International Dairy Journal 84 (2018) 6e14

Contents lists available at ScienceDirect

International Dairy Journal

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

Application of a decanter centrifuge for casein fractionation on pilot

scale: Effect of operational parameters on total solid, purity and yield
in solid discharge
Thomas Schubert a, *, Asutay Meric a, Remko Boom b, Jo
€ rg Hinrichs a, Zeynep Atamer a
a
University of Hohenheim, Institute of Food Science and Biotechnology, Department of Soft Matter Science and Dairy Technology, Stuttgart, Germany
b
Wageningen University, Laboratory of Food Process Engineering, Wageningen, The Netherlands

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

Article history: Individual casein fractions are of growing interest because of their multifunctional applications. The
Received 21 November 2017 fractions of caseins (aS-, b- & k-) possess a wide range of bio- and techno-functional properties. Although
Received in revised form various isolation and purification methods to obtain casein fractions have been reported, there is still a
6 April 2018
need for improvement, especially on a technical scale. The aim of this study was to develop and establish
Accepted 7 April 2018
Available online 13 April 2018
a continuous process for the fractionation of caseins. The fractions were obtained from micellar casein by
means of selective precipitation. The separation process was performed using a temperature-controlled
decanter centrifuge. Enrichment of the fractions was optimised by altering the operational parameters of
the decanter (inner weir diameter (Øweir), centrifugal force (FZ), differential speed (Dn), flow rate (V)). A
95% pure casein fraction containing aS- and b-casein was obtained. The purity and yield of the k-casein
fraction were 54% and 83%, respectively.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction chemicals. The relative simplicity of the method, combined with a

suitable separation technology, enables the isolation of casein
Interest in casein and casein fractions has been growing due to fractions in large quantities. Based on the method of Law and Leaver
their various techno-functional properties, such as solubility, (2007), Post and Hinrichs (2011) produced casein fractions from
emulsification, foam formation, stability, water-binding, gelation, micellar casein powder in small pilot scale (5 L) and large pilot scale
heat and acid stability (Broyard & Gaucheron, 2015) and bio- (85 L) utilising a nozzle-centrifuge. The yield and purity of b-casein
functional properties (Korhonen & Pihlanto, 2006; Korhonen, were reported as 10% and 90%, respectively, for a process volume of
2009), as well as non-food applications such as coating agents 5 L. However, the increase of the process volume from 5 L to 85 L
(Audic, Chaufer, & Daufin, 2003) and glues (Strube, Rüdiger, & reduced the yield (8%) and the purity (85%) of b-casein due to
Bremser, 2015). higher dead volume and start-up procedures in large-scale isola-
Different methods for the isolation and purification of casein tion (Post & Hinrichs, 2011). The environmental factors, such as
fractions can be found in the literature, as reviewed by Atamer et al. temperature, pH and CaCl2 concentration, were shown to have an
(2017). Law and Leaver (2007) developed a method in which the impact on the purity and the yield of b-casein (Post, Ebert, &
fractionation of caseins was performed based on selective precip- Hinrichs, 2009). Therefore, the effect of environmental factors on
̇
itation of casein fractions in the presence of calcium ions. In this the purity and the yield of the casein fractions should be further
method, aS- and b-caseins were precipitated by adding CaCl2 so- investigated. Similarly, adapting a suitable industrial separation
lution and were separated from the k-casein enriched supernatant technology can still provide an opportunity for further improve-
using a separation technology. The fractions obtained by this ment of the yield of casein fractions.
method are suitable raw materials for food and pharmaceutical In the dairy industry, different separation technologies, e.g.,
industries, since the fractionation is realised without toxic centrifuges such as disc-stack, imperforate basket, solid-bowl and
tubular centrifuges, are used (Beveridge, 2000). Decanter centri-
fuges meet three main criteria, which are important for casein
* Corresponding author. Tel.: þ49 711 45924383. fractionation: (i) high separation performance by generating high
E-mail address: [email protected] (T. Schubert).

https://doi.org/10.1016/j.idairyj.2018.04.002
0958-6946/© 2018 Elsevier Ltd. All rights reserved.
 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14 7

g-forces up to 4500  g, (ii) higher operation capacity on a solid discharge and the time it remains in the clarification and
continuous basis compared with other industrial solideliquid drying zones. Low differential speeds should be applied for better
separators such as nozzle separators or disc stack centrifuges, and dewatering of the solid phase (Jackson, 1980). High differential
(iii) the availability of adjustable operational parameters depending speeds shorten the residence time and lead to less dewatering on
on the physical and chemical properties of the feed or the desired the beach zone, but also prevent a build-up of solids within the
product (Altieri, Di Renzo, & Genovese, 2013; Dircks & Madsen, decanter and, therefore, reduce the risk of plugging, which would
2007; Records & Sutherland, 2001a). result in decreased operating efficiency (Jackson, 1980; Records &
A decanter centrifuge can be used to separate the solid phase, Sutherland, 2001a).
containing casein precipitates, from the liquid phase, as centrate or Centrifugal acceleration is the driving force for the separation by
supernatant (Le Magnen & Maugas, 1995). It is a horizontal density. It determines the separation rate and promotes further
centrifuge that utilises a screw conveyor mounted within the bowl dewatering of compressible solids on the drying zone. These
to actively convey the solids from the centrifuge unit (Letki, 2000). compressible solids, such as casein precipitates, are compressed
Furthermore, it can be used to realise two- or three-phase sepa- further by the centrifugal forces (Lopes & Keshavarz-Moore, 2013).
rations (Records & Sutherland, 2001a). A schematic representation According to Gleiss and Nirschl (2015), the efficiency of a
of a decanter centrifuge is shown in Fig. 1A. decanter centrifuge is largely dependent on the centrifugal force,
The feed is separated by high g-forces in the cylindrical section feed rate, weir diameter and the properties of the particles them-
of the bowl (clarifying zone) and solids are transported by the selves. A model using the approach of material functions for the
screw conveyor to the conical section of the bowl (beach zone), hindered settling points and gel points was developed and vali-
where solids are dewatered further and subsequently pushed to the dated by experimental data for a plastic polymer particle (polyvinyl
solid discharge (Records & Sutherland, 2001a). The separation and chloride, PVC P 70). The developed model suggests that Stokes’ Law
dewatering depends on the properties of the feed, such as con- is not sufficient to predict the separation efficiency of a decanter
centration of solids, viscosity and particle properties, as well as centrifuge, as it does not consider the hindered settling of particles.
operation parameters, such as feed rate and differential speed This new approach indicates that separation efficiency drops
(difference in rpm between the screw conveyor and the bowl). As a significantly when a critical feed rate is exceeded.
result, the transportation of solids can be tailored to specific ap- As decanter centrifuges can be adjusted very well to different
plications (Dircks & Madsen, 2007). operation requirements, they are used for a wide variety of appli-
The separation of the phases depends on decanter design and cations and different industrial branches, such as wastewater
different operating variables, including the centrifugal force exer- treatment, refining and dewatering of petrochemicals, mineral
ted on the feed, the pond depth, the differential speed and the flow processing and vegetable oil & dairy processing (Altieri, 2010;
rate. The separation is also influenced by feed properties, such as McCarthy, 2011; Records & Sutherland, 2001b).
the concentration of solids, viscosity, particles size and distribution, Decanter centrifuges are used in the production of olive oils,
as well as the difference in density between particles and liquid because of their capacity to process large volumes in short time
(Day, 2005). Stokes’ law and the residence time can be used to spans. Within this content, an increasing number of studies on
estimate the effect of different operating variables on the separa- optimising the efficiency of decanter centrifuges have been con-
tion. The separation rate (WZ; m s1) is determined by the cen- ducted (Altieri, Renzo, & Genovese, 2013). They are also already
trifugal forces exerted on the particles, together with the properties used in dairy industries in situations where total solid concentra-
of the particles and the liquid phase, such as the particle diameter tions are too high for conventional disk bowl separators. The dairy
squared (dP; m), the difference in density between the particles and industry also utilizes decanter centrifuges for three-phase separa-
the liquid phase (D9; kg m3), the angular velocity (u; rad s1), the tions, such as separation of melted butter into butter oil, buttermilk
radius of the decanter centrifuge bowl (r; m) and the dynamic and solids. Other applications include the dewatering and separa-
viscosity (h; Ns m2) of the suspension and can be calculated as tion of whey from casein curd and the separation of lactose crystals
shown in equation (1). from whey and whey permeate (McCarthy, 2011).
The aim of this study was to develop and establish a continuous
d2P  D9 2 process for the fractionation of caseins to increase the yield and
WZ ¼ u r (1)
18h purity of the obtained casein fractions. To achieve this, the separation
processes in the presented fractionation method were performed
The residence time (t, min) is defined as ratio of the process using a temperature-controlled decanter centrifuge. The effect of the
volume of the decanter centrifuge (VR, L) to the feed rate (V; L h1)
operational parameters (centrifugal acceleration (FZ), feed rate (V),
differential speed between screw conveyor and bowl (Dn) and weir
VR
t¼ (2) diameter (Øweir)) on the efficiency of the decanter operation was
V_ evaluated considering the following criteria: total solid content, yield
The process volume of the decanter is determined by the pond and casein composition of the obtained solid fractions.
depth, which is set by the inner weir diameter. At a constant feed
rate the residence time is increased for increasing pond depths 2. Materials and methods
(Records & Sutherland, 2001a). The length of the clarifying and
drying (beach) zone is also influenced by the inner weir diameter 2.1. Production of micellar casein
(Fig. 1B). By increasing the inner weir diameter, the pond depth is
decreased and therefore the total process volume is reduced. The micellar casein was produced in-house using a combined
Additionally, the length of the clarifying zone is decreased, whereas microfiltration and ultrafiltration process (Fig. 2) (Kersten, 2001),
the length of the drying zone or beach zone is increased, leading to which has been described previously in detail (Thienel et al., 2017).
a higher total solid content in the solid discharge, but also less Skim milk was concentrated using a microfiltration process (pore
clarified liquid phases. For better clarification of the liquid phase, a size: 0.1 mm). To reduce the whey protein content, the skim milk
smaller inner weir diameter is more appropriate. concentrate was subjected to a diafiltration process using ultrafil-
The differential speed between screw conveyor and bowl de- tration permeate. The obtained whey protein reduced concentrate
termines the rate at which the solid phase is pushed towards the was then further concentrated (pore size: 0.1 mm) and subsequently
8 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14

spray-dried (Post & Hinrichs, 2011). The ultrafiltration permeate according to VDLUFA C 10.6.8 (VDLUFA, 2011). Lactose content of
was obtained by subjecting the microfiltration permeate to an ul- the micellar casein powder was determined in a 10% (w/w) solution
trafiltration process (10 kDa cut-off). using a LactoScope FTIR Advanced FTA-3.0 (Delta Instruments B.V.,
Drachten, Netherlands). Total solid content was determined via
2.2. Decanter centrifuge weight loss after drying 3 g of powder for 5 h at 102  C according to
the method VDLUFA C 35.6 (VDLUFA, 2011).
A decanter centrifuge (model MD 80-S, Lemitec GmbH, Berlin, The total casein content and the separated casein fractions
Germany) was used for the separation steps in the presented casein were determined by reversed-phase high performance liquid
fractionation method (section 2.3.; Fig. 3). The decanter bowl has a chromatography (RP-HPLC) (Agilent 1260 Infinity, Agilent Tech-
diameter of 80 mm and a length of 308 mm. The conical section is nologies System, Santa Clara, CA, USA), equipped with a quater-
angled at 7. Weir diameters of 56, 60, 64 and 68 mm can be nary pump (Agilent 1260 Series, G1311B), auto sampler (Agilent
selected in order to adjust the pond depth, with a clearance be- 1260 Series, G1329B), thermostat with a column compartment
tween the bowl and screw conveyor of 0.5 mm. The decanter (Agilent 1290 Series, G1330B; Agilent 1260 Series, G1316B) and
centrifuge can generate a maximum centrifugal force of 4400  g at a variable wavelength UV/VIS detector (Agilent 1260 Series,
10,000 rpm and differential speeds between 1 and 200 rpm. G1315D). The reversed-phase column VDSpher OptiBio Pur
Depending on the feed solution it can be operated at feed rates 300C4-SE (150  4.6 mm) with a particle size of 5 mm and a pore
between 5 and 100 L h1 with a maximum particle density of diameter of 300 Å (VDS Optilab, Berlin, Germany) was used. For
5.0 g m L1 and particle diameters of up to 2 mm. Operation the analysis, a solvent gradient (shown in Table 2) at a flow rate of
environmental and product temperatures range from 5 to 35  C and 0.7 mL min1 was chosen. Eluent composition was 1% (v/v)
5e80  C, respectively. acetonitrile (ACN) and 0.1% (v/v) trifluoroacetic acid (TFA) in
demineralised water for Eluent A and 1% (v/v) demineralised
2.3. Fractionation on technical scale water and 0.072% (v/v) TFA in ACN for Eluent B. The column
temperature was set at 30  C and the detection wavelength was
For the fractionation of bovine milk on technical scale, the set at 214 nm. The dried casein fraction samples were dissolved
process of Post et al. (2009) was customised for higher protein and liquid samples were diluted (at least 2:1) in a buffer solution
contents (7%, w/w) in the starting suspension. Therefore, recon- of 6 M guanidine hydrochloride, 0.02 M dithiothreitol and 0.005 M
stitution time, temperature and calcium concentration for precip- sodium citrate and allowed to incubate for 1 h at room temper-
itation were adjusted for maximum yields of casein precipitates ature (Bordin et al., 2001) before the analysis. Casein standards,
after the separation steps 1 and 3 (Fig. 3; data not shown). For the C6780 a-casein, C6905 b-casein and C0406 k-casein, were pur-
purposes of this study, which were to ascertain the effects of cen- chased from SigmaeAldrich (St. Louis, MO, USA) and used as
trifugal force, flow rate, differential speed and inner weir diameter standards. Whey protein content was determined via RP-HPLC
on total solid, purity and yield of the solid discharge, only separa- according to ISO 13875jIDF 178:2005 (ISO, 2005). All analyses
tion steps 1 and 3 were carried out and can be seen in Fig. 3. were performed as triplicates.
The casein suspension was prepared in aliquots of 20 kg to a
final casein concentration of 7% (w/w) with demineralised water at
2.5. Calculations
30  C and stirred for 2 h. The pH was increased from its native pH to
pH 11 with a 5 M sodium hydroxide solution and readjusted after
The purity of the casein fraction (Pn-CN; %) was defined as the
15 min. Calcium chloride was added to a final concentration of
ratio of casein fraction content (Cn-CN, fraction) to the total casein
0.05 M and stirred for 10 min. The pH was adjusted with 5 M hy-
content in the isolated fraction (Ctotal-casein, fraction) as determined by
drochloric acid to 7 and readjusted after 15 min. The precipitated
RP-HPLC.
suspension was transferred to the feed tank and fed to the decanter
centrifuge at feed rates of 4.0 and 9.0 L h1. The precipitated caseins cnCN; fraction
(aSþb-casein) were separated from the k-casein-rich liquid phase PnCN ¼ $100 % (3)
Ctotalcasein; fraction
using the decanter at centrifugal forces between 2000 and
4400  g. The differential speed was set between 10 and 50 rpm The fraction-based yield (YnCN=total nCN ) was defined as the
and the inner weir diameter was set at 56, 64 and 68 mm. After a ratio of the obtained and freeze-dried casein fraction to the total
separation, the total solid content of the solid discharge was amount of casein fraction (as determined by RP-HPLC) in the
determined and the sample was freeze-dried for further analysis. applied micellar casein powder.
The liquid discharge was collected and used for separation 3 (Fig. 3)
to generate a k-casein-enriched precipitate. The suspension was mnCN; fraction $PnCN
YnCN=total nCN ¼ $100 % (4)
acidified to a pH of 3.8 using 5 M hydrochloric acid and readjusted mMCN $CnCN; MCN
after 15 min. The precipitated casein fraction was separated from
The protein-based yield was defined as the ratio of the obtained
the liquid by the decanter centrifuge at feed rates of 4.0 and
and the freeze-dried casein fraction to the total amount of protein
9.0 L h1. The centrifugal force and differential speed were set be-
(as determined by the Dumas method) in the applied micellar
tween 2000 and 4400  g and 10e50 rpm, respectively.
casein powder.
2.4. Analysis mnCN; fraction $PnCN
YnCN=total protein ¼ $100 % (5)
mMCN $Ctotal protein; MCN
The protein content was determined by a nitrogen analyser
(Dumatherm-DT, C. Gerhardt GmbH & Co. KG, Ko € nigswinter, Ger- The total solid content (TS) was defined as the ratio of the mass
many) following the ISO 14891jIDF 185:2002 Dumas method (ISO, of the sample after drying (mdried sample) to the mass before drying
2002). A transformation factor of 6.38 for milk and milk proteins (msample)
was used. The fat content was determined according to VDLUFA C
15.3.6 (VDLUFA, 2011). Calcium content was determined by com- mdried sample
TS ¼ x 100 % (6)
plexometric titration with ethylenediaminetetraacetic acid (EDTA) mSample
 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14 9

2.6. Statistical analysis

Figures show the mean values and the standard deviations for
triplicate analyses. An analysis of variance (ANOVA) and Tukey's
range test were performed. A value of P < 0.05 was considered
statistically significant. The statistical analysis was performed using
SigmaPlot 12.5 (Systat Software Inc., San Jose, CA, USA).

3. Results and discussion

3.1. Micellar casein powder

The results of the chemical analysis of the micellar casein

powder are shown in Table 1. The casein content was 68.8% casein
(total protein content ¼ 70.9%) with only very little whey protein
(0.4%) present. The calculated ratio of the aS-, b-, and k-casein
fractions was 3.3:3.1:1. These obtained values are in accordance
with the previous results for the production of micellar casein (Post
et al., 2009). The difference between the protein and the casein
content could be due to casein fragments that are not detected by
the RP-HPLC. Non-proteinogenic nitrogen sources also contributed
to this difference.

3.2. Fractionation on technical scale and influence of the operating Fig. 2. Process for the production of micellar casein. UF: ultrafiltration, MF:
parameters of the decanter centrifuge on the separation process microfiltration.

Unless specifically stated otherwise, the results presented refer

to results for the precipitate (aSþb-casein) obtained in separation 1. (Dn ¼ 10 rpm) and a weir diameter (Øweir) of 56 mm, whereas a
The total solid content (TS) of the solid discharge was influ- decrease of 1.2% (w/w) and 2.6% (w/w) in TS was recorded at higher
enced by centrifugal force, differential speed, feed rate and weir differential speeds of Dn ¼ 30 and Dn ¼ 50, respectively. The TS of
diameter. Fig. 4A illustrates the non-linear increase of the TS with the solid discharge obtained was also adversely affected by an in-
increasing centrifugal force for separation 1 from 20.4% (w/w) at crease of the differential speed. With a higher differential speed,
2000  g to 37.1% (w/w) at 4400  g. A higher centrifugal force the precipitate spent less time on the drying zone of the decanter
leads to a faster separation (Equation (1)) and better dewatering and the time for dewatering was reduced. This was also shown by
of the solid discharge. This effect was also observed by Jiang and Kinnarinen and Ha €kkinen (2013). The magnitude of this effect was
Xu (2015). The obtained TS for separation 3 showed the same dependent on the flow rate and is shown in Table 3. An increase of
trend as separation 1, but with much lower TS, ranging from 10 to weir diameter from 56 to 64 mm seemed to have only minor effects
25%. This can be attributed to the lower protein content, ranging on the TS, as only a small increase was recorded. However, the
between 0.4 and 2.5%. change to a weir diameter of 68 mm had a much larger effect,
An increase of the feed rate resulted in a decrease of the TS of the increasing the TS content of solid discharge by up to 13%. The
solid discharge, which is in agreement with Equation (2), where a magnitude of this change was interdependent on the flow rate and
higher feed rate results in lower residence times. This effect was the differential speed. These results are supported by previous
also observed by Jiang and Xu (2015) and Day (2005). Furthermore, findings in the literature (Day, 2005; Peeters & Weis, 2004).
the magnitude of the change seemed to be dependent on the set As stated, there are two major factors that influence the sepa-
weir diameter, differential speed and feed rate. As shown in Table 3, ration within a centrifuge; the separation rate, which is determined
an increase of the flow rate from 4.0 to 9.0 L h1 resulted in a minor by the centrifugal force and residence time, which is influenced by
decrease of 0.1% (w/w) in TS for low differential speeds the flow rate, the differential speed and the weir diameter. The

Fig. 1. Schematic representation of (A) a decanter centrifuge: 1, bowl with openings for solid and liquid discharge; 2, screw conveyor with feed inlet; 3, solid discharge, 4; drying
zone or beach zone; 5, clarifying zone or pond zone; 6, liquid discharge; 7, pitch; 8, feed and (B) the decanter centrifuge and the effect of weir diameter (Øweir) on clarifying zone
(red) and drying zone (yellow).
10 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14

Fig. 3. Process for the production of a combined aS-þb-casein fraction and an enriched k-casein fraction using micellar casein. CN: casein.

recorded change in the TS of the solid discharge, caused by the solids, such as the precipitated casein particles, were further
centrifugal force, largely matched the expected results. By compressed and reached a “gelling point” upon which they formed
increasing the centrifugal force, the angular velocity also increased, a continuous gel within the decanter bowl, which was not trans-
leading to increased separation rates and higher TS in the solid ported by the screw conveyer to the solid discharge and led to
discharge. However, at higher centrifugal forces, compressible blockages. This phenomenon was mainly observed for centrifugal
forces above 3500  g.
By increasing the flow rate or increasing the differential
speed, the amount of time the suspension was located within
Table 1
Chemical composition of the micellar casein powder.
the decanter directly decreased; in other words, was conveyed
to the solid discharge more quickly. This can be seen in Equation
Component Composition (%)
(2). Therefore, the precipitates had less time to be separated
Total solids 95.6 from the liquid phase and dewatered; therefore, the TS content
Protein content 70.9 in solid discharges decreased. By adjusting the inner weir
Casein content 68.8
diameter, the pond depth changed, and thereby the total volume
Whey protein content 0.4
Fat content 1.5 held within the decanter changed, as well as the length of the
Calcium content 2.6 drying zone (e.g., with lower total volume and longer drying
Lactose content 12.3 zones for larger weir diameters). The lower total volume within
the decanter would suggest a lower TS in the solid discharge.
However, this effect was overshadowed by the lengthening of
Table 2 the drying zone.
Solvent gradient for the determination of the casein content using reversed phase- Interestingly, the magnitude of the recorded changes observed
high performance liquid chromatography.
in the TS content suggest an interdependence of centrifugal force,
Time (min) Eluent A (%) Eluent B (%) feed rate, differential speed and weir diameter. While an increase in
0.0 72.0 28.0 the differential speed adversely affected the TS for all parameter
21.5 62.4 37.6 combinations tested, the magnitude of this effect increased with
22.5 62.4 37.6 increasing flow rate. The interdependency could also be seen for
26.0 54.0 46.0
different weir diameters (for constant feed rates) and different feed
28.0 0.0 100.0
29.0 0.0 100.0 rates (for constant weir diameters), as the magnitude of the change
30.0 72.0 28.0 in TS decreased with higher differential speeds at different weir
35.0 72.0 28.0 diameters but increased with higher differential speeds at different
 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14 11

Fig. 4. Effect of centrifugal force (FZ) on (A) the total solid content (TS) and (B) the yield (Y) of solid discharge obtained in separation 1 (differential speed (Dn) ¼ 30 rpm; feed rate
_ ¼ 4.0 L h1; weir diameter (Øweir) ¼ 56 mm).
(V) DM; fraction-based yield (Yasþb-CN/total asþb-CN); protein-based yield (Yasþb -CN/total protein).

Table 3
Total solid content (TS) and fraction-based yield (Yasþb-CN/total asþb-CN) of the solid discharge from separation 1.a

Øweir (mm) Dn (rpm) TS (%) Yasþb-CN/total asþb-CN (%)

V_ (4.0 L h1) V_ (9.0 L h1) V_ (4.0 L h1) V_ (9.0 L h1)

a c d
56 10 29.6 29.5 76.8 95.5e
30 27.3a 26.1b 76.2d 97.5e
50 26.1a 23.5a 81.1d 57.2b,c
64 10 30.6a 29.6c 65.3c 79.8d
30 26.5a 25.6b 35.5a 62.6c
50 25.7a 23.7a 77.8d 32.5a
68 10 43.0c 38.0d 41.5a,b 63.0c
30 39.2b,c 33.4d 52.5b 55.6b,c
50 35.4b 30.1c 71.8d 49.1b
a _ ¼ 4.0 and 9.0 L h1; wir diameters (Øweir) ¼ 56, 64, 68 mm. Different
Separation 1: centrifugal force (FZ) ¼ 3000  g; differential speeds (Dn) ¼ 10, 30, 50; feed rates (V)
superscript letters within a column indicate significant difference (P < 0.05).

flow rates. This indicates a multifactorial dependence of those pa- protein in the applied micellar casein powder was used for the
rameters on the TS of the solid discharge. These results are in protein-based yield (Yn-CN/total protein). The different values for the
agreement with separation theory (Equations (1) and (2)) and respective yields of the obtained casein fraction originate from
different literature sources (Day, 2005; Jiang & Xu, 2015; those reference points (total protein content and total casein con-
Kinnarinen & Ha €kkinen, 2013). tent in applied micellar casein powder). Unless specifically stated
The yield for the obtained aSþb-casein fraction did not show the otherwise, the presented results refer to the fraction-based yields.
same increase as the TS with increasing centrifugal force within the The effects of the operational parameters on the yield for each
chosen experimental parameters. As shown in Fig. 4B, an increase obtained fraction were more difficult to predict, as the precipitated
was detected up to the maximum yield of 76.7% at 4000  g and casein particles varied in size, shape, and density, and furthermore,
then the yield decreased upon further increase of centrifugal force. were compressible. Increasing the centrifugal forces from 2000 to
The influence of the flow rate, the differential speed and the 4000  g led to higher yields, since the separation rate increased
inner weir diameter on the yield was found to be largely dependent with increasing angular velocity. This effect can be explained by
on the respective other parameters (flow rate, differential speed using Equation (1) and was also observed by Gleiss and Nirschl
and inner weir diameter). The calculated yields did not only differ (2015). At higher centrifugal forces (higher than 4000  g), the
in their absolute values, but also showed different trends when the effect of the increased separation rate might be overshadowed by
other operational parameters were set at different values. This ef- greater shear forces and stronger turbulences caused by the high
fect was observed for increasing differential speeds at different angular velocity. Additionally, a built up of solids that ultimately led
flow rates. While an increase in the yield was recorded for to a blockage might have occurred during the experiment. This
increasing differential speeds from 10 to 50 rpm at a flow rate of could explain the decreasing yield when centrifugal forces of
4.0 L h1, the recorded yield decreased once the differential speed 4000  g were exceeded. To better understand this phenomenon,
was increased above 30 rpm at a feed rate of 9.0 L h1. The results more experiments at different “secondary parameters” (i.e., Øweir, V
for different parameter combinations are shown in Table 3. and Dn) should be considered.
The protein- and fraction-based yields were calculated accord- The effect of the pond depth (set by the chosen weir diameter)
ing to Equations (4) and (5), respectively. The effect of the opera- was the easiest to predict. Smaller weir diameters led to longer
tional parameters was identical for both calculated yields, as only clarifying zones (Fig. 1B) and longer residence times within the
the point of reference was different. The total amount of casein decanter (Equation (2)). This allowed for a better separation of solid
fraction in the applied micellar casein powder was used for the and liquid phases, and therefore, a better yield for the solid
fraction-based yield (Yn-CN/total n-CN), whereas the total amount of discharge and less turbid liquid discharge (Table 2; Fig. 5). This has
12 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14

Fig. 6. Reversed-phase high performance liquid chromatography profiles of: (A)

micellar casein (B) aSþb-casein fraction (C) k-casein fraction [obtained from separation
1 and 3: centrifugal force (FZ) ¼ 3000  g; differential speed (Dn) ¼ 10 rpm; feed rate
(V) ¼ 4.0 L h1; weir diameters (Øweir) of 56 mm (green), 64 mm (blue) and 68 mm
Fig. 5. Turbidity change in the samples in the liquid discharge during separation 1 by (red)].
decanter centrifuge operated with: a centrifugal force (FZ) ¼ 3000  g; differential
_ ¼ 4.0 L h1; weir diameters (Øweir) of (A) 56 mm (B)
speed (Dn) ¼ 50 rpm; feed rate (V)
64 mm and (C) 68 mm. content of 70.9% and a ratio of aSþb-casein fraction to k-casein
fraction of 6.4:1 (which corresponds to approximately 86% aSþb-
also been reported by Beveridge, Harrison, and Gayton (1992), Day casein fraction). When compared with the raw material, an increase
(2005) and Gleiss and Nirschl (2015). So even with lower TS in the in the protein content and an enrichment of the desired casein
solid discharge, more of the desired solid fraction was recovered. fraction were achieved. The minor differences in the purity implies
Interestingly, higher flow rates seemed to result in higher yields, as that the precipitation itself was mainly responsible for the purity of
long as the decanter's capacity to sufficiently separate the two the desired fraction. Even with higher moisture contents and lower
phases was not exceeded. This phenomenon was also predicted and TS, implying theoretically higher impurity, the relatively small
observed by Gleiss and Nirschl (2015). It is therefore critical to find amount of k-casein in the applied powder was not sufficient to
these optimal parameter combinations to maximise the effective- heavily influence the purity of the obtained aSþb-casein fraction.
ness of the process. However, this did not apply to the k-casein fraction obtained,
The effects of differential speed and flow rate showed a strong where the protein content ranged between 83.4 and 85.4% and the
influence on each other, and can cause major decreases in the achieved fraction purities were in the range of 46.2e51.8%, as
separation efficiency. For parameter combinations (mainly flow shown in Table 4. The results presented in this study showed almost
rate and differential speed) that did not overload the decanter's a fourfold increase in the purity of the k-casein fraction when
capacity to separate the precipitate from the liquid phase, a minor compared with the composition of the micellar casein powder and
or a major increase in the yields were recorded for increasing dif- ȧ major improvement when compared with the achieved purity
ferential speeds with Øweir ¼ 56 mm and 68 mm, respectively, since
more precipitate was pushed to the solid discharge. If, however, a
critical residence time was exceeded, the yield decreased drasti-
cally (Table 2). The feed was not sufficiently separated and the
casein precipitate was washed out with the liquid phase (Gleiss &
Nirschl, 2015). It is therefore paramount to adjust residence times
above this critical point. For the optimised operational parameters
(FZ ¼ 3000  g; Dn ¼ 30 rpm; V ¼ 9.0 L h1; Øweir ¼ 56 mm), the
output for solid and liquid discharge is 2.7 and 5.7 kg h1 and a
protein content of 20.4 and 0.4%, respectively.
The purity of the obtained fractions was determined using RP-
HPLC and calculated as shown in Equation (3). The chromato-
graphic profile of the applied micellar casein powder, as well as the
aSþb-casein fraction and k-casein fraction, obtained with different
weir diameters are shown in Fig. 6. The TS had no significant effect
on the purity or the total protein content of the collected samples. A
purity varying from 95.3 to 96.0% and a total protein content
Fig. 7. Effect of total solid content in the solid discharge on the purity of the fraction
ranging from 79.7 to 80.2% for the obtained combined aSþb-casein (PaSþb-CN) and the total protein content (ctotal protein, fraction) for aSþb-casein fraction
fraction were achieved for all parameter combinations. These re- obtained in separation 1. Purity and total protein content in the feed suspension is
sults are shown in Fig. 7. The micellar casein powder had a protein shown at 10% TS; PaSþb-CN; ctotal protein, fraction.
 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14 13

Table 4
Total solid content (TS), total protein content (ctotal protein, fraction), casein content (ctotal casein, fraction), purity (Pk-CN), protein-based yield (Yk-CN/total protein) and fraction-based yield
(Yk-CN/total k-CN) for k-casein isolation.a

Øweir in mm TS (%) ctotal protein, fraction (%) ctotal casein, fraction (%) Pk-CN (%) Yk-CN/total protein (%) Yk-CN/total k-CN (%)

56 18 83.4 56.4 53.8 11.8 82.9

64 22 85.4 52.1 48.2 5.5 38.7
68 25 84.5 60.8 55.3 4.9 34.6
a
Separation 3: centrifugal force (FZ) ¼ 3000  g; differential speed (Dn) ¼ 30 rpm; feed rate (V) _ ¼ 4.0 L h1; weir diameters (Øweir) ¼ 56, 64, 68 mm. Values for purity (Pk-CN),
protein-based yield (Yk-CN/total protein) and fraction-based yield (Yk-CN/total k-CN) from Post and Hinrichs (2011) were 31%, 2% and 11%, respectively.

and yield by the process proposed by Post and Hinrichs (2011), in Atamer, Z., Post, A. E., Schubert, T., Holder, A., Boom, R. M., & Hinrichs, J. (2017).
Bovine b-casein: Isolation, properties and functionality. A review. International
which a nozzle separator was used.
Dairy Journal, 66, 115e125.
Audic, J. L., Chaufer, B., & Daufin, G. (2003). Non-food applications of milk compo-
4. Conclusion nents and dairy co-products: A review. Lait, 83, 417e438.
Beveridge, T. (2000). Centrifugation: Large-scale centrifugation. In I. Wilson,
C. F. Poole, & M. Cooke (Eds.), Encyclopedia of separation science (pp. 320e329).
The aim of this study was to develop a continuous process for the Oxford, UK: Academic Press.
fractionation of caseins using a decanter centrifuge. In this study, a Beveridge, T., Harrison, J. E., & Gayton, R. R. (1992). Decanter centrifugation of apple
mash: Effect of centrifuge parameters, apple variety and apple storage. Food
decanter centrifuge was shown to be suitable for the intended Research International, 23, 125e130.
process, as several steps have been realised with good results. The Bordin, G., Cordeiro Raposo, F., De La Calle, B., & Rodriguez, A. R. (2001). Identifi-
separation of a pure aSþb-casein fraction with a good yield and a k- cation and quantification of major bovine milk proteins by liquid chromatog-
raphy. Journal of Chromatography A, 928, 63e76.
casein-enriched fraction were achieved. The purity and yield for the Broyard, C., & Gaucheron, F. (2015). Modifications of structures and functions of
k-casein obtained were improved when compared with previous caseins: A scientific and technological challenge. Dairy Science & Technology, 95,
findings (Post & Hinrichs, 2011). The effects of different operational 831e862.
Day, N. (2005). Centrifuge focus: Evaluation, testing and optimisation. Filtration &
parameters on the total solid content, the yield and the purity were Separation, 42, 22e24.
examined, and the relationships between operating parameters and Dircks, K., & Madsen, B. (2007). Decanter centrifuge with a screw conveyor having a
the target values were identified. To achieve maximum yield of a varying pitch. Alfa Laval Copenhagen A/S, Denmark. US Patent 7156801B2.
Gleiss, M., & Nirschl, H. (2015). Modeling separation process in decanter centrifuges
desired fraction, process parameters should be chosen as such:
by considering the sediment build-up. Chemical Engineering & Technology, 38,
moderate feed rates, small weir diameters, moderate to low differ- 1873e1882.
ential speeds and moderate to high centrifugal forces. For high total ISO. (2002). Milk and milk products e determination of nitrogen content e routine
solid content, low feed rates, large weir diameters, low differential method using combustion according to the Dumas principle (ISO 14891jIDF 185:
2002). Geneva, Switzerland: International Standardisation Organisation.
speeds and high centrifugal forces should be applied. ISO. (2005). Liquid milk e determination of acid-soluble beta-lactoglobulin content e
However, more information is needed to identify and predict reverse-phase HPLC method (ISO 13875jIDF 178:2005). Geneva, Switzerland: In-
critical residence times and optimal operating conditions for better ternational Standardisation Organisation.
Jackson, J. F. (1980). Solid bowl decanter centrifuges of the scroll discharge type. US
separation efficiency. Additionally, cleaning steps could be imple- Patent 4240578A.
mented to increase the purity of the k-casein-enriched fraction Jiang, J., & Xu, G. (2015). Application of decanter centrifuges on sludge dewatering in
obtained. Total solid and protein content in the starting suspension, urban lakes. Chinese Journal of Environmental Engineering., 9, 453e458.
Kersten, M. (2001). Proteinfraktionierung mittels Membrantrennverfahren. (Protein
as well as different operating temperatures might also affect the fractionation by membrane separation). PhD Thesis. Technical University of
separation procedure and could be adjusted to further increase the Munich. Germany: Faculty of Science Weihenstephan for Nutrition, Land Use
separation efficiency of the decanter. These should be studied in and Environment.
Kinnarinen, T., & Ha €kkinen, A. (2013). Experimental study on the influence of
more depth. selected process variables on the separation of a fine particle suspension with a
In future work, the suitability of the decanter centrifuge for a pilot scale decanter centrifuge. Particulate Science & Technology, 31, 603e611.
cold extraction, and subsequently, separation at low temperatures Korhonen, H. J. (2009). Bioactive milk proteins and peptides: From science to
functional applications. Australian Journal of Dairy Technology, 64, 16e25.
will be examined and the three fractions will be generated as
Korhonen, H., & Pihlanto, A. (2006). Bioactive peptides: Production and function-
intended in Fig. 3. ality. International Dairy Journal, 6, 945e960.
Law, A. J. R., & Leaver, J. (2007). Methods of extracting casein fractions from milk and
caseinates and production of novel products. Hanna Research Institute. assignee.
Acknowledgements WO Pat. No. 03003847.
Le Magnen, C., & Maugas, J. J. (1995). Method for obtaining beta-casein. Eurial - Parc
This research project was supported by the German Ministry of Club du Perray, assignee. US Pat. No. 5397577.
Letki, A. G. (2000). Centrifugation: Decanters. In I. Wilson, C. F. Poole, & M. Cooke
Economics and Energy (via AiF) and the FEI (Forschungskreis der (Eds.), Encyclopedia of separation science (pp. 1e10). Oxford, UK: Academic Press.
Ern€ahrungsindustrie e.V., Bonn) Project AiF 18816 N. The authors Lopes, A. G., & Keshavarz-Moore, E. (2013). Ultra scale-down device to predict
would like to thank the staff of the Dairy for Research and Training, dewatering levels of solids recovered in a continuous scroll decanter centrifuge.
Biotechnology Progress, 29, 1494e1502.
as well as Fabian Heck, Birgit Greif and Daniel Tritschler at the McCarthy, O. J. (2011). Plant and equipment. Centrifuges and separators: Applica-
University of Hohenheim for their analytical and technical assis- tions in the dairy industry. In J. W. Fuquay, P. F. Fox, & P. L. H. McSweeney (Eds.),
tance. Dr. Stefan No€ bel and Johannes Scha€fer are acknowledged for Encyclopedia of dairy sciences (2nd ed., pp. 175e183). San Diego, CA, USA: Aca-
demic Press.
their critical discussions. We also would like to thank Thomas Veer Peeters, B., & Weis, S. (2004). Relationship between pool depth and internal
at GEA Westfalia Separator, as well as Christian Otto and his team at washing on the beach of a solid bowl decanter centrifuge. Filtration & Separa-
Lemitec for their valuable suggestions and support. tion, 41, 36e40.
Post, A. E., Ebert, M., & Hinrichs, J. (2009). b-casein as a bioactive precursor - pro-
cessing for purification. Australian Journal of Dairy Technology, 64, 84e88.
References Post, A. E., & Hinrichs, J. (2011). Large-scale isolation of food-grade b-casein.
Milchwissenschaft, 66, 361e364.
Altieri, G. (2010). Comparative trials and an empirical model to assess throughput Records, A., & Sutherland, K. (2001a). Decanter design. In A. Records, &
indices in olive oil extraction by decanter centrifuge. Journal of Food Engineering, K. Sutherland (Eds.), Decanter centrifuge handbook (pp. 15e119). Amsterdam,
97, 46e56. The Netherlands: Elsevier.
Altieri, G., Di Renzo, G. C., & Genovese, F. (2013). Horizontal centrifuge with screw Records, A., & Sutherland, K. (2001b). Applications. In A. Records, & K. Sutherland
conveyor (decanter): Optimization of oil/water levels and differential speed (Eds.), Decanter centrifuge handbook (pp. 121e146). Amsterdam, The Netherlands:
during olive oil extraction. Journal of Food Engineering, 119, 561e572. Elsevier.
14 T. Schubert et al. / International Dairy Journal 84 (2018) 6e14

Strube, O. I., Rüdiger, A. A., & Bremser, W. (2015). Buildup of biobased adhesive VDLUFA. (2011). Verband Deutscher Landwirtschaftlicher Untersuchungs- und For-
layers by enzymatically controlled deposition on the example of casein. Inter- schungsanstalten (Hrsg.): Handbuch der Landwirtschaftlichen Versuchs- und
national Journal of Adhesion and Adhesives, 63, 9e13. Untersuchungsmethodik (VDLUFA- Methodenbuch), Bd. VI. Chemische, physikali-
Thienel, K. J. F., Holder, A., Schubert, T., Boom, R. M., Hinrichs, J., & Atamer, Z. (2017). sche und mikrobiologische Untersuchungsverfahren für Milch, Milchprodukte und
Fractionation of milk proteins on pilot scale with special focus on beta-casein. Molkereihilfsstoffe. Darmstadt, Germany: VDLUFA Verlag.
International Dairy Journal, 79, 73e77.