Faculteit Bio-ingenieurswetenschappen Academiejaar 2010 – 2011 Methods for monitoring fat crystallization under shear for margarine applications Elien Verstraete Promotor: Prof. dr. ir. Koen Dewettinck Tutors: ir. Nathalie De Clercq dr. ir. Veerle De Graef Masterproef voorgedragen tot het behalen van de graad van Master in de bio-ingenieurswetenschappen: Levensmiddelenwetenschappen en voeding
81
Embed
Methods for monitoring fat crystallization under shear for … · 2012-03-14 · Faculteit Bio-ingenieurswetenschappen Academiejaar 2010 – 2011 Methods for monitoring fat crystallization
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Faculteit Bio-ingenieurswetenschappen
Academiejaar 2010 – 2011
Methods for monitoring fat crystallization under shear for margarine applications
Elien Verstraete Promotor: Prof. dr. ir. Koen Dewettinck Tutors: ir. Nathalie De Clercq dr. ir. Veerle De Graef
Masterproef voorgedragen tot het behalen van de graad van
Master in de bio-ingenieurswetenschappen: Levensmiddelenwetenschappen en
voeding
The author and promotor give permission to put this thesis to disposal for consultation and to copy
parts of it for personal use. Any other use falls under the limitations of copyright, in particular the
obligation to explicitly mention the source when citing parts out of this thesis.
De auteur en de promotor geven de toelating dit werk voor consultatie beschikbaar te stellen en delen
ervan te kopiëren voor persoonlijk gebruik. Elk ander gebruik valt onder de beperkingen van het
auteurs recht, in het bijzonder met betrekking tot de verplichting uitdrukkelijk de bron te vermelden bij
het aanhalen van resultaten van dit werk.
Gent, juni 2011
The promoter,
Prof. dr. ir. Koen Dewettinck
The author,
Elien Verstraete
Woord vooraf | I
Woord vooraf
Het is zo ver, na 5 jaar studeren en een thesis ben ik gekomen aan het einde van deze studie en het is
echt voorbij gevlogen! Ook dit jaar was een hele ervaring en heb ik immens veel bijgeleerd. Dit heb ik
natuurlijk niet op mijn eentje gedaan en daarom wil ik hier enkele mensen bedanken.
In de eerste plaats wil ik mijn promotor, prof. dr. ir. K. Dewettinck bedanken om mij de kans te geven
mij te mogen verdiepen in de margarine wereld. Het was een uitdagend onderwerp maar dit heeft er
voor gezorgd dat dit een unieke ervaring werd.
Daarnaast wil ik zeker en vast mijn tutors, Nathalie en Veerle, bedanken. Ik kon altijd op hen rekenen
en geen vraag was hen teveel. Ik wil hen ook bedanken voor de opmerkingen en tips bij het schrijven
van mijn thesis, zonder hen zou dit boekje er niet gelegen hebben.
Deze thesis zou ook niet mogelijk geweest zijn zonder de samenwerking met Vandemoortele NV. Zij
hebben mij de kans gegeven om naast het labowerk aan de universiteit ook al eens te proeven hoe het
er in de industrie aan toe gaat. Hierbij wil ik vooral Ans bedanken. Ik kon altijd terecht bij haar met al
mijn vragen en ze gaf me veel goede raad, ook bij het schrijven van mijn thesis. Ook de mensen van het
R&D labo, en in het bijzonder Joost, om mij wegwijs te maken in het labo en met te helpen bij mijn
analyses. De mensen van de pilot mag ik ook niet vergeten te bedanken om de pilot testen in goede
banen te leiden.
Verder wil ik ook de doctoraatstudenten en de laboranten bedanken voor de leuke sfeer en de gezellig
babbels in het labo, ook voor de hulp wanneer er eens iets niet lukte. In het bijzonder wil ik Benny
bedanken voor het oplossen van alle praktische zaken en voor de hulp tijdens de Schröder testen.
Ook de andere thesisstudenten mag ik niet vergeten, bedankt voor alle toffe babbels en om tussen het
‘serieuze’ labowerk ook voor wat ontspanning te zorgen. Ik wil speciaal hierbij Liesbeth bedanken voor
de vele leuke gesprekken, Ik kon altijd op je rekenen, merci hiervoor!
Zowel deze thesis als de voorbije 5 jaar zou niet gelukt zijn zonder de steun van mijn ouders. Ook de
laatste weken toen de stress zijn maximum bereikte, stonden zij klaar voor mij.
MERCI ALLEMAAL!!
Elien
Table of content | II
Table of content
Woord vooraf .............................................................................................................................................. I
Table of content ......................................................................................................................................... II
List of abbreviations ................................................................................................................................... V
List of figures ............................................................................................................................................. VI
List of tables ............................................................................................................................................ VIII
Abstract ..................................................................................................................................................... IX
Samenvatting .............................................................................................................................................. X
Figure 4.21: The evaluation of the different steps in (a) PilotFTE and in (b) PilotVDM for margarine
after 1 week ........................................................................................................................................... 56
Figure 4.22: Photographs of the loaf size of sponge cake with (a) PilotMarg1, (b) CTSU15°C_30’,
(c)CTSU15°C_1h, (d)PilotMarg2, (e)CTSU20°C_30’ and (f)CTSU20°C_1h ............................................. 58
Figure 4.23: (a) Hardness of the cakes made with different margarines and (b) the linear correlation
between hardness and the dough density ............................................................................................ 59
List of tables | VIII
List of tables
Table 2.1: Average Composition of European Type Margarines ............................................................. 3
Table 3.1: Composition of the different reference samples (Vereecken, 2010) ................................... 15
Table 4.1: Composition of the fat blends .............................................................................................. 24
Table 4.2: Fatty acid (%) profile of the different fat blends .................................................................. 24
Table 4.3: TAG (%) composition of the different fat blends .................................................................. 25
Table 4.4: Comparison of the RVA and the CTSU technique ................................................................. 37
Table 4.5: The different temperatures used on the CTSU to compare with the products on both pilot
margarine. The sponge cake batters were prepared with a Kenwood Major kneader (Kenwood,
Vilvoorde, Belgium) starting with the kneading of the margarine. The cakes were baked by placing
the batters during 45 minutes in an oven at 175°C.
3.6.2 Photographic images of the cake volumes
Sample: Slices of a thickness of one cm were taken of the cake.
Apparatus information: The photographs were taken using a Sony DSLR-A390 camera (Sony
Corporation, Tokyo, Japan).
3.6.3 Hardness
Sample: Slices of two cm were taken of each cake.
Apparatus information: The hardness was measured with a TAplus Texture analyzer (Lloyd
Instruments, Hampshire, United Kingdom) with a 500N load cell with a cylindrical probe with a
diameter of 13mm (CNS Farnell, Hertfordshire, United Kingdom).
Measurements: The probe descends at 10mm/min and goes down for 10mm with a trigger of
0,2N. The hardness is expressed as the maximum load (N) during the 10mm of penetration. Ten
repetitions were performed on each sample.
3.7 Statistical analyses
The statistical analyses were executed with the program SPSS 15.0 (Illinois, USA). The measurements
were statistically compared with a One Way Anova test. First, the Levene test was done to test the
hypothesis of equal variances. If the hypothesis was accepted, a Tukey test was used to check
significant differences on a 95% significance level. If the hypothesis was rejected, a Dunnett’s T3 was
used.
Results and discussion | 23
4 Results and discussion
The aim of this research was to develop a method to produce margarine on lab scale with similar
properties as the product on pilot scale. From the schematic overview in Figure 4.1, it can be seen
that the research was divided in three main parts; the first part was the characterization of the fat
blends and a more fundamental research of fat crystallization under shear. In the second part the
aim was to find a suitable method on lab scale and in the last part, the results of the lab scale with
the controlled temperature shearing unit (CTSU), as selected in the previous part, were compared
with these of the pilot of Vandemoortele (PilotVDM) and the pilot of FTE (PilotFTE). The research was
in cooperation with Vandemoortele NV.
Figure 4.1: The schematic overview of the research
4.1 Part 1: Crystallization under shear: fundamental study
4.1.1 Characterization of the fat blends
The fatty acid profile, the triacylglycerol (TAG) composition and the non-isothermal SFC were studied
for five selected samples. The four reference samples (Shst15, Shst20, Pst15 and Pst20) contained of
shea stearin (Shst) or palm stearin (Pst) with palm olein as diluting oil (see Table 4.1). The
composition of these samples was as such that the SFC equals 35% after 24 hours at 15°C (Shst15 and
Pst15) or 20°C (Shst20 and Pst20) (Vereecken, 2010). Palm stearin and shea stearin were used as an
alternative for some unhealthy fats, like hydrogenated fats that can contain a lot of trans
Part 1
• Chemical characterization of the fat blends
• Fundamental research
• Rheo-NMR and oscillatory rheology
Part 2
• Search alternative for production on pilot scale:
• RVA (+ post treatment)
• CTSU: influence of the rotational speed - length of the shearing phase
Part 3
• Comparison of the different scales with fat blends and margarine
• CTSU vs PilotFTE
• CTSU vs PilotVDM
Results and discussion | 24
unsaturated fatty acids, in the margarine production. The cake margarine based samples contained
palm oil and palm stearin (see Table 4.1). The composition was based on a standard cake margarine
fat.
Table 4.1: Composition of the fat blends
Shst15 75,1% palm olein + 24,9% shea stearin
Shst20 56,5% palm olein + 43,5% shea stearin
Pst15 74,5% palm olein + 25,5% palm stearin
Pst20 50,1% palm olein + 49,9% palm stearin
Margarine fat 80% palm oil + 20% palm stearin
4.1.1.1 Chemical composition
Table 4.2 presents the fatty acid profile of the different fat blends, executed by gas chromatography.
The major fatty acids in both the fat blends with shea stearin, palm stearin and margarine fat were
palmitic acid (C16:0), stearic acid (C18:0), oleic acid (C18:1) and linoleic acid (C18:2cis) but at
different ratios. The amount of saturated and unsaturated fatty acids was similar (see last line of
Table 4.2). All the fat blends contained low amounts of trans fatty acids (<1%).
Table 4.2: Fatty acid (%) profile of the different fat blends
Shst15 Shst20 Pst15 Pst20 Margfat
C16:0 31,13 21,00 45,27 50,06 46,72
C18:0 17,65 26,99 4,27 4,51 4,27
C18:1cis 40,16 40,87 39,02 34,96 37,53
C18:2cis 7,85 8,37 8,49 7,52 8,65
Others* 3,21 2,77 2,95 2,95 2,83
Unsaturated 49,04 50,66 48,54 43,42 47,01
*sum of the fatty acids with amounts < 1%
Table 4.3 gives the TAG composition of both the reference and cake margarine based samples. Shea
stearin is a source of SOS and this TAG was thus also a major component of the fat blends with shea
stearin. In these fat blends, there was also a high amount of POP and POO/SOL, the amount of
symmetric TAGs present was high. Symmetric TAGs are known to have a rather slow crystallization
rate (Timms, 2003).
Palm stearin is a source of PPP triacylglycerol and this TAG was an important component of the fat
blends with palm stearin and the margarine fat. Besides PPP, the fat blends also contained large
Results and discussion | 25
amounts of POP and POO/SOL. PPP is a seeding agent and will have an effect on symmetric
triacylglycerols like POP. PPP promotes crystallization and will cocrystallize with POP; meanwhile POP
will reduce the crystallization rate (Vereecken, 2010 and Smith et al, 2005).
In the case of margarine fat, the high amounts of palm oil and in the case of the other fat blends, the
high amounts of palm olein, were responsible for the high amount of POO/SOL. For the other blends,
the high amounts of palm olein were responsible
Table 4.3: TAG (%) composition of the different fat blends
TAG Shst15 Shst20 Pst15 Pst20 Margfat
PLL 1,60 1,67 1,98 1,59 2,00
LOO 1,63 1,70 1,86 1,70 1,86
PLO/SLL 8,28 8,59 9,56 8,33 9,30
PPL 7,74 5,72 9,65 8,58 9,04
OOO 3,79 3,80 4,26 4,50 4,18
POO/SOL 19,39 19,40 22,89 19,55 21,61
PPO/POP/PSL 22,33 11,15 29,85 29,82 28,23
PPP 1,18 0,22 7,43 13,16 9,93
SOO 3,53 5,25 2,53 2,28 2,57
PSO 7,50 7,21 5,27 5,33 5,05
PPS 0,25 ** 1,43 2,64 2,10
SOS 18,73 32,43 0,65 0,62 1,02
Others* 4,05 2,86 2,64 1,90 3,11
*sum of the TAGs with amounts < 1% **no detectable amounts
4.1.1.2 Non-isothermal SFC
Figure 4.2 shows the SFC-profile of the different fat blends. The fat blends with shea stearin melt
more quickly than the fat blends with palm stearin, which contained a higher amount of saturated
fatty acids (see Table 4.2). Shst15 had a melting point around 35°C, which was slightly lower than the
melting point of Shst20 that was about 40°C. This difference was due to the higher amount of PPP in
Shst20. The fat blend Pst20 had the highest melting point, namely between 50°C and 55°C, followed
by margarine fat (around 50°C) and then Pst15 (between 45°C and 50°C). The higher the amounts of
PPP and saturated fatty acids and the lower the amounts of POO/SOL, the higher the melting point of
these fat blends (see Table 4.2 and Table 4.3).
Results and discussion | 26
Figure 4.2: SFC as a function of temperature (°C) for the different fat blends
4.1.2 Fundamental study of crystallization under shear
In margarine production the applied shear is an important process parameter that has a significant
effect on the structure development. It is thus of paramount importance to gain insight in the
crystallization behaviour of fat systems under shear. The effect of shear on the crystallization of fat
blends will be studied by rheo-NMR and oscillatory rheology.
4.1.2.1 Rheo-NMR
With the rheo-NMR technique, the SFC was studied during one hour of shear and also compared with
the SFC without shear. This technique was applied to the four reference samples at their respective
temperatures and the results are shown in Figure 4.3. The first data point had a negative isothermal
time because the sample was cooled from 70°C to the crystallization temperature (15°C or 20°C)
during the first 2 minutes.
In Figure 4.3(a), the SFC profile of the fat blends with shea stearin is shown with and without shear.
For the fat blends Shst20 there was a fast increase of SFC to remain constant afterwards. In the
absence of shear, it took more time to reach the maximum SFC of 6%. So the crystallization at 20°C
stayed constant after 2 minutes when the fat blend underwent shear and after 4min for the sample
without shear. The crystallization process was enhanced when shear was applied (De Graef, 2009).
For the sheared samples the SFC plateau showed some fluctuations. This can be explained as
continuous crystallization and dissolving of the sample due to shearing (Vereecken, 2010). Although
the SFC seemed to have reached an equilibrium, the crystallization process still proceeded. The slow
crystallization rate is due to the fact that the fat blends were designed to reach 35% SFC after 24h of
crystallization. For the fat blends crystallized at 15°C, bigger differences were observed between the
sample with shear and the one without shear. The curve of fat blend Shst15 without shear showed
Results and discussion | 27
the same shape as the ones crystallized at 20°C. The crystallization process was here again so slow
that it seems that the crystallization reached a maximum. The fat blend Shst15 with shear had a
different shape. After one hour, the crystallization of the sample with shear was still increasing. The
crystallization rate of this sample was the highest of the four samples because it contained less
symmetric TAGs and shear enhanced the crystallization. Both samples had at the end a higher SFC
than the samples crystallized at 20°C due to a lower crystallization temperature.
In the Figure 4.3(b), the crystallization during one hour of the fat blends with palm stearin is shown,
both with and without shear. The same procedure was followed, thus the crystallization temperature
of Pst15 was 15°C and for Pst20, 20°C. The curves ended all at similar SFC, namely 35%. The
crystallization proceeded very fast. After less than one hour the expected SFC, normally after an
equilibrium time of 24h, was already achieved. This was due to the high amount of PPP in both
samples. PPP is a seeding agent and will accelerate the crystallization (Vereecken, 2010 and Smith et
al, 2005). Fat blend Pst20 reached faster the equilibrium than fat blend Pst15 which was again due to
the higher amount of PPP in fat blend Pst20 (see Table 4.3). As for the samples with shea stearin, the
curves of the samples with shear were slightly steeper in the beginning and sooner reached the
equilibrium value than the samples without shear. That can again be explained by the fact that
crystallization was enhanced by shear.
(a) (b)
Figure 4.3: Rheo-NMR of the fat blends with (a) shea stearin and (b) palm stearin
Results and discussion | 28
The curves of the fat blends with shea stearin did not reach a similar value as in the case of the palm
stearin blends (see Figure 4.3(b)) and the SFC was also much smaller than 35%. The crystallisation
rate was thus much slower than these of the fat blends with palm stearin. This was due to the large
amount of symmetric triacylglycerols, like SOS and POP and the rather small amounts of PPP (see
Table 4.3). The large amount of SOS is typically for shea stearin, this was the reason why Shst20, with
more shea stearin, had a lower SFC than Shst15. (Vereecken, 2010)
4.1.2.2 Oscillatory rheology
The crystallization under shear was also studied with rheology. Four shear rates (75s-1-150s-1-300s-1-
500s-1) were applied for 4 shear times (0min-15min-30min-60min) after which crystallization under
static conditions was further monitored up to 120min isothermal time. As a reference, the
crystallization without shear was also recorded. Although measurements were executed in triplicate,
the plotted rheological results represent one exemplary curve as very little variation was observed
between the repetitions.
The complex modulus (|G*|) in function of the isothermal time for the different fat blends are shown
for crystallization without shear in Figure 4.4. All the fat blends showed a curve in two steps. These
two steps could be related on the one hand to the polymorphic transition of α crystals in the first
step and a transformation in β’ crystals in the second step or on the other hand to fractionated
crystallization. After two hours the curves reached a similar end value, except for the fat blend
Shst15 where |G*| was still increasing. The |G*| of the samples with palm stearin started to increase
before one minute isothermal time. Pst20 increased slightly faster and reached fast the equilibrium
value. The more PPP, thus the more palm stearin present, the faster the product will crystallize and
the |G*| will increase sooner. The samples with shea stearin started to increase after 2 and 3
minutes isothermal time for respectively Shst20 and Shst15. These samples thus crystallized slower
and the end point was reached later. This slow crystallization rate is due to the high amounts of
symmetric triacylglycerols like SOS and the low amounts of PPP; which could also be concluded from
the rheo-NMR results.
Results and discussion | 29
Figure 4.4: Complex modulus (|G*|) as a function of isothermal time for all fat blends for crystallization without shear
The apparent viscosity, recorded during the shearing step at isothermal conditions, is shown as a
function of the isothermal time in Figure 4.5. Shear was applied for 15, 30 and 60 minutes at the four
shear rates. For all samples, an effect of shear rate on the initial apparent viscosity could be
observed, which was due to the fact that shear was already applied during the cooling (De Graef,
2009). The initial apparent viscosity at 75s-1 was the lowest, followed by 150s-1 and the highest initial
apparent viscosity was at the highest shear rates, namely 300s-1 and 500s-1. Upon later isothermal
times the apparent viscosity increased substantially at all shear rates. The increase was due to the
formation of primary crystals but also due to the aggregation into a crystal network (De Graef, 2009).
All fat blends showed an earlier increase in apparent viscosity at higher shear rates due to a faster
crystallization. However, the higher the shear rate, the smaller the increase in apparent viscosity,
resulting in a lower equilibrium value for the samples at higher shear rates. No large differences were
observed for shearing at 300s-1 compared to shearing at 500s1. According to the study of Tarabukina
et al (2009), 300s-1 seems to be a critical shear rate: from 300s-1 onward aggregation of the crystals is
no longer possible as the contact time between the colliding entities is too short to enable
aggregation. The same effect will happen with a shear rate of 500s-1 and that was why the curves
were very similar. The apparent viscosity increased in two steps for every fat blend, indicated by
numbers in Figure 4.5. The parts 2 and 4 represent the two increases and the parts 1, 3 and 5 show
the apparent viscosity plateaus.
For the fat blends crystallized at 15°C (Figure 4.5(a) for Shst15 and Figure 4.5(b) for Pst15) the two
steps were clearly defined. Looking at the different isothermal periods, it can be seen that after
Results and discussion | 30
15min only the first increase (2) in apparent viscosity had occurred. Consequently, a lower apparent
viscosity value was recorded after 15 min compared to longer shearing times. After 30min the
second increase (4) was just finished and the apparent viscosity was only slightly lower than after
60min. Between 30min and 60min (5) the apparent viscosity was still increasing but not as strong as
in the second step (4). The final apparent viscosity was the highest for the lowest shear rate, which is
due to the fact that shear does not only enhance the primary crystallization but also influence the
microstructural development of the fat structure. The apparent viscosity was not only influenced by
the amount of solids but also by the crystal size distribution and the crystal interactions (De Graef,
2009 and Kellens et al, 2007). The higher the shear rate, the more break down of aggregates will take
place.
(a) (b)
(c) (d) Figure 4.5: Apparent viscosity recorded during the shearing step in function of the isothermal time of (a)
Shst15, (b) Pst15 (c), Shst20 and (d) Pst20
Results and discussion | 31
For the fat blends Shst20 (Figure 4.5(c)) and Pst20 (Figure 4.5(d)) crystallized at 20°C, the two steps
were not well distinguished and the plateau between the steps (3) was not clearly visible. The final
apparent viscosity was also lower than the samples crystallized at 15°C. This can be explained by a
lower driving force for crystallization at higher temperatures. For the shea stearin blends, it could
also be due to the amounts of SOS or PPP; the sample Shst15 had a lower amount of SOS and higher
amount of PPP. As concluded before, Shst15 will crystallize faster and have a higher apparent
viscosity. Subsequent to the second apparent viscosity increase (4) of the samples crystallized at
20°C, a small decrease could be observed, indicating some structural breakdown of the primary
network due to the shear forces (De Graef, 2009). As a constant apparent viscosity was reached
before 15min, all the different shearing periods will result in similar final apparent viscosity values.
Following the sheared crystallization, monitored in terms of apparent viscosity, crystallization
proceeded under static conditions. During this static crystallization the |G*| was measured in
function of the isothermal time. The results of the fat blends with shea stearin are shown in Figure
4.6. Figure 4.6(a) and Figure 4.6(d) represent the results after 15min of shear for respectively the fat
blends Shst15 and Shst20. The shape of the curves was similar, but the increase goes on earlier and
steeper for Shst15, which was due to the lower amount of symmetric triacylglycerols in this fat blend.
As already mentioned, symmetric triacylglycerols will slow the crystallization rate. The higher the
shear rate, the steeper the increase will be. This could be explained by the shear rate that enhanced
the crystallization rate. For Shst15, shearing at 300s-1 or 500s-1 did not affect the |G*| values at the
end of the isothermal period. Both shear rates are high and as mentioned before, aggregation is not
possible during shearing. After shearing, the crystallisation and network forming continued similar.
For both blends, similar final |G*| were observed regardless of shear rates applied in the first part of
the crystallization process.
For graph (b), (c), (e) and (f) in Figure 4.6, the same trend could be observed: a steep increase
followed by a constant |G*|. After the period of shear, these samples already underwent a second
increase in apparent viscosity which was not the case after 15min of shearing, as seen in Figure 4.5.
The initial |G*| after 30min and 60min of shear was ten times higher than after 15min of shear,
more crystals or a stronger network was formed resulting in a higher |G*|. At the end of the
isothermal period, the |G*| reached the same value.
Although at the end of shear step, the samples crystallized at 15°C showed a higher apparent
viscosity compared to crystallization at 20°C, their initial complex modulus values were lower.
Results and discussion | 32
Figure 4.6: |G*| in function of the isothermal time of Shst15(d, e, f) and Shst20(a, b, c) recorded after a shear step of 15min (a, d), 30min (b, e) and 60min (c, f)
Figure 4.7 shows the crystallization under static conditions of the fat blends with palm stearin after a
shear step. Different lengths of this shear step did not result in large differences in initial |G*|,
although the final apparent viscosity after 15min of shear was much lower than after 30 and 60
minutes. The samples crystallized at 15°C and 20°C did also not show big differences in initial |G*|,
although the final apparent viscosity differed a lot at the step with shear. The apparent viscosity of
the samples at 15°C was higher than these at 20°C (Figure 4.5) but this translated into differences in
the initial |G*|. The |G*| was increasing in time, but more gradually than the samples with shea
Results and discussion | 33
stearin. It can also be observed that the higher the shear rate in the period of shear, the higher the
final apparent viscosity will be.
Figure 4.7(a), (b) and (c) show the samples crystallized at 15°C after respectively 15, 30 and 60
minutes of shear. After 15min of shear, a clear difference could be observed between the two
highest and two lowest shear rates. The samples of 75s-1 and 150s-1 started at low |G*|, increased
first fast and around 25min the slope was decreasing. For the two higher shear rates the |G*| started
higher but the first increase was not as steep. At the end, the |G*| all reached a similar value. After
30 and 60 minutes of shear the increase was more gradually and the two steps are not as visible. The
longer the period of shear, the more the curve of 150s-1 shifted to the curves of 300 and 500s-1. The
samples of 150, 300 and 500s-1 still achieved a similar end value, whereas the curve of 75s-1 ended
lower with longer periods of shear. This could be explained by the shorter time to crystallize as the
period of shear was longer, thus less crystallization has taken place, resulting in a lower |G*|.
For crystallization at 20°C (Figure 4.7(d), (e), (f)), the difference between the two highest and the two
lowest shear rates was also clearly visible. The shape of the curves was different of the samples
crystallized at 15°C. In the first part, the increase was not as steep. For the samples sheared at 75s-1
and 150s-1 for 15min and at 150s-1 for 30min a steep increase of the |G*| occurred after 90min. The
other curves did not show this second increase. After 15min of shearing the two highest shear rates
reached a similar value and the two lowest went to one. At longer times of shear, the two highest
shear rates still went to a similar value but the two lowest shear rates lie further apart.
The curves of the shea stearin and palm stearin did show a lot of differences, both the shape, begin
and final |G*| values differed between the two samples. At the end of the isothermal period, the
samples with shea stearin reached an equilibrium which was not observed for the palm stearin
blends. These blends were still increasing at the end, although the crystallization rate of the palm
stearin would be expected to be higher than for the shea stearin blends due to the high amount of
PPP in the palm stearin blend.
Results and discussion | 34
Figure 4.7: |G*| in function of the isothermal time of Pst15 (d, e, f) and Pst20 (a, b, c) recorded after a shear step of 15min (a ,d), 30min (b,e) and 60min (c,f)
Results and discussion | 35
4.2 Part 2: Crystallization under shear: applied study
The first objective was to find a suitable method to produce margarine on a lab scale as alternative for
the production on pilot scale. The first applied technique was the RVA and the RVA with a post-
treatment, this part was a continuation of the PhD of Jeroen Vereecken (2010). The second technique
that was used, was the controlled temperature shearing unit (CTSU).
4.2.1 Rapid Viscosity Analyzer (RVA)
The RVA can crystallize the fat blends under shear. Only 30ml of products is necessary instead of 20kg
on the PilotFTE or 70kg on the PilotVDM. The RVA had a similar geometry as the starch paste cell of the
rheometer; it also measures the apparent viscosity. In practice, the RVA is mostly used to investigate
the gelatinization.
The samples were crystallized at 15°c or 20°C with a shear rate of 160rpm or 75 s-1 for 15min, after that
the rotor is taken out and the can with the sample was stored under static conditions at the
crystallization temperature for two weeks. The sample was analyzed by NMR to measure SFC, Texture
analyzer to measure hardness and the appearance of grains was visually studied after one day, one
week and two weeks. The used time-temperature profile is shown in Figure 4.8.
Figure 4.8: Time-temperature profile used for the crystallization of the samples at 15°C under shear (after Vereecken, 2010)
The main disadvantage of the RVA is that only one rotational speed can be used. A long isothermal
crystallization period was not possible because at the end of the experiment, the rotor had to be
removed without damaging the fat structure. When the state of crystallization is too far, the product
will be too hard and the rotor cannot be taken out.
The results of the experiments were similar as in the one of the PhD of Jeroen Vereecken (2010). The
fat blends with shea stearin showed already grains after one day of storage. The fat blends with palm
Results and discussion | 36
stearin showed no grains after two weeks of storage. To have more crystallization during the process, a
post-treatment was used after the procedure of the RVA.
4.2.2 RVA followed with a post treatment
The RVA can be compared with a cooling step from the margarine production process and the TNO
extrusion cell with a post treatment in the pilot process. The TNO cell will give a similar treatment to
the samples as a pin worker. The extrusion cell consists of a cylinder; in this cylinder a stick with a disk
of 6 holes can be fitted. The sample, crystallized in the RVA, was brought as fast as possible in the
cylinder and the measurement in the TNO was started by moving the probe up and down for 100 or
200 times.
This procedure was carried out for the fat blend Shst20. Immediately after the whole procedure the
sample was taken out of the cylinder and there were clearly big and hard grains visible. The TNO
treatment accelerated the formation of the big grains. As seen in the RVA, the fat blends with shea
stearin tending to form β crystals during the storage. The crystallization of the β polymorph was
enhanced by the TNO cell that applied shear on the product. This method was not further developed.
4.2.3 Controlled Temperature Shearing Unit (CTSU)
A second alternative for the pilot production was the CTSU. The CTSU is normally used to temper
chocolate, but can also be used to crystallize fat and margarine samples under shear. The CTSU can
handle a range of rotational speeds of 0 rpm to 115 rpm and a big advantage is the possibility to stir
much longer than the RVA. For every batch, 1kg of sample was used; this is more than with the RVA
but still very small in comparison with the pilot productions.
4.2.4 Comparison of RVA and CTSU
To compare the samples, the dimensions of the devices should be considered. The tip speed gives an
indication of the shear forces. To compare the results, the procedure of the crystallization and the
same tip speed of the RVA were used on the CTSU. The rotational speed of the CTSU was calculated by
Equation 1.
Equation 1: (a) general formula; (b) tip speed of the RVA; (c) rotational speed of the CTSU
(
) ( )
( )
( )
(
)
( )
( )
⁄ ( )
Results and discussion | 37
Table 4.4 shows the comparison of the RVA and CTSU technique.
Table 4.4: Comparison of the RVA and the CTSU technique
RVA CTSU
30ml of sample 1l of sample
1 rotational speed: 160rmp Range of rotational speeds: 0-115rpm
Short period of crystallization under shear (15min) Long period of crystallization under shear possible
(2h)
Fast cooling: 10°C/min Fast cooling: 10°C/min
In the following tests, it was chosen to only discuss the fat blend Shst20 because it was most sensible
to the formation of grains and this was the most important problem that has to be avoided in the
margarine production. For every part, the hardness, SFC and the appearance of grains were studied
after one hour, one day, one week and two weeks of storage.
4.2.5 CTSU with similar conditions of the RVA
This part will compare the results of the RVA and the CTSU, executed with the same procedure (see
Material and methods, 3.4.1 and 3.4.2) and the same tip speed. The results are shown in Figure 4.9.
The difference in SFC between the RVA and CTSU technique were smaller than the difference in the
hardness. After one hour the SFC of the RVA was very low (lower than 10%) so during the RVA
treatment, not much sample was crystallized. This was due to the slow crystallization rate of the fat
blend by the high amounts of the symmetric triacylglycerol SOS (see before). After one day the SFC had
increased to 35%. This was also a control to see if the fat blends were made correct because they were
composed to have a SFC of 35% after 24h. After one day, one week and two weeks, the SFC increased
slightly but no differences were observed between the two techniques.
After one hour, the hardness of both techniques was rather low; no crystal network was formed yet.
After one day the hardness had increased especially for the sample produced with the CTSU. The
increase of the hardness with the RVA was not as big. After one and two weeks of storage, the
hardness continued to rise for the sample with the CTSU, although there was no difference in SFC. This
can be explained by sintering of the fat crystal network during post crystallization and this will lead to a
harder fat crystal network. Sintering is the formation of solid bridges between fat crystals and is
promoted by mismatches in the crystal network and by a more heterogeneous composition (Johansson
Results and discussion | 38
and Bergenstahl, 1995). In this case, the sample had a heterogeneous composition because the fat
blend consisted of approximately 50% of shea stearin and 50% of palm olein. In contrast, the hardness
of the RVA decreased in time thus the difference between the two hardness measurements became
bigger when the samples were stored longer. During the storage, sandiness appeared for both
samples, after one day. In the sample of the CTSU, the sandiness still appeared after one and two
weeks but in the sample with the RVA, a lot of big and hard grains appeared and their amount
increased in time. Vereecken (2010) identified these grains as fat crystals in β polymorph. This could
explain the decrease of the hardness in the samples of the RVA. The hard grains were embedded in a
soft matrix. It will be the hardness of the soft matrix that was measured instead of the hardness of the
grains. The hardness was thus underestimated.
Figure 4.9: Comparison RVA and CTSU @ 51rpm on fat blend Shst20 (NG=no grains, SA=sandiness, BG=big grains)
For the fat blend Shst15, there was a similar trend in both hardness and SFC, but the hardness was
more than two times higher. The grains formed after one week and two weeks were smaller with the
RVA technique and no grains were observed with the CTSU method. The fat blends with palm stearin
showed also the same trend in hardness. After one hour the SFC was already higher than 25%, many
crystals were formed during the treatment. This was due to the high amounts of PPP. As mentioned
before PPP is a seeding agent and it provides a fast crystallization. There was also no appearance of
grains during storage, which was due to the high amounts of palmitic acid in these fat blends which
stabilize the β’ polymorph.
Results and discussion | 39
4.2.6 CTSU - influence of the rotational speed
During storage at static conditions, the crystals have more time to form a regular crystal network and
this will favour the formation of β crystals and thus the formation of grains (Ghotra et al, 2002).To
avoid post crystallization, the crystallization had to be enhanced during the process by increasing the
shear forces. Higher shear forces will enhance the primary crystallization and influence the crystal
network (De Graef, 2009). This can be done by increasing the rotational speed. The rotational speeds
that were used to compare are 51 rpm, the same speed as before, and 115 rpm, which is the maximal
rotational speed of the CTSU. The highest rotational speed had a clear effect on the hardness that was
about three times higher than at the lower speeds, as seen in Figure 4.10. Especially after one hour,
there was a large increase of the hardness when higher speeds were applied. The crystal network was
already formed in the CTSU, which was not the case at 51 rpm. There was still a formation of some
grains at higher speeds, but the sandy feeling appeared after one week instead of after one day, like
the case of lower rotational speeds. The different rotational speeds had only a small influence on the
SFC. During the storage, sintering occurred and this will not change the SFC.
Figure 4.10: Comparison CTSU @ 51rpm and @ 115rpm on fat blend Shst20 (NG=no grains, SA=sandiness)
4.2.7 CTSU – influence of the length of the shearing phase
Another possibility to have a more completed crystallization during the process was increasing the
shear time so less post-crystallization will occur. From the oscillatory rheology (see 4.1.2.2, Figure 4.5),
it was seen that the crystallization was not completed after 15min of shear. The apparent viscosity of
the isothermal shear step at 15°C was not at the maximum after 15min and the second step had not
yet occurred. It was only after 30min that the apparent viscosity reached the maximum. Next to 15min
Results and discussion | 40
of shear, as in the previous parts, a shearing time of 30min, 1h and 2h was studied. The results of the
fat blend Shst20 is shown in Figure 4.11.
It can be concluded that the hardness decreased when longer shear times were used. The SFC did not
follow this trend and was mostly the same for the different shear times. This could be explained by
breakdown of the fat crystal structure when shearing longer. Another possibility of the lower hardness
is the intake of air bubbles during the shearing. The CTSU has a headspace with a volume as big as the
volume of the sample. During stirring, air can go from the headspace in the sample and air bubbles are
then retained by the crystal network.
The appearance of grains was also investigated. In the previous parts, it was seen that this was
especially a problem at fat blend Shst20. After already 30min of shearing, there was still no appearance
of graininess after 2 weeks of storage for this fat blend. Also in the other samples, there was no
appearance of grains.
Figure 4.11: Shst20 - longer period of shear (NG=no grains, SA=sandiness)
Both the RVA and the RVA with the TNO cell were not suitable to make a product on lab scale that is
similar the pilot scale; during the storage, grains appeared in the products of the shea stearin blend.
The second technique was the CTSU. The procedure was optimized by using higher shear rates and
longer shear times which resulted in samples without grains in the shea stearin blend during storage.
This technique was further used in part 3.
Results and discussion | 41
4.3 Part 3: Lab scale versus pilot scale
Up to now, only the pilot scale can be used to produce margarine products similar to industrial scale
but these tests need relatively large amount of fat ingredients and time. To do more tests with less
sample in less time, a lab scale process is needed. In the previous part, an alternative on lab scale was
found namely the CTSU. This part presents the comparison between the results of the CTSU and two
pilot scale processes: the Pilot of FTE (PilotFTE) on the one hand and the Pilot of Vandemoortele
(PilotVDM) on the other hand. Four blends were used in this evaluation: two reference samples
(Shst20 and Pst20) and two cake margarine based samples (margarine fat and margarine).
4.3.1 General overview of the three crystallizers and the experimental set-up
Figure 4.12 presents the set-up of each crystallizer. PilotVDM consists of a mixing tank and a high
pressure pump that forwards the emulsion to the first SSHE (CS1), after which different cooling steps
and treatments are possible. Sequence of the different steps can be changed. At the end of the
process, the product can go to a resting tube for further crystallization. During the experiments the
sequence of the steps was kept constant (see Figure 4.12): four scraped surface heat exchangers
(SSHE) ensured different cooling steps (CS), followed by a pin worker for post treatment (PT) of the
crystallized sample. For the margarine samples a resting tube (RT) was included after the pin worker as
further hardening was necessary for packaging. For one test with Pst20 an intermediate treatment (IT)
was added after the second cooling step. For each run on this pilot crystallizer around 70kg of the
sample was used. For each fat blend and margarine different cooling regimes were applied in the
SSHE’s. Depending on these temperatures the fat will crystallize faster or slower at respectively lower
or higher cooling temperatures, resulting in end products with different properties
The pilot of FTE (PilotFTE) is smaller and consists out of one cooling step and a post-treatment (see
Figure 4.12). The amount of sample needed for this pilot was around 25kg.
The CTSU, on lab scale consists of one cooling cell and no further treatment. Around 1kg of sample was
necessary for this set-up.
Results and discussion | 42
PilotVDM PilotFTE CTSU
Figure 4.12: Scheme of the different steps in the PilotVDM, PilotFTE and CTSU
After every step in both pilots, samples were collected in tubs and the product temperature was
measured. These tubs were subsequently stored at 20°C for two weeks. The storage stability of these
samples was monitored by measuring the SFC and hardness after one day, one week and two weeks.
To be able to compare the results of the PilotVDM and the PilotFTE, the samples were grouped based
on their temperature immediately after the process. For each fat blend different temperatures were
selected from the different steps in the PilotVDM and the PilotFTE. Additionally, these fat blends were
crystallized in the CTSU at these temperatures. For example, for the production of a margarine: the
temperature measured immediately after filling was 15,6°C for CS2, and 15,4°C for CS4. These samples
were than compared with a product made at 15°C in the CTSU. The different temperatures chosen for
the different steps in both pilots are shown in Table 4.5.The temperatures of the different parts of the
pilots and the different tests are shown in appendix 1.
Table 4.5: The different temperatures used on the CTSU to compare with the products on both pilot scales
In addition to the monitoring of the SFC and hardness during storage, the microstructure was also
studied by polarized light microscopy (PLM) and, for the margarine samples the water droplet size
distribution was determined. To compare the cake margarines on a consumer level, cakes were
produced with the margarines of the PilotVDM and the CTSU. Structure as well as hardness of the
cakes was examined.
To be able to compare the results from these two pilots with the CTSU, the shear forces of the
different processes were taken into account as different shear forces lead to different product
properties. To have an indication of these shear forces, the tip speed was calculated for each process,
as seen in Equation 2. Both pilots have a higher tip speed then the CTSU. A higher tip speed is
indicative for higher shear forces.
Equation 2: General formula of the tip speed(a), calculation of the tip speed of the pilots(b) and of the CTSU(c)
(
) ( )
( )
( )
(
)
( )
(
)
( )
4.3.2 Microstructural characterization of the fat blends
The microstructure of the different samples made on the CTSU, PilotVDM and PilotFTE was compared
using PLM.
In Figure 4.13(a) the first (CS1) and last step (PT) of the PilotVDM were compared for the sample
Shst20. The left part of the image shows the microstructure of the post-treatment that is characterized
by a regular crystal network with finer crystals than after the first cooling step. The sample after the PT
experienced more shear forces as it went through the five steps of the pilot process and therefore
crystals were finer.
The comparison between a shear time of 30 minutes and 2h for Shst20 at 20°C in the CTSU is shown in
Figure 4.13(b). Both samples showed a uniform network of fine crystals that only differs in the amount
of air bubbles that are present. These air bubbles are indicated by the arrows. They show a
characteristic black edge around the air bubbles. After thirty minutes of shearing some air bubbles
were observed but after 2h of shearing the amount of bubbles significantly increased making it difficult
to distinguish the crystal structure. So it can be concluded that more air was trapped with increased
shear time. This is possible because the CTSU has a large headspace filled with air. Trapping of the air
Results and discussion | 44
in the sample was possibly also influenced by the shape of the stirrer or by turbulence during stirring.
These air bubbles were not observed on pilot scale.
PILOT PT – CS1 CTSU 2h – 30’
(a) (b)
Figure 4.13: Comparison of the microstructure (PLM) of samples made with the CTSU, PilotFTE and PilotVDM. (a) Shst20 PilotVDM PT (left) – CS1 (right), (b) Shst20 CTSU 2h (left) – 30’ (right). The arrows indicate some air bubbles.
Figure 4.14 illustrates the differences in microstructure of margarine made in the CTSU and the two
pilots. The left side of the images always presents the sample of the CTSU. The comparison was made
between margarine made at 25°C and sheared for 1h30, and margarine after CS1 of the PilotVDM in
Figure 4.14(a). Figure 4.14(b) visualizes the comparison between margarine made at 15°C for a shear
time of 2h and margarine after PT of the PilotFTE. The samples were prepared by taking a small
amount of the two different margarines, putting them next to each other on a slide, adding a cover slip
and applying an equal pressure on both margarines with another slide. It can be seen that the right
side of the images were darker than the left side, as the thickness of the margarine layer originating for
the pilot was considerably higher. This was due to the higher hardness and plasticity of the pilot-
margarines and so it was not possible to obtain a thin layer making the comparison with the CTSU
more difficult. However, it was clear from the images that the margarines of the CTSU and the
margarines of the two pilots had fine crystals with an uniform distribution.
Results and discussion | 45
MARGARINE: CTSU – PILOT VDM MARGARINE: CTSU – PILOT FTE
(a) (b)
Figure 4.14: Comparison of the microstructure (PLM) of samples made with the CTSU, PilotFTE and PilotVDM. (a) Marg25 CTSU (left) – PilotVDM CS1 (right), (b) Marg15 CTSU (left) – PilotFTE PT (right). The arrows indicate some air bubbles.
Figure 4.15 shows the comparison between a fat blend made with the CTSU and a fat blend made with
respectively the PilotVDM (Margarine fat) and the PilotFTE (Pst20). The samples were prepared similar
to the margarine samples. For both images (a and b), the air bubbles between the fat crystals are
clearly visible for the samples prepared in the CTSU as indicated with arrows. Margarine fat crystallized
at 10°C for a shear time of 2h and Margarine fat after CS2 of the PilotVDM was compared in Figure
4.15(a). The microstructure was more regular for the sample made by the CTSU, but the crystals were
bigger. The high shear forces applied on the sample of the PilotVDM created more small crystals.
Figure 4.15(b) visualizes the comparison between Pst20 made at 10°C and sheared for 30min and Pst20
after CS of the PilotFTE. It can be seen that the sample produced on the CTSU contained two different
entities, namely a regular network of fine crystals in which darker grains (encircled in Figure 4.15(b))
are embedded. Although the sample looked as a smooth sample, but microstructural, grains were
present. As these grains were mostly harder than the matrix, it was not possible to get a thin layer. The
layer of the grains will thus be thicker and darker because it was more difficult for the light to pass
through. Both the samples of the CTSU and the PilotFTE had fine crystals. There was a uniform
distribution of the crystal network in the sample of the PilotFTE and no grains were observed.
Results and discussion | 46
FAT BLEND: CTSU – PILOT VDM FAT BLEND: CTSU – PILOT FTE
(a) (b) Figure 4.15: Comparison of the microstructure (PLM) of samples made with the CTSU, PilotFTE and PilotVDM. (a) MargFat10 CTSU (left) – PilotVDM CS2 (right), (b) Pst10 CTSU (left)– PilotFTE CS (right). The arrows indicate some air bubbles and the circles indicate some grains.
4.3.3 Evaluation of the fat blends and margarine samples as function of
storage time
This part presents the comparison of quality parameters between the samples made at both pilots and
the CTSU as function of storage at 20°C. After 1 day, 1 week and 2 weeks, both the SFC and hardness
were measured and the results were compared
4.3.3.1 CTSU versus PilotFTE for the different fat blends
This part presents the results of the comparison between the CTSU and the PilotFTE. For the three fat
blends, Pst20, Shst20 and MargFat, the trends were almost similar so the results of Pst20 are discussed
as a representative example. Figure 4.16 shows the comparison between the CTSU and the PilotFTE for
the palm stearin blend (Pst20). Only the two extreme values of the shear times (30’ and 2h) in the
CTSU are shown, the values of the intermediate shear followed the results of these extreme values.
The results of the other fat blends are shown in appendix 2.
Observations for SFC
Palm stearin blend
In the cooling step (Figure 4.16(a)), the SFC of the palm stearin blend didn’t show significant
differences between the different set-ups. For both the CTSU at 10°C and the CS, the SFC remained
constant during storage which was due to the high crystallization rate of palm stearin (see 3.3). The
SFC of the PT in the palm stearin blend (Figure 4.16(b)) remained constant during two weeks of storage
due to high crystallization rate of palm stearin and the high shear forces. The SFC of the CTSU samples
Results and discussion | 47
produced at 15°C increased during storage for one week. Only the samples with short shear times
continued to increase till the SFC was significantly higher than the SFC of the PilotFTE. The SFC of the
samples produced at longer shear times remained constant during the second week and was similar to
the SFC of the PT.
Margarine fat
The SFC showed the same trends in the CTSU and the PilotFTE for both steps. The SFC of the PilotFTE
remained constant during two weeks of storage. The SFC of the CTSU samples increased the first week
and remained constant in the second week. The crystallization of the margarine fat samples was not
yet completed after the process although the blend contained also palm stearin, but the amount of
palm stearin was much lower than in the blend with palm stearin (20% instead of 50%). The SFC was
similar for both set-ups at the CT but was significantly higher for the CTSU samples at the second step.
Shea stearin blend
During the first step, the SFC of the blend with shea stearin increased during the two weeks for both
the PilotFTE and the samples of the CTSU but the SFC of the PilotFTE was significantly higher than the
SFC of the CTSU. The crystallization was not yet completed after two weeks of storage due to the slow
crystallization rate of shea stearin (see 3.3). During the second step, the SFC increased for the PilotFTE
but decreased for the CTSU samples
Observation for hardness
Palm stearin blend
From Figure 4.16(a) of the first step in the process, it can be seen that the hardness of the PilotFTE is
much higher than the hardness of the samples produced in the CTSU, although the difference in SFC
was small. The lower hardness was probably due to the high amount of air bubbles in the samples of
the CTSU (see 4.3.2 Microscopy). It was not possible to measure the hardness of the CTSU samples
after one day, the hardness was too low due to the air bubbles in the samples.
After one day, the hardness of the CTSU samples was also very small in the second step of the process
of PilotVDM as seen in Figure 4.16(b). Both the hardness of the CTSU and the PT of the PilotFTE of the
palm stearin samples increased significantly during the first week of storage. After two weeks the
hardness of the PilotFTE significantly decreased but the hardness of the CTSU samples slightly
increased and the hardness became similar to the samples at low shear times in the CTSU. The
decrease in hardness after two weeks of the PilotFTE samples can be caused by the formation of grains
or by Ostwald ripening. Ostwald ripening is the growth of larger crystals at the expense of smaller
Results and discussion | 48
crystals and is related to the solubility gradient found between small and large crystals (Rousseau,
2000; Oijo et al, 2004).
For every blend it was concluded that with an increased shear time, the hardness decreased as at
higher shear times, more air bubbles were trapped in the sample. The difference in hardness between
the samples of the CTSU and the PT of the PilotFTE were also smaller than at the CS. As the
crystallization temperature of the samples in Figure 4.16(b) was higher than in Figure 4.16(a), the
crystallization goes slower and less crystals were formed during the process and thus more sintering
occurs in the samples of the CTSU during storage (see Table 4.5 and Figure 4.16).
Margarine fat
The hardness in the cooling step showed the same trend as the palm stearin blend. The hardness of
the blends with margarine fat remained constant in the PilotFTE but the hardness of the CTSU samples
increased during the first week of storage and remained than constant. The hardness of the CTSU
samples was significantly higher than the sample of the PT from one week of storage.
Shea stearin blend
For both steps of the process, the same trends in hardness can be seen as for the margarine fat. Only
the hardness remained significantly lower during the storage.
10°C 15°C
(a) (b)
Figure 4.16: Hardness (N) and SFC (%) in function of the storage time for the comparison of (a) CTSU Pst10 – PilotFTE CS and (b) CTSU Pst15 – PilotFTE PT
Results and discussion | 49
4.3.3.2 CTSU versus PilotVDM for the different fat blends
The trends for hardness and SFC of the CTSU versus PilotVDM were similar as the ones between CTSU
and PilotFTE. In the values of the intermediate shear followed the results of these extreme values. , the
results of the palm stearin blend were compared with the CTSU and the PilotVDM. The results of the
other two fat blends are shown in appendix 3. Only the two extreme values of the shear times (30’ and
2h) in the CTSU are shown, the values of the intermediate shear followed the results of these extreme
values.
Observations for SFC
Palm stearin blend
For both the palm stearin blend and the margarine fat, the SFC showed different trends when the
crystallization temperatures of the CTSU samples was lower than 15°C (see Table 4.6) than samples
produced at temperatures of 15°C or higher. For the palm stearin blend, it can be seen that the SFC
remained constant during the storage for the samples produced at temperatures lower than 15°C. The
crystallization was already completed during the process. At crystallization temperatures of 15°C or
higher, the SFC of the samples of the PilotVDM remained constant during storage. The SFC of the CTSU
on the other hand still increased after 1 week storage. The same was observed for the samples with a
shorter shear time, crystallization continued until the second week. This increase was due to an
incomplete crystallization during the shearing process as it was crystallized at a higher temperature.
The SFC of the PilotVDM was similar or slightly higher than that of the CTSU for all crystallization
temperatures.
Margarine fat
For the margarine fat at all crystallization temperatures, there was a small increase of the SFC for the
samples made in the CTSU. At crystallization temperatures lower than 15°C, the SFC of the PilotVDM
was higher than the samples of the CTSU, but reached the same point after two weeks of storage. At
temperatures higher than 15°C, the SFC of the PilotVDM was lower for the whole storage time but the
difference between the SFC of the two set-ups became smaller at the end of storage.
Shea stearin
The SFC of the shea stearin blend increased during storage and the crystallization rate of the samples
of the CTSU and PilotVDM was similar. The SFC of both CTSU and PilotVDM was similar after two weeks
of storage.
For all the three blends, the final SFC after two weeks of storage was between 30 and 35%, as
established for the blending of the samples (see 3.1.1 Material and Methods).
Results and discussion | 50
Observations for hardness
In Table 4.6 the results are shown of the comparison between the CTSU and the PilotVDM for the palm
stearin blend. Also the hardness was influenced by the crystallization temperature. At temperatures
lower than 15°C, the hardness of the samples made at the CTSU was around 1N and much lower than
the hardness of the samples of the PilotVDM after two weeks of storage (CS2: 13,9N, CS3: 9,1N and
CS4: 9,8N). When there were crystallized at temperatures higher or equal to 15°C, the hardness was
around 5N and the difference with the hardness of the samples of the PilotVDM (CS1: 12,9N, PT: 9,7N)
was smaller. At the CTSU, the samples with a short shear time showed a higher hardness than the
samples of long shear times. This can again be explained by the amount of air bubbles in the sample as
this increased when the shear time was longer (see 4.3.2 Microscopy). It can also be seen that for
every blend, both for the CTSU and the PilotVDM samples the hardness remained almost constant or
showed a limited increase during storage.
Table 4.6: The comparison between the CTSU and the PilotVDM for the blend with palm stearin
CTSU 2h 0,1±0,0(a,A) 3,5±0,3(b,B) 2,4±0,6(b,C) 27,1±1,6(a,A) 30,4±1,3(a, B) 28,1±1,7(b,B)
PILOT PT 9,8±1,2(b,A) 9,0±1,0(c,A) 9,7±1,0(c,A) 31,8±0,8(b,A) 33,6±0,9(a, B) 32,3±0,0(c,A,B)
*hardness was not detectable In a column, the results with the same lower-case letter (a,b,c) are not significantly different (p < 0,05) In a row, the results with the same capital letter (A,B,C) are not significantly different (p < 0,05)
Results and discussion | 51
4.3.3.3 The evaluation of the different steps in the pilot processes during storage
This part presents the evaluation of the different steps in the PilotFTE and PilotVDM for a selected fat
blend. At a particular step in the pilot process, both the hardness and SFC will be influenced by the
process parameters but also by the parameters of the previous steps. This effect was not present for
the samples of the CTSU because this is a batch process.
Figure 4.17 presents the two steps of the PilotFTE together with the corresponding sample of the CTSU
for the palm stearin blend. The SFC was not significant different and it remained constant during the
process, the hardness however increased. More network formation and more sintering occurred
during the last step. The temperature at the last step was also higher than the first step. As mentioned
before, the sintering during storage was more distinct at higher crystallization temperatures.
Figure 4.17: The evaluation of the different steps in the PilotFTE for the palm stearin blend after 1 week
Figure 4.18 visualizes the process of the PilotVDM with the corresponding samples of CTSU for
margarine fat. After the second cooling step, the SFC increased to remain stable afterwards. The SFC of
the CTSU stayed constant during the whole process but was lower than the SFC of CS2, CS3 and CS4 of
PilotVDM. The hardness of the four cooling steps of the PilotVDM was similar but at the last step, the
PT, the hardness decreased. The PT kneaded the product and the fat crystal network was partially
broken down resulting in a lower hardness. The hardness of the CTSU did not follow the same trend:
after CS1 and PT the hardness was higher than at the other steps due to the higher crystallization
temperature leading to more sintering during storage.
Results and discussion | 52
Figure 4.18: The evaluation of the different steps in the PilotVDM for margarine fat after 1 week
The hardness of the CTSU did not show the same trend as the hardness of the PilotVDM for the
different steps. The production of a sample of the CTSU had only one step and was not influenced by
another step. The SFC was similar between the two methods; there was no big influence of the
previous steps on the SFC of the samples of the pilot processes.
From these results it can be concluded that there are still a lot of differences between the fat blends
produced on the CTSU and on both pilots. Although some similarities were found in SFC, the hardness
was much smaller at the CTSU samples than in both pilots. These differences are due to the capturing
of air in the samples reducing the hardness.
4.3.3.4 CTSU versus PilotFTE for margarine
The comparison between the PilotFTE and the CTSU for margarine is shown in Figure 4.19. The
differences between the hardness of the CTSU and the PilotFTE became smaller when the storage time
increased but they were still significant. There was no significant difference between the SFC of the
PilotFTE and the CTSU, they remained both constant.
Results and discussion | 53
15°C
Figure 4.19: Hardness (N) and SFC (%) in function of the storage time for the comparison of CTSU Marg15 – PilotFTE CS, PT
4.3.3.5 CTSU versus PilotVDM for margarine
In the last step a cake margarine was produced by the CTSU. The comparison was made between the
CTSU and both the PilotFTE and PilotVDM.
Observations for SFC
The first cooling step of the pilot was performed at 25°c and simulated in the CTSU (Figure 4.20(a)) The
SFC of the CTSU samples did not change the first week but increased in the second week. Figure 4.20(a)
shows the comparison between the CS1 of the PilotVDM and the CTSU samples made at 25°C. Both
set-ups showed the same trend, there was a large increase in SFC during the first week and in the
second week, the SFC remained constant. After two weeks, the SFC of the CS1 was significantly lower
than the SFC of the CTSU samples. The higher the crystallization temperature, the bigger the increase
in SFC will be, because of the formation of less crystals during the process. The CS2, CS3 and CS4 of the
PilotVDM for margarine are compared with the samples of the CTSU crystallized at 15°C, as shown in
Figure 4.20(b). The SFC of CS2 is lower than the other samples but increased during the first week to
reach a similar value as the samples of the CTSU and the PilotVDM (CS3 and CS4). Both the SFC of the
samples of the CTSU and CS3 and C4 remained constant during the storage at a similar SFC. After 2
weeks of storage, the samples of the CTSU and the PilotVDM reached the same level. In Finally, the SFC
of the RT was compared with the SFC of margarine crystallized in the CTSU at 20°C. The SFC was similar
Results and discussion | 54
for the PilotVDM and CTSU samples after 1 day and 2 weeks and overall, the SFC increased during the
storage for both set-ups but with a different progress. This can be seen in Figure 4.20(c). The SFC of the
RT increased in the first week and remained constant in the second week.
15°C 20°C
(a) (b)
25°C
(c)
Figure 4.20: Hardness (N) and SFC (%) in function of the storage time for the comparison of (a) CTSU Marg15 – PilotVDM CS2, CS3, CS4, (b) CTSU Marg20 – PilotVDM RT, (c) CTSU Marg25 - PilotVDM CS1
Results and discussion | 55
Observations for hardness
Both the hardness of the samples of the CTSU crystallized at 25°C and the CS1 increased during the
storage time, as seen in Figure 4.20(a). There were no similarities between the two set-ups but the
differences were again small, especially at the shorter shear times. The CS2, CS3 and CS4 of the
PilotVDM were compared with samples of the CTSU in Figure 4.20(b). The hardness of the samples in
the PilotVDM was higher than the hardness of the CTSU samples during the first week but after two
weeks, the hardness of the samples of the CTSU at low shear times had a similar hardness of the
samples of CS3 and CS4, the hardness of CS2 remained the highest. During the storage time, the
hardness of the PilotVDM remained constant, but the samples of the CTSU increased, due to sintering,
so the differences in hardness between the PilotVDM and the CTSU became smaller during storage.
The hardness was again higher for low shear times due to less capturing of air at short shear times than
for long shear times. In Figure 4.20(c) can be seen that both the hardness of the CTSU samples at 20°C
and the sample of the RT increased during storage. After one week of storage, the hardness of the
samples with low shear times in the CTSU were similar to the sample of the RT, the hardness of the
other CTSU samples was significantly lower than the samples of the RT. However the differences in
hardness were not as large as in the fat blends. At the samples crystallized at 20°C and 25°C, both the
hardness and the SFC increased during the storage. The crystallization was not completed during the
shearing process due to the high crystallization temperatures.
4.3.3.6 The evaluation of the different steps in the pilot processes during storage of
margarine
The SFC was similar for the two process steps (CS and PT) and also for the samples of the CTSU and the
PilotFTE. However Figure 4.21(a) shows that there was a big difference in hardness between the two
set-ups; the hardness of the CTSU was much lower than the hardness of the PilotFTE. For the samples
of the PilotFTE, the hardness was smaller for the PT. There was some breakdown of the crystal network
due to the working of the PT.
The different steps of the PilotVDM and the samples of the CTSU were compared. The results are
shown in Figure 4.21. The SFC was equal for both set-ups in the first two steps. Thereafter, the SFC of
samples produced on the pilot the PilotVDM increased at CS3 and remained unchanged for the
following steps. The SFC of the CTSU remained constant during the four cooling steps but decreased
during resting (RT). This decrease was probably due to the higher crystallization temperature in the PT.
The hardness of the different steps in the PilotVDM remained constant; there was no breakdown of
the network in the last step as at the PilotFTE. The hardness of the samples of the CTSU was
significantly lower than the hardness of the samples of the PilotVDM.
Results and discussion | 56
(a) (b)
Figure 4.21: The evaluation of the different steps in (a) PilotFTE and in (b) PilotVDM for margarine after 1 week
The results of the margarine samples made at the CTSU had more similarities with the results of the
pilots than the fat blends. The SFC of the CTSU showed a lot of similarities but the hardness was still
small in comparison with the pilot samples. The results of the microscopy showed also air bubbles in
the margarine made with the CTSU. When the crystallization temperatures were above 15°C, the
hardness of the samples produced in the CTSU at short shear times was similar to the margarine
produced on the pilots.
4.3.4 Water droplet size distribution
The water droplet size distribution of margarine is an important quality parameter. Therefore the
water droplet size distribution of the two pilot set-ups and the CTSU were compared. The samples that
are crystallized at 15°C or 20°C with a shear time of 30min or 1h in the CTSU were chosen due to their
similarities with the PilotVDM and PilotFTE. For the PilotVDM and PilotFTE the samples of respectively
the CT and PT, and the CS4 and RT are compared with the CTSU. The results of the D3,3 or average
diameter, the standard deviation and the 97,5% value of the samples are shown in Table 4.7. The
water droplet size distribution of the samples of the PilotVDM was too small to measure as in a cake
margarine the diameter is around 0,2µm (van Duynhoven et al, 2002). Margarine can be considered as
a microbiological safe product as the water droplet size distribution is smaller than a micro-organism
(0,5 – 5µm). The small water droplets were the result of the high shear forces on the emulsion, the
emulsifier will stabilize the small droplets in the fat phase. At the PilotFTE, the geometrical weighted
mean diameter was around 3,5µm, 10 times larger than at the PilotVDM, which was due to the lower
shear forces. Additionally only 2 steps were present in this pilot resulting in bigger water droplets. The
Results and discussion | 57
droplet size was for 97,5% of the water droplets around 8,9µm and 11,6µm for respectively the first
and second step of the PilotFTE. This means that micro-organism can grow and multiply in the product.
The samples made with the CTSU had a water droplet size distribution that was more than 100 times
bigger than the droplet size at the PilotVDM, they had also a very large distribution, some droplets
were even bigger than 1mm. The large size and broad distribution is due to the rather low shear forces
in the CTSU. The water droplet size and the distribution of the samples crystallized at 15°C were
smaller at longer shear times. When the crystallization was done at 20°C there was no difference
between 30min and 1h of shear. The droplet size distribution of the samples crystallized at different
temperatures is similar.
Table 4.7: The D3,3, standard deviation and 97,5% values (µm) of different margarine samples
D3,3 (µm) stdev (µm) 97,5% (µm)
CTSU 15°C 30min 77,8 5,0 1840,0
CTSU 15°C 1h 31,4 3,6 392,0
CTSU 20°C 30min 59,4 5,5 1670,0
CTSU 20°C 1h 63,0 5,1 1559,0
PilotFTE CS 3,5 1,6 8,9
PilotFTE PT 3,4 1,9 11,6
PilotVDM test 1 CS4 * * *
PilotVDM test 1 RT * * *
*not measurable
4.3.5 Cake tests
In this part, sponge cakes were prepared with the different cake margarines to assess the effect of the
crystallization procedure on the final cake structure. Six cakes were prepared: two with the final
margarines from trail 1 (PilotVDM_Marg1) and 2 (PilotVDM_Marg2) and four with margarines from the
CTSU. These four margarines were selected based on their similarities with the margarines made on
the PilotVDM (see 4.3.3.4 and 4.3.3.5): two crystallized at 15°C with 30 minutes (CTSU15°C_30’) and
one hour (CTSU15°C_1h) of shearing, and two crystallized at 20°C with 30 minutes (CTSU20°C_30’) and
one hour (CTSU20°C_1h) of shearing. The six cakes were visually evaluated by comparing the crumb
structure and the hardness was measured by texture analysis.
The photographs of a slice taken from the six cakes are shown in Figure 4.22. These photos give an idea
of the difference in texture and the loaf size. It can be that for cakes prepared with the margarines
Results and discussion | 58
from the PilotVDM (Figure 4.22(a) and (d)) the air bubbles were bigger and less homogenously
distributed in the cakes, resulting in a coarser texture. No visual differences were observed between
the cakes prepared with margarine of the CTSU. These cakes were characterized by a finer texture with
small air bubbles. Some bigger air bubbles are still present but in much lesser extent than in the cakes
made of the margarine of the pilot.
(a) (b) (c)
(d) (e) (f)
Figure 4.22: Photographs of the loaf size of sponge cake with (a) PilotMarg1, (b) CTSU15°C_30’, (c)CTSU15°C_1h, (d)PilotMarg2, (e)CTSU20°C_30’ and (f)CTSU20°C_1h
In general, the quality characteristic of the cakes didn’t show many differences. Their volume and
crumb were similar; although the density of the dough was smaller for the margarines made with the
CTSU (see Figure 4.23(a)). This can be explained by the air bubbles that were captured in the margarine
of the CTSU (see before) giving a lower density to the dough.
In addition to the visual comparison, the hardness of the loaf size was measured with a penetration
test of which the results are shown in Figure 4.23(a). The hardness of the cakes of PilotVDM_Marg1
was significantly different from that of the cakes with margarines of the CTSU, while the hardness of
the cakes with PilotVDM_Marg2 were only different with the cakes made by margarine of the CTSU
crystallized at 15°C. The hardness of the cake with Marg15_1h was significantly lower than the cakes
Results and discussion | 59
made with Marg20_30’ and Marg20_1h. This trend was also seen for the hardness of the margarines.
The margarines crystallized at 15°C had a lower hardness than these crystallized at 20°C. Furthermore,
the hardness of the cakes was plotted against the density of the cake (see Figure 4.23(b)) and a linear
correlation (R² = 0,97) could be found. This means that a lower density of the cake will give a lower
hardness of the cake. Both parameters were influenced by the capture of air in the margarine made
with the CTSU. There will be more air bubbles in the dough because there was already air in the
margarine. This will cause, after the baking, a more aerated cake characterized by a lower hardness.
(a) (b)
In the bars: the results with the same lower-case letter (a,b,c) are not significantly different (p < 0,05)
Figure 4.23: (a) Hardness of the cakes made with different margarines and (b) the linear correlation between hardness and the dough density
a,b c
b,d b,c,d cC
a
Conclusions | 60
5 Conclusions
Since the last years, the health issue about saturated and trans unsaturated fatty acids has encouraged
the food industry to make new margarines with less saturated and trans fatty acids. These new
margarines have to be tested before the industrial production can start. This is done on pilot scale but
these tests still need a lot of (fat) ingredients and time.
In this study a method was developed to produce margarine products on lab scale as an alternative to
pilot scale. In the first part a fundamental study was executed on crystallization under shear by
measuring the SFC and rheological parameters. It was clear that shear enhanced the crystallization and
this resulted in a faster increase of the SFC and a higher final SFC. Especially the slow crystallizing shea
stearin blend (at 15°C) showed a much steeper SFC profile when shear was applied. Next to monitoring
the SFC, also oscillatory rheology was used to study the effect of shear. After cooling the sample, a
shear step with different shear rates and with different shear times was applied. This step was
followed by an isothermal period at static conditions. Both the shear rate and time had a large
influence on the apparent viscosity in the shear step. All fat blends showed an earlier increase in
apparent viscosity at higher shear rates due to a faster crystallization. However, the higher the shear
rate, the smaller the increase in apparent viscosity, resulting in a lower equilibrium value for the
samples at higher shear rates. It could be seen that the apparent viscosity was similar for the shear
rates of 300s-1 and 500s-1. At these shear rates no aggregation is possible because the contact time
between the colliding entities was too short to aggregate. The crystallization proceeded in two steps
but this was less visible for the samples crystallized at 20°C, they tended to go to one step. At the end
of the shear step, the final apparent viscosity was the highest for the lowest shear rates as there was
less structural breakdown. The |G*| of the shea stearin blend in the period without shear went fast to
an equilibrium for all shear rates. At the end of the isothermal period, the curves reached a similar
value. For the palm stearin blend, the crystallization proceeds different at the period without shear.
The increase is much slower and there is a clear difference between the two lowest (75s-1 and 150s-1)
and the two highest shear rates (300s-1 and 500s-1).
In the second part of the research a method was developed and optimized on lab scale as alternative
for the pilot scale. The first technique, the RVA was not suitable as it was not possible to adapt the
procedure due to the limitations of the device. Also a post treatment with the TNO cell lead to the
development of grains (β crystals) in the shea stearin blends during the storage. A second technique,
that was tested, was a controlled temperature shearing unit (CTSU) developed for the production of fat
blends and margarine on lab scale. The procedure of the CTSU was optimized by using higher shear
rates and longer shear times. The results with the shea stearin blend showed no grains during storage
Conclusions | 61
of two weeks. As the CTSU technique seemed suitable for crystallization under shear on a small scale,
the next step was to compare its performance with the PilotVDM and PilotFTE.
Next to the palm stearin and shea stearin blend, also a cake margarine fat and cake margarine were
compared between the pilots and the CTSU. From the images of the PLM, it was seen that air was
captured in the samples of the CTSU. This phenomenon had a large influence on the hardness of the
samples, especially for the fat blends. The hardness of the CTSU samples was much smaller compared
to the blends processed in both pilots. The biggest differences were observed for the samples
crystallized at temperatures below 15°C. The SFC was also significantly lower in the CTSU but the
differences between the CTSU and the pilots were not that big. For the margarine samples, there were
still some differences in hardness but much smaller than for the fat blends. The SFC during the whole
process was similar between the different pilots and the CTSU. The most similarities, in both hardness
and SFC, were seen at lower shear times (30min). The water droplet size distribution of the margarine
samples for the different set ups were examined. The smallest water droplet size distribution was
found for the samples made by the PilotVDM, followed by the samples of the Pilot FTE and the biggest
water droplet size distribution was for the CTSU samples. Higher shear forces will result in a smaller
water droplet size distribution. At the end, the margarines were compared on consumer level by
preparing cakes with the margarines of the CTSU and the PilotVDM. The hardness of the cakes was
similar or slightly lower for the cakes with the margarine of the CTSU. The loaf size of the different
cakes was visually evaluated. The loaf size of the cakes made with margarine of the PilotVDM showed
larger air bubbles and a more heterogeneous structure. However the loaf size of the cakes with
margarine of the CTSU showed a homogeneous structure with small air bubbles. There were no
differences seen between the cakes of the margarines made at different temperatures and shear times
at the CTSU.
The CTSU was thus not suitable to crystallize fat blends that are similar with the pilot scale. However
the margarine products on the CTSU showed a lot of similarities with the margarine on the pilots and
even on consumer level, almost no differences were observed. So the CTSU seemed to be a valuable
tool to produce margarine on lab scale as alternative for pilot scale.
Further research | 62
6 Further research
In this research it is concluded that there were many similarities between the margarine products in
the CTSU and the pilots, but this was not the case for the comparison with the fat blends. Optimization
is needed to improve the crystallization of both the fat blends and the margarine on lab scale. As seen
in the research, the biggest problem was the capturing of air in the samples. This can be avoided by
reducing the headspace in the CTSU or by using another stirrer type to trap less air in the sample.
Another possibility to improve the method is to use a continuous process with several steps, instead of
a batch system. During this process, different temperature profiles are imposed on one sample. The
TNO cell as post treatment did not give the desired properties, but other post treatments can be tested
which are similar to a pin worker as used on pilot scale.
In this study, only a cake margarine was tested. In further research, it can be tested whether this setup
is suitable for other sorts of margarine.
References| 63
7 References
Bockisch, M. (1998). Fats and Oils Handbook. Champaign, AOCS Press, 838p.
Braipson-Danthine, S. & Deroanne, C. (2004). Influence of SFC, microstructure and polymorphism on
texture (hardness) of binary blends of fats involved in the preparation of industrial shortenings. Food
Research International, 37, 941-948.
Campos, R., Narine, S. S & Marangoni, A. G. (2002). Effect of cooling rate on the structure and
mechanical properties of milk fat and lard. Food Research International, 35, 971-981.
Carter, C., W. Finley, J. Fry, D. Jackson, and L. Willis (2007). Palm oil markets and future supply.
European Journal of Lipid Science and Technology, 109, 300 -314.
Chen, C. W., O. M. Lai, H. M. Ghazali, and C. L. Chong (2002). Isothermal crystallization kinetics of
refined palm oil. Journal of the American Oil Chemists' Society, 79, 403-410.
De Graef, V., Goderis, B., Van Puyvelde, P., Foubert, I. & Dewettinck, K. (2008). Development of a
rheological method to characterize palm oil crystallizing under shear. European Journal of Lipid Science
and Technology, 110, 521-529.
De Graef, V. (2009). Microstructural properties of isothermal palm oil crystallization. PhD Thesis. Ghent
University, Belgium, 181p.
De Graef, V., Van Puyvelde, P., Goderis, B. & Dewettinck, K. (2009). Influence of shear flow on
polymorphic behavior and Microstructural development during palm oil crystallization. European
Journal of Lipid Science and Technology, 111, 290–302.
deMan, J. (1998). Functionality of palm oil in food products. Journal of food lipids, 5, 159-170.
Dewettinck, K. & Fredrick, E. (2010-2011). Course: Technology of vegetable products. Ghent University,
Belgium.
Freeman, I. P. (2000). Margarines and shortenings. Ullman’s Encylopedia of Industrial Chemistry.
Ghotra, B. S., Dyal, S. D. & Narine, S. S. (2002). Lipid shortenings: a review. Food Research International,
35, 1015-1048.
Hui, Y. H. (1996). Edible oil and fat products: products and application technology. Bailey’s industrial oil
and fat products, Volume 3.
References| 64
Hui, Y. H. & Clark, S. (2007). Handbook of food products manufacturing: principles, bakery,. beverages,
cereals, cheese, confectionary, fats, fruits, and functional foods. John Wiley & Sons.
Johansson, D. & Bergenstahl, B. (1995). Sintering of fat crystal networks in oil during post-
crystallization processes. Journal of the American Oil Chemists’ Society, 72, 911-920.
Kellens, M., V. Gibon, M. Hendrix, and W. De Greyt (2007). Palm oil fractionation. European Journal of
Lipid Science and Technology, 109, 336-349.
Kaphueakngam, P., Flood, A. & Sonwai, S. (2009). Production of cocoa butter equivalent from mango
seed almond fat and palm oil mid fraction. Asian Journal of Food and Agro-Industry, 2,441-447.
Liu, Y., Meng, Z., Zhang, F., Shan, L. & Wang X. (2010) Influence of lipid composition, crystallization
behaviour and microstructure on hardness of palm-based margarines. European Food Research
Technology, 230, 759-767.
Miskandar, M. S., Che Man, Y. B., Yusoff, M. S. A. & Rahmann, R. A. (2002a). Effect of emulsion
temperature on physical properties of palm oil-based margarine. Journal of the American Oil Chemists’
Society, 79, 1163-1168.
Miskandar, M. S., Che Man, Y. B., Yusoff, M. S. A. & Rahmann, R. A. (2002b). Effect of scraped-surface
tube cooler temperatures on the physical properties of palm oil margarine. Journal of the American Oil
Chemists' Society, 79, 931-936.
Miskandar, M. S., Che Man, Y. B., Yusoff, M. S. A. & Rahmann, R. A. (2004). Effect of flow rates on the
storage properties of palm-oil based margarine. Journal of Food Lipids, 11, 1-13.
Miskandar, M. S., Che Man, Y. B., Yusoff, M. S. A. & Rahmann, R. A. (2005). Quality of margarine: fats
selection and processing parameters. Asia Pacific Journal of Clinical Nutrition, 14, 387-395.
Moustafa, A. (1992). Bakery fats. American soybean association, Madrid, Spain.
Mozair, C., deMan, J. M., & deMan, L. (1989). Effect of tempering on the physical properties of
shortenings. Canadian Institute of Food Science and Technology, 22, 238–242.
Narine, S. S. and K. L. Humphrey (2004). A comparison of lipid shortening functionality as a function of
molecular ensemble and shear: microstructure, polymorphism, solid fat content and texture. Food
Research International, 37, 28-38.
Njumbe Ediage, E. (2007). Crystallization and melting behavior of triacylglycerols and their mixtures.
Thesis. Ghent University, Belgium, 90p.
References| 65
Nor Aini, I. & Miskandar, M. S. (2007). Utilization of palm oil and palm products in shortenings and
margarines. European Journal of Lipid Science Technology, 109, 422-432.
Ojijo, N., Kesselman, E., Shuster, V., Eichler, S., Eger, S., Neeman, I. & Shimoni, E. (2004). Changes in
microstructural, thermal, and rheological properties of olive oil/monoglyceride networks during
storage. Food Research International, 37, 385-393.
Rousseau, D. (2000) Fat crystals and emulsion stability – review. Food Research International, 33, 3-14.
Smith, K. W., Cain, F. W. & Talbot, G. (2005). Crystallization of 1,3-dipalmitoylglycerol and
tripalmtoylglycerol and their mixtures from acetone. European Journal of Lipid Science and Technology,
107, 583-592.
Tarabukina, E., Jego, E., Haudin, J.-M., Naward, R. & Peuvrel-Disdier, E. (2009). Effect of shear on the
rheology and crystallization of palm oil. Journal of Food Science, 74, 405-416.
Timms, R. E. (2003). Confectionery Fats Handbook - Properties, production and application.
Bridgewater, The Oily Press, 441p.
Van Dalen, G. (2002). Determination of the water droplet size distribution of fat spreads using confocal
scanning laser microscopy. Journal of Microscopy, 208, 116-133.
Van Duynhoven, J. P. M., Goudappel, G.-J. W., Van Dalen, G., Van Bruggen, P. C., Blonk, J. C. G. &
Eijkelenboom, A. P. A. M. (2002). Scope of droplet size measurements in food emulsions by pulsed field
gradient NMR at low field. Magnetic Resonance in Chemistry, 40, 51-59.
Vereecken, J. (2010). Effect of acylglycerol composition on the Microstructural and functional
properties of bakery fats and margarines. PhD Thesis, Ghent University, Belgium, 251p.
Young, N. W. G. & Wassell, P. (2008). Food emulsifiers and their applications. Springer.
Appendix| 66
Appendix
Appendix 1 Processparameters of the PilotVDM and the PilotFTE
Temperatures of the samples taken at the different steps in the process for the PilotVDM
Temp. (°C) CS1 CS2 CS3 CS4 PT/RT
Shea stearin blend 11,7 9,3 16,1 12,9 15,6
Palm stearin blend 17,7 10,0 5,4 10,0 13,0
Margarine fat 15,6 10,6 6,6 11,8 15,5
Margarine 24,2 15,6 13,7 15,4 18,6
Temperatures of the samples taken at the different steps in the process for the PilotFTE
Temp. (°C) CS PT
Shea stearin blend 9,8 22,5
Palm stearin blend 10,8 18,9
Margarine fat 7,7 21,1
Margarine 11,6 16,2
Appendix| 67
Appendix 2 CTSU versus PilotFTE of the shea stearin blends and the margarine fat
The results of the comparison between the CTSU and the PilotFTE for the shea stearin blend.
The comparison is showed between Shst20 crystallized at 10°C and the CS of the PilotFTE (a) and
between Shst20 crystallized at 20°C and the PT of the PilotFTE.
(a) (b) The results of the comparison between the CTSU and the PilotFTE for the palm stearin blend
The comparison is showed between MargFat crystallized at 7°C and the CS of the PilotFTE (a) and
between MargFat crystallized at 20°C and the PT of the PilotFTE.
(a) (b)
Appendix| 68
Appendix 3 CTSU versus PilotVDM of the shea stearin blends and the margarine fat
The results of the comparison between the CTSU and the PilotVDM for the shea stearin
PILOT PT 6,6±0,6(c,A) 12,4±0,8(c,B) 7,0±1,0(b,A) 28,3±1,5(b,A) 30,0±1,5(a,A,B) 31,9±0,4 (a,B)
In a column, the results with the same lower-case letter (a,b,c) are not significantly different (p < 0,05) In a row, the results with the same capital letter (A,B,C) are not significantly different (p < 0,05)
Appendix| 69
The results of the comparison between the CTSU and the PilotVDM for margarine fat.
PILOT PT 5,6±0,4(a,A) 9,9±0,6(c,B) 8,8±0,7(a,B) 28,6±1,0(b,A) 36,5±1,3(a, B) 34,4±0,2(b,C)
In a column, the results with the same lower-case letter (a,b,c) are not significantly different (p < 0,05) In a row, the results with the same capital letter (A,B,C) are not significantly different (p < 0,05)