well as several injection times. They concluded
that dosage strategies and times do not have a significant impact on
the grinding efficiency as long as the grinding aid is dosed completely
prior to the period of fine grinding. In the same study, they showed a
minor effect of temperature on particle size distribution and flow prop-
erties but greater impacts on specific surface area, product adhesion and Fig. 1. Setup of the laboratory, vertical stirred media mill for dry fine grinding applications.

cement strength. A strong increase of the product adhesions with rising

temperature in dry fine grinding of cement in a dry horizontal stirred
media mill has also been demonstrated by Altun et al. [21]. Moreover, the specific energy Em is calculated according to Eq. (1)
Within this work, the investigation of dry fine grinding in a vertical
stirred media mill is combined with the evaluation of the product
P−P 0
behavior as a consequence of the particle stabilization. Therefore, lime- Em ¼ ∙ tG ð1Þ
mP
stone was milled in batch-wise experiments using different liquid
grinding aids as stabilizing additives. Furthermore, process and mill
parameters as well as grinding aid dosage strategies have been varied with the electrically measured power draw P as well as no load power
in order to investigate their impacts on the grinding efficiency. Thereby, draw P0, the product mass mP and the grinding time tG.
the interdependence of the mill parameters and the product behavior As the mill did not exceed a gross power draw of 190 W within this
was evaluated. study, the temperature inside the grinding chamber did not increase
by N4 °C in the grinding experiments. Thus, possible temperature effects
2. Experimental are neglected within the following discussion.

2.1. Vertical stirred media mill 2.2. Grinding parameter

The dry fine grinding experiments were carried out in a self- The impact of the stress conditions inside the mill on the grinding
constructed batch-wise operated vertical stirred media mill shown in result was examined by varying the media size dGM, grinding media
Fig. 1. The agitator is driven by a Dispermat AE-06 by VMA-Getzmann, density ρGM and stirrer tip speed vt. Therefore, three different kinds of
Germany. grinding media materials as well as a number of media sizes were inves-
The mill consists of a 1.6 L stainless steel grinding chamber (2). The tigated in this study (see Table 1), while the grinding media filling ratio
chamber was designed with rounded edges to ensure a sufficient mate- ϕGM was kept constant at 40%. Additionally, the stirred tip speed was
rial mixing, even in the lower mill segments. A double-shell (3) enables varied in a range of 1–4 m/s.
the jacket cooling for the temperature control of the mill charge. Further mill parameters, which do not have a primary effect regard-
Further, an air stream can be passed into the mill via an air connector ing the stress energy of the grinding beads, were held constant in all
(4) at the bottom of the grinding chamber. Prior to this process, the experiments (see Table 2). Within this study, the grinding experiments
air is led through a cooling coil (5) which is placed inside the cooling were run without any air volume flow through the mill. Solely in one
jacket. Thereby, the air stream is tempered at the same value as the test series, where the grinding aid was added to the mill via air stream
walls of the grinding chamber. This combination offers a better temper- at the bottom of the grinding chamber, an air flow of 10 L/h was used.
ature control of the milling process. Using water as cooling medium,
process temperatures of 5 to 80 °C can be investigated. The shaft
(1) can be equipped with up to four cross stirrers (6) depending on Table 1
the filling ratio on the mill. Both, shaft and stirrers are also made of Grinding media.
stainless steel. The air stream is discharged from the mill via a filter Material Average size dGM [mm] Solid density ρGM [kg/L]
(8) on top of the chamber lid (7). Separated particles are removed
Steel 3.2 7.86
from the filter surface by tapping and led back to the grinding chamber Zirconia 3.1 6.07
by gravity. The addition of liquid grinding aids is realized either by an Alumina 1.4 3.62
interface in the chamber lid or by an injection into the air stream within 2.0
the connection part at the chamber bottom (4). The dosage is controlled 3.2
3.8
by a syringe pump 540060 of TSE Systems.
116 P. Prziwara et al. / Powder Technology 335 (2018) 114–123

Table 2 dispersion unit and the lens system R2 for analyzing particles in a
Mill parameters kept constant for the grinding experiments. range of 0.25–87.5 μm. For the analysis, the powder samples were
Further mill parameter dispersed in 96% ethanol and treated with ultrasonic (60 s) prior to
Air volume flow
the analysis. Samples for the particle size analysis were taken from the
Standard 0 L/min total amount of ground material. 3 measurements were done for each
Grinding aid addition via bottom 10 L/min test setting.
Temperature 30 °C
Grinding chamber volume VGC 1.6 L
2.5. Powder flowability
Grinding media filling ratio ϕGM 40%

The powder flowability was analyzed using a ring shear tester RST-
2.3. Material XS (Dietmar Schulze Schüttguttechnik, Germany) and a 30 mL shear
cell. Measurements were performed at ambient conditions directly
Limestone Durcal 40 with a mean particle size of x50 ≈ 20 μm after the grinding experiment. At the beginning of the shear test,
was ground in the presented vertical stirred media mill. The initial the powder sample was pre-sheared under a normal stress in order
solid and bulk densities of this material are ρS = 2.73 kg/L and ρB = to load the bulk material with a major consolidation stress σ1. In the
1074 kg/L, respectively. The amount of limestone was calculated present case, a comparatively high normal stress of 10,000 Pa was
according to Eq. (2). chosen, as it enabled the most distinct differentiation between the
different grinding aids, while maintaining lower standard deviations
! compared to other consolidation settings at the same time. After consol-
mGM
mP ¼ φP ∙ V GC ∙ φGM∙ − ∙ ρB ð2Þ idation, the powder sample is successively sheared under reduced
ρS;GM
normal stresses of 2500, 5000 and 7500 Pa. Each resulting value pair
of normal and shear stress is used to construct the yield locus in order
where φP is the filling ratio of limestone related to the void volume to determine the unconfined yield strength σC of the powder sample.
within the media bulk. Within this study, φP was varied between 25 In this study, the flowability index ffC, which is defined as the ratio
and 100%. Moreover, φGM, mGM and ρS,GM are the filling ratio, mass and between major consolidation stress σ1 and unconfined yield strength
solids density of the grinding media. The entire mill filling was removed σC, is used to qualify the powder flowability. According to Jenike [22],
from the mill after reaching the required grinding time. After separating the ffC can be used to classify the powder flow behavior into five groups
the limestone from the grinding media by sieving, a sample was taken (see Table 4).
from the entity of product volume in order to analyze the particle
size distribution and powder flowability. The statistical reliability of 3. Results and discussion
the results was improved by refilling the mill with fresh limestone
and grinding aid for each investigated grinding time. 3.1. Powder flowability and grinding mechanism
In order to investigate the impact of the product powder flowability
on the grinding process, the flow behavior of the powder was specifi- Within this study, DEG and HepAc were chosen as grinding aids
cally adjusted by adding different liquid grinding aids. According to since they are known to cause a significantly different flow behavior
a former study [14], diethylene glycol (DEG) and heptanoic acid of fine limestone powders [14]. Fig. 2 shows the impact of these grind-
(HepAc) were chosen as additives, as they decisively demonstrated dif- ing aids on the particle size x90,wet as a function of the specific energy
ferent effects on the powder flowability index ffC of ground limestone. Em (left). Additionally, the results of grinding without the addition of
As also shown in this previous study, powder flowability is strongly additive are included. Hereby, identical mill parameters and grinding
dependent on the grinding aid concentration. Thus, the grinding aid times were used. Furthermore, the change of the powder flowability
concentration was set to a comparatively high value of 0.1 wt% related index ffC measured by ring shear tests is illustrated as a function of the
to the mass of ground limestone in order to maintain different flow particle size x90,wet (right).
properties over a wide product fineness range. The main properties of It can be seen, that the grinding result strongly depends on the par-
the two chemicals are shown in Table 3. ticle stabilization achieved by grinding aid molecules. While DEG affects
Within the standard procedure, the entire amount of grinding aid a more efficient grinding in the beginning of the process, HepAc shows
was dropped on the limestone inside the mill prior to the grinding the better results at high product finenesses (approx. x90 b 15 μm). Also,
experiment. In an additional section (see Chapter 3.4), different dosage at low specific energies, HepAc is even inferior to grinding without
strategies were evaluated. Thereby, the additive was added either any additive. This was not to be expected since HepAc had delivered
by pre-mixing to the limestone using a ploughshare mixer or by adding the best grinding results in a former study [14]. However, grinding with-
to the grinding chamber using continuous dosage strategies. Among the out additive becomes less efficient with progressing grinding time fairly
continuous strategies, different kinds of dosage rates like constant rates, soon. Here, no high product finenesses are achieved which underlines
steadily rising rates or stepwise additive addition were investigated. the necessity of grinding aids in the matter of dry fine grinding.
Interestingly, certain similarities are observed between the grinding
2.4. Particle size analysis results (left) and the powder flowability index ffC (right). On the one
hand, the low grinding efficiency affected by HepAc in the beginning
Particle size distributions were measured by laser diffraction using a of the process corresponds to high powder flowability indices of ffC
Helos instrument (Sympatec, Germany). The device fulfills ISO 13320
and is equipped with a laser with a wavelength of 632.8 nm, a SUCELL
Table 4
Powder flow behavior according to Jenike [22].
Table 3
Used grinding aids with manufacturer specifications. ffC size range Jenike flow behavior

ffC b 1 Not flowing

Name Molecular Purity Density Boiling point
1 b ffC b 2 Very cohesive
formula [%] [kg/L] [°C]
2 b ffC b 4 Cohesive
Diethylene glycol (DEG) C4H10O3 N99.0 1.12 244 4 b ffC b 10 Easy-flowing
Heptanoic acid (HepAc) C7H14O2 N96.0 0.918 223 10 b ffC Free-flowing
 P. Prziwara et al. / Powder Technology 335 (2018) 114–123 117

Fig. 2. Particle size x90,wet as a function of the specific energy Em (left) for grinding with and without grinding aid, as well as the corresponding powder flowability indices ffC (right).

N 6. Whilst grinding, the powder flowability index decreases due to the by stressing particle beds. Thereby, the particle displacement deter-
rising product fineness, and thus, the higher extent of particle-particle mines the active mass of captured and stressed product particles mA,
interactions within the bulk. At this point, the grinding in turn becomes and thus, the grinding conditions inside the mill. Therefore, stress
more efficient. On the other hand, grinding also becomes less efficient energy as well as contact frequency are recommended to be related to
when the flow behavior of the bulk material falls below a certain the active mass, like proposed by Burmeister et al. (see Eqs. (5) and
value, which is especially illustrated by the grinding experiment with- (6) [24]).
out any additive. Moreover, grinding with DEG down to very small par-
ticles leads to similar albeit smaller effects. Regarding the diagram, the SEGM
SI ¼ ð5Þ
slopes of the grinding curves decrease at corresponding ffC values mA
below 3–3.5 (i.e. at a cohesive flow behavior). However, an overlapping
mA
effect is assumed when grinding without any additive: Since the grind- CF P ¼ CF GM ð6Þ
mP
ing aid molecules are missing and consequently the particle stabiliza-
tion is non-existent, the grinding efficiency may also decline due to a
Early works by Hanisch and Schubert [25] as well as Schönert [26]
higher extent of agglomeration.
have already aimed to quantify the amount of particles that are cap-
It is known from wet operated stirred mills, that the grinding result
tured and stressed between grinding balls in common ball mills. As a re-
strongly depends on the stressing energy of the grinding beads SEGM
sult, Schönert presented a geometrical model in order to quantify the
as well as the frequency of their collisions CFGM. Thereby, the stressing
amount of stressed particles (see Fig. 3). He described the amount of
energy of the beads is determined by their kinetic energy Ekin,GM,
captured and stressed particles by an active volume VA, which can be as-
which is primarily affected by the bead size dGM, the grinding media
sumed to be proportional to the active mass mA. The active volume is
density ρGM and the stirrer tip speed vt. The collision frequency CFGM is
further approximated by the grinding bead diameter dGM, a capturing
defined as the number of bead collisions CNGM per time t, and is there-
angle α0 and the particle bed height h.
fore a function of stirrer revolutions per time nS and total number of
grinding beads NGB (see Eqs. (3) and (4) by Kwade [23]). π 2
VA ¼ ðdGM ∙ sinα 0 Þ ∙ h ð7Þ
4
3
SEGM ¼ Ekin;GM ∝ dGM ∙ ρGM ∙ v2t ð3Þ
Schönert gave estimations of α0 ≈ 0.08–0.12 and h ≈ 4 ∙ xmax for dry
CNGM V GC ∙ φGM ∙ ð1−εÞ fine grinding in ball mills. Even though Schönert also mentioned the
CF GM ¼ ∝ ns ∙ NGB ∝ ns π 3 ð4Þ
t impact of friction and drag forces on the particle capturing, especially
∙d
6 GM in the case of fine products of particle sizes b10 μm [27], parameters
like the powder flow behavior or the contact velocity of the balls are
where ε is the porosity of the grinding media bulk. Rácz and Csöke [10] not included in this model. However, a study by Oettel and Husemann
confirmed the importance of these parameters when describing the
grinding conditions in dry operated stirred media mills. However,
simulation studies by Burmeister et al. [24] on dry operated planetary
ball mills indicated, that those parameters influence the stress energy
and frequency to a different extent in dry operated mills than in wet
operated stirred mills.
Even though these machine-related parameters were held constant
in Fig. 2, significantly different results are observed between the three
different formulations. Hence, the question arises, how product-
related values like the powder flowability may change the grinding
mechanism inside the mill. Hereby, a deeper insight into the occur-
rences between the grinding beads needs to be considered: In contrast
to wet stirred milling, where high fluid drag forces displace the majority
of suspended product particles from the active grinding zone between Fig. 3. Model by Schönert for quantifying the particle capturing between grinding balls
the beads, dry fine grinding in media mills is assumed to happen mainly (modified from [26]).
118 P. Prziwara et al. / Powder Technology 335 (2018) 114–123

[28] showed that the powder flowability has a crucial impact on the the coatings are nearly prevented by the added grinding aids: In order
powder displacement/capturing between the grinding tools. It empha- to determine the impact of product adhesions on the grinding process,
sizes that the different grinding results shown above are a consequence the material coatings on the beads were determined for selected exper-
of the different powder flow behavior. Here, the particle capturing iments by sieving (10 s at 1000 μm mesh size) and subsequent weighing
between two approaching grinding beads may be strongly affected by of the beads. For instance, material adhesions per bead surface were
the powder flowability (see scheme in Fig. 4). As long as the product compared for the experiments shown in Fig. 2 after 16 min of grinding.
powder possesses high powder flowabilities, the particles are easily Here, values of 3.4 g/m2, 3.9 g/m2 and 28.7 g/m2 were obtained for
pushed out of the active grinding zone between the approaching the grinding with DEG, HepAc and without additive, respectively. In
beads. Thus, less particles are captured and stressed between colliding general, grinding without any additive leads to significantly higher
beads. On the one hand, this circumstance results in comparatively adhesions of product particles on bead and mill surfaces compared to
high stress intensities SI since the stress energy of the grinding beads the application of grinding aids, especially due to the missing stabiliza-
is transferred only to a small mass of product particles mA. On the tion of the product particles. These higher amounts of adhesions also
other hand, only a low mass of particles is stressed at each contact contribute to the inefficient grinding, as material coatings lead to both,
event leading to a low product related stress frequency CFP. If the a further increase of the active mass as well as a damping of the colliding
product powder flowability is low, more particles are captured and beads, and thus, less efficient stressing events. As no significant differ-
stressed between the beads. Hence, lower stress intensities but higher ences regarding the sticking behavior are seen for DEG and HepAc, the
product related stress frequencies are achieved. Below a certain powder varying grinding performances of the two additives are rather attrib-
flowability, the captured particle mass becomes too high in relation to uted to the flow phenomena explained above. It underlines that the
the stress energy provided by the beads, making the grinding process grinding performance is rather an interaction of the stress conditions
less efficient. provided by the mill (i.e. SEGM and CFGM) and the bulk behavior of the
In other studies [10,29], the active mass mA was solely approximated product, and thus, the particle stabilization obtained by applying grind-
by the coating of product particles on the grinding beads. In this study, ing aid molecules. Hence, a sole consideration of the product adhesions

Fig. 4. Schematic illustration of the particle capturing between approaching grinding beads depending on the product powder flowability.
 P. Prziwara et al. / Powder Technology 335 (2018) 114–123 119

on the mill equipment is not sufficient to describe those dry grinding grinding-relevant values like the stressing energy and contact fre-
processes, within which the major part of the product material is rather quency of the grinding beads. On the other hand, the energy dissipation
available as free powder and not as adhesions on the milling equipment. within the bulk due to friction may increase especially for cohesive
Therefore, the impact of the product adhesions on the grinding perfor- powders. This assumption is supported by the measured power draw
mance is neglected in the later discussion when comparing the two of the stirring unit, which is slightly but constantly higher in the case
different grinding aids. of the DEG formulation. Furthermore, a “damping” of the beads caused
by high amounts of captured particles may influence the breakage be-
3.2. Variation of the product filling ratio havior within the stressed particle bed, but also the motion behavior
of the beads after their collision. Altogether, it can be assumed that a
Fig. 5 shows the impact of the limestone filling ratio on the grinding less cohesive flow character leads to a lower extent of frictional losses
result using the two different grinding aids, a tip speed of 2 m/s and within the bulk as well as a more efficient grinding media motion
3.2 mm alumina beads as grinding media. which results in higher frequencies of bead contacts and contact
It can be stated that the influence of the different powder flow velocities.
behaviors on the grinding result also depends on the powder filling As already suggested, the relation between stress intensity and
inside the mill. At low filling ratios of 25 and 50% the two additives affect breakage behavior of the particles stressed within the active volume
almost the same x90,wet, whereas the filling ratio of 50% leads to a much needs to be considered when describing the grinding process. Indica-
more effective grinding. If the product filling is further increased, tions for this kind of dependencies can already be derived from the
the grinding efficiency stays almost constant with the application of present study. As it can be seen from Fig. 6, the different characteristic
HepAc. However, the product fineness decreases again with higher values of the particle size distribution, x10, x50 and x90, depend on the
product fillings when using DEG. In contrast to the results shown in added additive, the higher the percentage value is (e.g. x90). It indicates
the previous chapter, HepAc leads to a more efficient grinding than that the mass of captured particles has a selective impact on the break-
DEG when using smaller beads (thus lower stress energies) in combina- age behavior within the stressed particle bed. In the case of high masses
tion with high product-to-bead-ratios, which supports the model of of captured particles, the comparatively low stress intensities might
particle capturing between the beads. Furthermore, these results indi- prevent an effective comminution of the coarser and more robust
cate that this particle capturing is not only affected by the flow behavior particles, thus leading to higher amounts of coarse particles within the
of the product powder, but also by the initial amount of bulk material product. A further damping of the beads by fine cohesive particles can
between the beads during their approaching: Especially in the case increase this effect.
of low flowabilities (here: DEG) more particles are captured at high Obviously, this impact on the breakage behavior alters the whole
product filling ratios, because the high amount of cohesive particles is particle size distribution. The impact of the two grinding aids on the
less prone to be pushed out of the active grinding zone. Consequently, width of the particle size distribution is shown in Fig. 7 for the different
the stress intensity falls below a critical value, which reduces the energy product filling ratios using the poly dispersity index PDI (see Eq. (8)).
efficiency. In turn, a high flowability assists the particle flow and pre-
vents unnecessarily high amounts of captured particles, which would x90 −x10
PDI ¼ ð8Þ
lead to inefficient grinding at a given stressing energy of the beads. x50
The results show that the consideration of the active mass mA includ-
ing the calculations from Eqs. (3)–(6) enables a realistic estimation of The diagram underlines the assumption that the powder flowability
optimum grinding parameters as a function of the product formulation. (i.e. the kind of grinding aid) influences the breakage behavior between
However, this model alone is not sufficient for a reliable prediction of the beads, and thus, the shape of the particle size distribution. Thereby,
the grinding process. For such a prediction, the impact of the powder low and high powder flowabilities lead to a broader and narrower par-
flow behavior needs to be taken into account on a further level: Similar ticle size distribution, respectively. In the case of DEG, the size distribu-
to the parameter “viscosity” in wet operated stirred media milling pro- tion becomes even wider with increasing specific energy input, while it
cesses, the flowability of the product material determines the motion stays almost constant or even decreases (φP = 50%) when using HepAc.
behavior of the total mill filling, including both product powder and As described before, the particle capturing between the beads can be
grinding media. On the one hand, the motion of the mill filling affects assumed to affect the breakage behavior within the stressed particle
bed. Especially in case of low stress intensities SI caused by high active

Fig. 5. Impact of the product filling ratio on the particle size x90,wet as a function of the
specific energy Em. Fig. 6. Comparison of x10, x50 and x90 when using different grinding aids.
120 P. Prziwara et al. / Powder Technology 335 (2018) 114–123

Fig. 7. Impact of the product filling ratio on the width of the particle size distribution, Fig. 8. Impact of the grinding bead size on the particle size x90,wet as a function of the
characterized by the poly dispersity index PDI, as a function of the specific energy Em. specific energy Em.

masses mA (e.g. DEG), an effective comminution of the coarser and According to Eq. (7), the active mass mA is also increased by the bead
more robust particles becomes less efficient, so that products with size. Therefore, the stress intensity, which is decisive for the particle
broader size distributions are obtained. This effect is further supported breakage, does not increase proportionately to the stress energy of
by the product filling ratio: A medium filling ratio of 50% tends to result the beads or even stays almost constant. Furthermore, the frequency
in a minimum width of particle size distribution for both additives. As of contact events CFGM decreases with larger bead sizes due to the
the filling ratio increases, generally more particles are captured between lower number of beads inside the mill. This circumstance may also pre-
the beads leading to slightly higher PDI values. However, the broader vent a further improvement of the grinding result. However, this effect
distribution at the lowest filling ratio of 25% is rather a consequence of is also partly compensated by the higher active mass at each stress event
high stress intensities promoting the accumulation of fines, and thus, (see Eqs. (5) and (7)).
widening the distribution towards the submicron particle size range. By using HepAc as grinding aid, much stronger dependencies of the
As only values like the x90 consider the coarse product fractions grinding process on the grinding bead size are observed. Again, the
within the product, they are the preferred characteristic values for smallest beads cause a comparatively inefficient process. However,
comparing different additives. Therefore, grinding results are hereafter better results are obtained than using the small beads in combination
shown via the x90 as a function of the specific energy. with DEG, which can be explained by a smaller mass of captured parti-
cles at each bead collision when using HepAc. Hence, the previously
3.3. Variation of the stress conditions very low stress intensities with DEG are slightly increased, making the
process more efficient. The same effect is achieved regarding the 2.0
In the following, a more detailed process study is presented in order and 3.2 mm beads, where the most efficient grinding, and thus, much
to identify how to adjust the mill parameters to the product flow behav- better results are observed than using DEG. However, the grinding effi-
ior. Certainly, the breakage behavior of the product particles entails ciency drops off again when using large beads of 3.8 mm. Here, the
even more overlapping impacts. On the ground that the material is stress energy of the beads is too high in comparison to the (low) mass
not varied in this study, the breakage behavior is not further considered of captured particles. In the case of overly high stress intensities, exces-
as a parameter. sive energy may dissipate into heat instead of being used for particle
breakage. Furthermore, the combination of large beads and high pow-
3.3.1. Grinding media size der flowability is accompanied with a low product related stress
The impact of the grinding media size on the grinding result was frequency, which might result in an insufficient quantity of mechanical
investigated by varying the bead size of the alumina grinding media. impacts on each product particle.
Thereby, the stirrer tip speed and the product filling ratio were held
constant at 2 m/s and 100%, respectively. Fig. 8 shows the resulting par- 3.3.2. Grinding media material
ticle size x90,wet as a function of the specific energy Em for the different Fig. 9 shows the product fineness in dependence on the specific
bead sizes as well as grinding aids. energy for different grinding media materials. Thereby, the grinding
As it is shown, the grinding result depends on the bead size, and media size was held nearly constant. The stirrer tip speed and the
thus, on the grinding conditions inside the mill. Thereby, different im- product filling were held constant as well.
pacts of the bead size can be identified for the two grinding aids. Regard- The diagram emphasizes the findings from the variation of the
ing DEG, the grinding performance is almost independent of the bead grinding bead size. Again, stronger impacts can be seen whilst using
size as long as a critical, minimum bead size is exceeded. At the smallest HepAc. While the best result is achieved by using alumina media, the
bead size of 1.4 mm, the stress energy of the grinding beads is too small beads with a higher density and thus a higher stress energy lead to a
to enable an efficient comminution of those particles, which are cap- less efficient grinding. On the one hand, the higher stress intensities
tured and stressed between the beads. In contrast to that, grinding may lead to a higher dissipation of excessive energy at collisions
becomes more efficient when providing higher stress energies by of those heavier beads. On the other hand, grinding times needed to
using coarser beads of 2.0 mm and more. A further increase of the be shortened for steel and zirconia media in order to introduce compa-
bead size does not lead to an additional improvement of the grinding rable amounts of energy to the mill chamber as a consequence of the
performance, even though the stress energy of the beads steadily higher power consumption of the mill. Consequently, the total number
increases. At this point, further aspects lead to counteracting effects: of bead contacts was reduced simultaneously. In combination with a
 P. Prziwara et al. / Powder Technology 335 (2018) 114–123 121

Fig. 9. Impact of the grinding media material on the particle size x90,wet as a function of the Fig. 10. Impact of the stirrer tip speed on the particle size x90,wet as a function of the specific
specific energy Em. energy Em.

high powder flowability, and thus, a low extent of particle capturing on the particle capturing is not well understood nowadays and will be
between the beads, the overall product related stress number is even studied in a future work.
lower. However, further impacts overlap the grinding results. This can In order to further investigate the impacts of the product flow
especially be seen from the results of the zirconia beads, which are not behavior on the grinding result at medium product filling ratios, the
in between the other two materials, as it was expected regarding the stirrer tip speed was also varied exemplary for a filling ratio of 50%.
material density. Here, other parameters like the surface roughness or Fig. 11 shows the resulting grinding progress as a function of the specific
the friction behavior of the beads may have overlapping influences on energy.
the grinding result, as it was already observed for wet stirred milling The product fineness mainly depends on the stirrer tip speed and the
processes [30]. In dry fine grinding processes, those parameters might specific energy input over a wide range, but less on the added grinding
especially affect the particle capturing between the beads or the effi- aid. Hence, the diagram underlines that the extent of the powder flow
ciency of the bead motion. Unfortunately, no data on friction and surface impact decreases at low filling ratios. However, in the beginning of the
parameters of the beads are available at this point. Thus, these impacts process as well as at small particle sizes obtained at high energies, slight
will be examined in future studies. differences between the two grinding aids are observed. While DEG
A “compensation” of the low number of bead contacts can be affects the better results at small energy inputs, the grinding efficiency
achieved by capturing higher amounts of product particles between is higher at very fine product sizes by using HepAc. Here, the impacts
the beads, as seen with DEG. As long as the stress energy of the beads of either too high (HepAc) or too low (DEG) powder flowabilities
is in an appropriate range to break those higher amounts of particles become obvious once again.
efficiently, nearly no disadvantages are caused by beads with higher As the product filling ratio in continuous production mills is difficult
densities. Even on the contrary, as beads with higher densities enable to control, the derivation of global rules regarding the choice of grinding
similar product qualities in smaller grinding times, they might be favor- aids is made even more difficult. The impact of different grinding aids
able especially in continuously operated grinding processes. From an as well as various mill parameters on the product filling ratio inside
energetic point of view, the best results with DEG are obtained with continuous operated dry stirred media mill is an ongoing work.
the moderately-heavy zirconia beads. Here, the combination of stress
energy (as well as contact number) and captured particles seems to
be preferable compared to the alumina and steel beads.

3.3.3. Stirrer tip speed

By using the 3.2 mm alumina media the impact of the stirrer tip
speed was investigated in a range of 1–4 m/s. Fig. 10 shows the grinding
result as a function of the specific energy for both grinding aids.
In this case, certain similarities can be observed for the two grinding
aids. In both cases, the worst grinding efficiency is found for the highest
tip speed, which indicates that the stress energy is too high, resulting in
an increase of dissipation as excess energy. Comparing the different tip
speeds, lower efficiencies are also observed for the smallest tip speed,
whereas HepAc achieves better results for 1 m/s due a lower captured
product mass, and thus, higher stress intensities. The optima are
obtained by medium tip speeds of 3 m/s (DEG) and 2 m/s (HepAc),
which is also consistent with the model scheme of particle capturing
regarding powder flowability. Therefore, an optimum stress intensity
exists at which the energy utilization has an optimum value as it is
already well known from wet stirred media milling [31] and individual Fig. 11. Impact of the stirrer tip speed on the particle size x90,wet as a function of the specific
dry grinding processes [10,24]. The impact of the bead collision velocity energy Em when using different grinding aids at a reduced product filling ratio of 50%.
122 P. Prziwara et al. / Powder Technology 335 (2018) 114–123

Fig. 12. Particle size x90,wet as a function of the specific energy Em (left) for grinding with initial and continuous grinding aid addition as well as the corresponding powder flowability
indices ffC (right).

3.4. Impact of grinding aid dosage strategy capturing of the coarse material is improved. Furthermore, an adequate
particle stabilization as well as sufficient retention times are guaranteed
In some industrial applications, the use of high stress energies may at the same time. As the product becomes finer as well as more cohesive
be preferred (e.g. due to shorter resulting process times), which might across the mill length, an additional continuous dosage of flow-assisting
make flow assisting grinding aids like HepAc less favorable. If one of additives like HepAc in later mill sections may be considered as a good
those grinding aids is the desired additive though (e.g. for affecting cer- option. Thus, optimal particle capturing, a good material transport as
tain product qualities), a continuous grinding aid addition might be the well as a superior bead motion are maintained. These issues remain
right choice. Thereby, high amounts of grinding aids are prevented, under investigation in current research.
especially in the beginning of the process where a high powder
flowability can be disadvantageous due to the coarse product. Fig. 12 4. Conclusion
shows the grinding results for both the initial as well as continuous
grinding aid addition when grinding with a medium level of stress The results emphasize a strong impact of the powder flowability on
energy (exemplary: 3.2 mm steel beads, 2 m/s tip speed). In addition, the dry fine grinding of limestone in a batch-wise operated stirred
the corresponding powder flowability indices are illustrated. media mill. Since the flow behavior of the product is more or less
For DEG, the grinding efficiency is slightly reduced by a continuous affected by liquid grinding aids, the grinding performance depends on
grinding aid addition, which may be explained by a lower state of both the process parameters of the mill as well as the particle stabiliza-
stabilization due to the lower additive amount. However, the powder tion by an additive. As it was shown, the effects are even more decisive
flowability does not changed significantly compared to that one at medium and high product filling ratios, which are favorable regarding
obtained by grinding with the complete addition prior to the process. high productivities in industrial processes.
In contrast to that, the powder flowability is strongly reduced by a con- The study revealed that it is crucial to adjust the process parameters
tinuous addition in case of HepAc. This in turn leads to a slightly better of the mill under recognition of the type of grinding aid concerning dry
grinding performance especially at small specific energy inputs because fine grinding in stirred media mills. Since the flow behavior of the prod-
those critical high flowabilities are prevented. uct powder determines the amount of particles captured between the
Table 5 summarizes the grinding results after six minutes of grind- grinding beads, the stress intensity as well as the product-related stress
ing, using the same stress energy but varying initial and continuous frequency inside the mill firmly depend on both process parameters and
dosage strategies. It can clearly be observed that the dosage strategy grinding aid. While a higher stress energy tends to be more favorable in
plays a subordinate role in the case of DEG. Here, only the method of the case of cohesive product powders, flow-assisting additives affect a
premixing in a ploughshare mixer for 1 h gives slightly better results. more efficient grinding at lower stress energies. Particularly, when
As mentioned above, continuous dosing of HepAc increases the grinding grinding to the lower micron range, and thus, dealing with cohesive
efficiency compared to the initial additive addition. However, the differ- powder, particle capturing between the beads as well as grinding per-
ent investigated dosage rates and injection positions do not have a formance are mainly optimized by grinding aids, which are improving
significant influence. the powder flowability. Besides particle capturing, also the overall
In industrial processes, a possible way of transferring those results motion of the grinding beads is affected by the product flowability. A
may be achieved by adding the grinding aid(s) at different injection too low flowability is assumed to cause a hindered motion of the
interfaces inside the mill. Thereby, an initial addition of flowability- beads resulting in lower stress energies and lower stress frequencies
irrevelant additives like DEG might be favorable, since the particle of the beads. The grinding results clearly show, that analogous to wet

Table 5
Product particle size x90,wet after 6 min of grinding with different initial and continuous grinding aid dosage strategies (3.2 mm steel beads, 2 m/s tip speed). The bold values correspond to
the experiments shown in Fig. 12.

Grinding aid Particle size x90 [μm] after 6 min of grinding

Addition prior to grinding Continuous addition

Drop-wise Premixing Constant dosage rate (via lid) Rising dosage rate (via lid) Step-wise rate (via lid) Constant dosage rate (via air stream)

0.1 wt% DEG 17.06 16.25 17.45 16.40 16.86 17.02

0.1 wt% HepAc 16.1 16.21 14.76 14.62 15.08 14.75
 P. Prziwara et al. / Powder Technology 335 (2018) 114–123 123

stirred media milling an optimum stress intensity, which leads to [9] Y. Wang, E. Forssberg, J. Sachweh, Dry fine comminution in a stirred media mill—
MaxxMill®, Int. J. Miner. Process. 74 (2004) S65–S74.
the highest energy utilization as well as the lowest specific energy [10] Á. Rácz, B. Csőke, Application of the product related stress model for product
consumption for a demanded product fineness, also exists for dry fine dispersity control in dry stirred media milling, Int. J. Miner. Process. 157 (2016)
grinding processes. Moreover, some parallels between the role of the 28–35.
[11] T.S. Sverak, C.G.J. Baker, O. Kozdas, Efficiency of grinding stabilizers in cement clinker
powder flowability and the viscosity in wet grinding processes can be processing, Miner. Eng. 43–44 (2013) 52–57.
found. [12] B. Hoffmann, C. Bernhardt, Zur Energieeffektivität bei der Mahlung von Kalkstein
Furthermore, the grinding aid affects the width of the particle size mit Monocarbonsäuren, Powder Technol. 41 (1985) 295–300.
[13] H. Schubert, Effects of fluid and additives on grinding processes, Aufbereitungstechnik
distribution as the stress intensity determines the breakage behavior 8 (1988) 115–120.
of coarse and more robust particles. Also, it could be demonstrated [14] P. Prziwara, S. Breitung-Faes, A. Kwade, Impact of grinding aids on dry grinding
that the dosage strategy of the grinding aid might be a helpful tool performance, bulk properties and surface energy, Adv. Powder Technol. 29 (2)
(2018) 416–425.
for optimizing the grinding efficiency. However, it only plays a minor
[15] H. Dombrowe, B. Hoffmann, W. Scheibe, Über Wirkungsweise und Einsatzmöglichkeiten
role compared to an adequate adjustment of process parameters and von Mahlhilfsmitteln, Zement Kalk Gips 11 (1982) 571–580.
grinding aid. [16] L. Sottili, D. Padovani, Effect of grinding admixtures in the cement industry, part 1,
In future research, the results from these batch experiments will be ZKG Int. 53 (10) (2000) 568–575.
[17] O. Altun, et al., Utilization of grinding aids in dry horizontal stirred milling, Powder
transferred to continuous grinding processes, where the fineness as Technol. 286 (2015) 610–615.
well as the cohesiveness of the powder increases across the mill length. [18] M. Deckers, W. Stettner, Effect of grinding aids with special consideration of the mill
Thereby, the powder flowability plays an additional role regarding axial conditions, Aufbereitungstechnik 10 (1979) 545–550.
[19] S.K. Moothedath, S.C. Ahluwalia, Mechanism of action of grinding aids in comminu-
material transport and retention time of the product inside the mill. tion, Powder Technol. 71 (1992) 229–237.
[20] W. Scheibe, B. Hoffmann, H. Dombrowe, Einige Probleme des Einsatzes von
Acknowledgement Mahlhilfsmitteln in der Zementindustrie, Cem. Concr. Res. 4 (1974) 289–298.
[21] O. Altun, H. Benzer, U. Enderle, Improving material transportation of dry horizontal
stirred mill in cement grinding, Conference Proceeding: 13th European Symposium
We acknowledge the German Federal Ministry for Economic Affairs on Comminution and Classification, Braunschweig/Germany, 2013.
and Energy (BMWi), ThyssenKrupp Industrial Solutions and BASF [22] A.W. Jenike, Storage and flow of solids, Bull. Utah Eng. Exp. Stn. 123 (1964).
[23] A. Kwade, J. Schwedes, Breaking characteristics of different materials and their effect
Construction Solutions for funding this research project (03ET1253A). on stress intensity and stress number in stirred media mills, Powder Technol. 122
(2002) 109–121.
References [24] C. Burmeister, et al., Dry grinding in planetary ball mills: evaluation of a stressing
model, Adv. Powder Technol. 29 (1) (2018) 191–201.
[1] W. Scheibe, B. Hoffmann, H. Dombrowe, Der Einsatz von Mahlhilfsmitteln als [25] J. Hanisch, H. Schubert, Zum Einfluß der Beanspruchungsgeometrie auf das
eine Möglichkeit zur Verbesserung von trockenen Feinmahlprozessen, Freiberg. Zerkleinerungsverhalten von nicht allseitig begrenzten Kornschichten, Neue
Forschungsh. A. 602 (1978) 61–70. Bergbautechnik 12 (11) (1982) 646–649.
[2] N.A. Madlool, et al., A critical review on energy use and savings in the cement indus- [26] K. Schönert, Advances in comminution fundamentals and impacts on technology,
tries, Renew. Sust. Energ. Rev. 15 (4) (2011) 2042–2060. Aufbereitungstechnik 32 (9) (1991) 487–494.
[3] A. Jankovic, W. Valery, E. Davis, Cement grinding optimisation, Miner. Eng. 17 (11−12) [27] K. Schönert, Aspects of very fine grinding, Proceedings of Challenges in Mineral
(2004) 1075–1081. Processing, Berkeley 1989, pp. 155–172 Chapter 9.
[4] F. Shi, et al., Comparison of energy efficiency between ball mills and stirred mills in [28] W. Oettel, K. Husemann, The effect of a grinding aid on comminution of fine lime-
coarse grinding, Miner. Eng. 22 (7–8) (2009) 673–680. stone particle beds with single compressive load, Int. J. Miner. Process. 74 (2004)
[5] C.C. Pilevneli, et al., Open and closed circuit dry grinding of cement mill rejects in a S239–S248.
pilot scale vertical stirred mill, Powder Technol. 139 (2) (2004) 165–174. [29] C.F. Burmeister, et al., Experimental and computational investigation of Knoevenagel
[6] O.Y. Toraman, Effect of chemical additive on stirred bead milling of calcite powder, condensation in planetary ball mills, Chem. Eng. Technol. 37 (5) (2014) 857–864.
Powder Technol. 221 (2012) 189–191. [30] B.Y. Farber, L. Knopjes, N. Bedesi, Advances in ceramic media for high energy milling
[7] O. Altun, H. Benzer, U. Enderle, Effects of operating parameters on the efficiency of applications, Miner. Eng. 22 (7–8) (2009) 704–709.
dry stirred milling, Miner. Eng. 43–44 (2013) 58–66. [31] A. Kwade, A stressing model for the description and optimization of grinding pro-
[8] G. Mucsi, Á. Rácz, V. Mádai, Mechanical activation of cement in stirred media mill, cesses, Chem. Eng. Technol. 26 (2) (2003) 199–205.
Powder Technol. 235 (2013) 163–172.