Advanced Powder Technology 24 (2013) 771779

Contents lists available at SciVerse ScienceDirect

Advanced Powder Technology

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

Original Research Paper

Antioxidant capacity of spray-dried plant extracts: Experiments

and simulations
T.A.G. Langrish , R. Premarajah
School of Chemical and Biomolecular Engineering, University of Sydney, NSW 2006, Australia

a r t i c l e

i n f o

Article history:
Received 7 November 2012
Received in revised form 26 March 2013
Accepted 27 March 2013
Available online 12 April 2013
Keywords:
ORAC
Bioactive
Simulation
Plug ow
Mathematical modelling

a b s t r a c t
The effects of different inlet air temperatures (70150 C) have been studied on the antioxidant retention
and yields of a spray-dried bioactive solution (Hibiscus sabdariffa L.) from a Buchi B-290 spray dryer and
compared with plug-ow spray drying simulations. Antioxidant retention has been tested using the Oxygen Reducing Antioxidant Capacity assay (ORAC). Experimentally, a peak yield of between 65% and 70% of
the solids fed to the dryer has been found at an outlet gas temperature of 6065 C and an inlet air temperature of 110 C, regardless of the batch of material or the liquid feed rate. The varying outlet gas temperatures did not signicantly affect the antioxidant retention of the sample, and the simulations
demonstrate that this result is due to the competing effects of increasing air temperature and decreasing
water activity (at higher inlet air temperatures) on the degradation kinetics. These results suggest that it
is more important to obtain greater product yields rather than minimising the degradation amount in this
spray-drying situation.
2013 The Society of Powder Technology Japan. Published by Elsevier B.V. and The Society of Powder
Technology Japan. All rights reserved.

1. Introduction
Spray dryers are frequently used in a variety of industries for
producing powder products [1]. Apart from reducing the moisture
content of the nal products, it has long been realised that other
powder properties are affected by the way in which the spray dryer
is operated, such as bulk density [2]. At a fundamental level, the
physical powder structure changes during spray drying, due to
changes in the degree of crystallinity [3], and reactions can also
take place, such as the degradation of vitamin C [4,5]. Modelling
approaches in spray drying based on plug ow and parallel ow
of gas and solids are t for the purpose of predicting trends in these
aspects (product crystallinity and reaction extent) with changes in
the operating conditions [6]. Plug-ow simulation approaches may
be useful [6] in conjunction with the more sophisticated but also
more demanding modelling approach [7] of Computational Fluid
Dynamics (CFD).
In terms of the applications for spray-dried powders, health,
food and nutrition are key outcomes. Heart disease and cancer
are amongst the leading causes of death in Australia and many
other industrialised countries [8], and oxidative stress is thought
to play a role in their development [8,9]. Oxidative stress is an
imbalance between reactive oxygen species (ROS) and antioxidant
defence and may lead to oxidative damage to body tissues [8,9].

Corresponding author.
E-mail address: timothy.langrish@sydney.edu.au (T.A.G. Langrish).

This can result from either an increase in ROS or a breakdown in

the level of antioxidant defence, and therefore the balance of antioxidants and ROS in the body is crucial. An increase in ROS within
the body can arise from common sources such as smoking, exposure to heat and UV light. The antioxidants used to protect against
ROS can come from internal sources, i.e. the antioxidants produced
within the human body, or external sources, such as antioxidants
obtained from fruits and vegetables. The use of traditional synthetic antioxidants, such as butylated hydroxyanisole (BHA) and
butylated hydroxytoluene (BHT), has been reduced in recent times
since these chemicals are suspected to be carcinogenic and to
cause liver damage [1013]. Therefore there is a signicant market
demand to replace these synthetic antioxidants with natural antioxidants derived from plant material.
In this study, a commercially available extract, namely rosella
extract (Hibiscus sabdariffa L.), which is high in antioxidants, has
been chosen as the sample material. The main antioxidants in H.
sabdariffa L. are anthocyanins, which are the largest group of natural pigments in plants and are responsible for many attractive colours in plants, such as owers, fruits (particularly berries) and
vegetables [14]. Drying the antioxidant-rich extract would facilitate increased shelf-life and reduced microbial degradation due
to the reduction of water content. Also, the reduction of weight
through the removal of water would allow the product to be transported at a greatly reduced cost, which may increase the likelihood
of the product being made available globally. Creating a powdered
product may also increase the ease of adding this antioxidant-rich
powder to consumer food products.

0921-8831/$ - see front matter 2013 The Society of Powder Technology Japan. Published by Elsevier B.V. and The Society of Powder Technology Japan. All rights reserved.
http://dx.doi.org/10.1016/j.apt.2013.03.020

772

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

There is signicant interest in the drying of natural biological

products, particularly foods and these types of biological extracts,
together with maximising the antioxidant levels in these materials
as a measure of their quality. For example, in the tray drying of red
peppers, Vega-Galvez et al. [15] found that the overall antioxidant
activity (as characterised by the total phenolic content [16,17]) of
the dried material was lower than the fresh material. However,
within the dried products, the antioxidant levels increased slightly,
but signicantly, when the air-drying temperature was increased
from 50 C to 90 C. At the same time, the vitamin C levels decreased with increasing temperature. There were at least two
explanations for the increase in overall antioxidant activity. The
rst explanation was that low air-drying temperatures correspond
to long air-drying times, which mean long periods where the material being dried is warm and wet, having a high water activity. It
was possible, but somewhat counter-intuitive, that this situation
may promote the degradation of antioxidant compounds relative
to high-temperature drying, which reduces the moisture content
and water activity of materials quickly. The degradation rates of
antioxidants, such as vitamin C, have been found to be high at both
high temperature and high water activities [5]. Whether low or
high-temperature drying results in the largest amount of antioxidant degradation depends on whether the temperature effect or
the water-activity effect on the degradation rate is greatest. A second explanation was that drying may have promoted the generation and accumulation of Maillard reaction products, such as
some melanoidins, which may have antioxidant properties.
Another work that reported a counter-intuitive temperature effect in drying (higher temperature giving higher antioxidant activity) was Que et al. [18]. They found higher antioxidant activity
from hot air drying than by freeze drying with pumpkin our, suggesting that the higher temperature conditions might have resulted in more antioxidant Maillard reaction products. They cited
Nicoli et al. [19], who suggested that the formation of Maillard
reaction products was responsible for the increase in the antioxidant activity for slightly-roasted coffee (after 10 min of roasting).
Yet another example of the motivation for this type of work
comes from the work of Piga et al. [20] on prunes, which result
from the drying of plums, so this work shows the effect of processing (drying) on a fruit (food) product. They point out that prunes
have high radical scavenging ability, the strongest of all fruit and
vegetable products in the human diet [21], and that prunes have
shown antimutagenic activity by in vitro tests [22]. In addition,
prunes contain substantial amounts of chlorogenic and neochlorogenic acids (collectively known as hydroxycinnamic acids), which
lower the glycemic index in humans [23,24] and inhibit LDL oxidation in vitro [25]. They studied the drying of two plum varieties, sugar and President, with air drying at two temperatures, 60 C and
85 C. The antioxidant activity increased signicantly at the higher
temperature (85 C) in the sugar variety. For the President variety,
drying at 60 C gave a lower antioxidant capacity than the fresh
fruit, but the antioxidant capacity was two and a half times greater
for fruit dried at 85 C than for fresh fruit. This surprising result
was attributed to the generation of a Maillard reaction intermediate product, hydroxymethylfural, which showed higher concentrations in the higher-temperature dried product and whose
concentrations were strongly correlated with the antioxidant
capacity.
Even more closely related to this work is that of Fang and Bhandari [26]. They spray dried bayberry juice with maltodextrin (DE
10), nding 96% retention of the total phenolic content and 94%
retention of total anthocyanins in the spray-dried products. However, they produced no model that simulated this behaviour, so
their explanation was qualitative in nature. Chiou and Langrish
[27] found that the encapsulation of the H. sabdariffa L. with natural fruit bres results in the production of a free-owing powder.

This work has continued the previous work to explore the retention of antioxidant activity. The previous work of Fang and Bhandari [26] has been extended to show how the effect of a key
controllable variable, the inlet gas temperature, on the antioxidant
activity of the spray-dried powder can be simulated and partly explained by a similar reaction kinetic expression to Goula and
Adamopoulus [5] with a parallel/plug ow simulation approach.
The experimental work will be described rst, before the modelling
and simulation approaches are presented.
2. Experimental
2.1. Materials and methods
2.1.1. Fibre and extract mixture
The ingredients used to make the slurry mixture used in the
spray-drying process consisted of ne-milled sugar cane bre
(<30 lm Fibacel, KFSU Pty., Ltd.) and rosella extract (H. sabdariffa
L., Vic Cherikoff Food Services Pty., Ltd.). The slurry mixture was
prepared using equal parts of bre and extract based on the mass
of the total dissolved solids. The combined extract was diluted
with water until the mixture had a solids concentration of 10%.
The slurry mixture was agitated, throughout the spray-drying process, with a magnetic stirrer to ensure that a homogenous solution
was fed into the dryer.
2.1.2. Drying conditions
The slurry mixture was fed into a Buchi B-290 mini spray dryer
with a height of 0.48 m. The settings were as follows: aspirator
(main drying air ow) rate 0.0127 kg/s (100%); inlet air temperatures ranging from 70 to 150 C; liquid feed pump rate of 1.9 mL/
min (5%); atomisation air ow rate 455 L/h (rotameter 35 to
the bottom of the gauge ball); and the nozzle cleaner set to
51 strikes/min (setting 9). These operating conditions resulted in
outlet air temperatures ranging from 41 to 82 C.
2.1.3. Extraction of samples
Two extraction techniques were employed throughout this
study. The rst was a very simple water extraction method, where
a spray-dried sample (1 g) was extracted in a 70 mL screw cap jar
with 20 mL of water. To ensure the powder was wetted, the jar was
gently shaken and then the solution was kept at room temperature
for 1 h. After the hour, the supernatant was removed, appropriately
diluted and analysed for antioxidant capacity via the Oxygen
Reducing Antioxidant Capacity (ORAC) assay [28,29]. The differences in solution concentrations were taken into account by
obtaining the total dissolved solids (TDSs) of the supernatant and
comparing the samples in terms of TDS.
The second technique was a slightly modied version of one
previously used by Prior and colleagues [28]. The spray-dried powder was mixed with a solution of acetone, water and acetic acid
(70:29.5:0.5). The mixture was vortexed for 30 s and then sonicated for 5 min, with the tube being inverted once during the sonication step to ensure that the powder was still adequately
suspended. After the sonication, the tube was left at room temperature for 10 min with occasional shaking. The tube was then centrifuged at 3500 rpm for 15 min before the supernatant was
removed and transferred to a 25 mL volumetric ask, where the
solution was made up to 25 mL through the addition of water.
2.1.4. Oxygen Reducing Antioxidant Capacity assay
The ORAC method used here was based on the one developed
by Prior and colleagues [28] since it was chosen by the US Department of Agriculture [29] as the reference method when assessing
other analytical techniques. The method used in this study in-

773

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

2.1.5. Calculations
The nal ORACFL values have been calculated by using a regression equation (Y = aX + b) between Trolox concentration (X) and
the net area under the FL decay curve (Y). Linear regression has
been used in the range of 6.2550 lM Trolox. The data have been
expressed as micromoles of Trolox Equivalents (TEs) per 100 g of
sample (lmolTE/100 g). The area under the curve (AUC) has been
calculated using the rectangle rule, and the net AUC has calculated
by subtracting the AUC of the blank from that of the sample.

0
0

Water Extraction

Prior Extraction

20

40

60

80

Prior Extraction

50

100

150

200

function of outlet air drying temperature. It shows a slight peakin-yield trend. Nevertheless, with the standard deviation in yields
in this and previous work of less than 4%, the peak is still signicant at the 90% condence level. Similar peak-in-yield trends have
been found in previous studies for biological materials such as fruit
juices, sucrose and a similar rosellabre mixture [3234], so this
peak yield has a rm precedent for sugar-rich foods, such as this
extract. Fig. 4 is from another batch of rosella juice and shows a
much more distinct peak-in-yield trend for a different batch of
the extract from the same type of raw material, showing the need
for care in generalising the yield result in Figs. 3 and 4 does support
the peak yield shown in Fig. 3. This highlights the differences in
trends that must be expected when dealing with natural food
products having biological variability that may be due to seasonal
uctuations or other factors. Nevertheless, the differences in composition have had no signicant effect on the peak yield. The picture does not change signicantly when the yield is plotted as a
function of the inlet gas temperature in Fig. 5 (for the same overall
data shown in Fig. 3, with the latest extract), because the inlet and
outlet temperatures are linked through heat losses and through the
heat of vaporisation, so low outlet temperatures correspond to low
inlet temperatures and high outlet temperatures to high inlet temperatures for the same spraying rate, as here.
An analysis of the variance between extracted powders (inputs
ve separate inlet gas temperatures and two extraction techniques;
outputs antioxidant activity; two replicates of each experiment;
therefore ANOVA sample size (n) = 20) was undertaken to assess
the effect of the different extraction methods on the amounts of
antioxidants extracted (the water technique and the Prior technique). The outcome was that there is no signicant difference between the two extraction methods with 95% condence.

60

Prior Extraction

Water Extraction

Fig. 2. Antioxidant capacity (lmol TE/100 g) as a function of outlet gas temperature

from the spray dryer.

15

Water Extraction

Inlet Temperature (C)

80

10

20

10

15

Yield (%)

Antioxidant capacity ('000

molTE/100g)

2.1.6. Experimental results and discussion

Fig. 1 shows the antioxidant content for every 100 g of the
rosella bre powder as a function of the outlet gas temperature.
The initial rosella extract had an antioxidant content of
32,100 lmol Trolox Equivalents (TEs)/100 g. The gure shows
(for the outputs) a mean of 12,600 with a minimum of 9000 and
a maximum of 16,200 lmol Trolox Equivalents (TEs)/100 g. The
expression of the results as a function of outlet temperature has
been selected since the outlet air temperature is more representative of particle temperatures throughout the spray-drying process
[30]. As discussed below, there is no signicant trend in these data.
Fig. 1 shows that there is no signicant effect of outlet gas temperature on the antioxidant content of the spray-dried rosellabre powder. Statistical analysis performed in Excel has shown a
very low R2 value of 0.01 and that with 95% condence the data
has a gradient between 0.1 and 0.1. This demonstrates a lack of
correlation within the data set, and hence it can be said that there
is no statistically signicant trend. The standard error in the ORAC
results (Fig. 1) is 17% of the mean, which is reasonable considering
biological variability. A similarly low degree of correlation is found
when the results are plotted as a function of inlet condition (Fig. 2),
highlighting the lack of correlation between the antioxidant capacity and the operating variables. Analysis of variance gives the standard error, which is most meaningfully given for the overall set of
conditions [31] as being related to the mean square error.
Fig. 3 shows the yields (powder recovery: solids out of cyclone/
solids in liquid feed  100%) produced from the spray dryer as a

20

Antioxidant capacity ('000

molTE/100g)

volved mixing 20 uL of appropriately diluted extract with 200 uL of

a uorescein solution (FL) and 37.5 uL of a 2,20 -Azobis(2-methylpropionamidine) dihydrochloride solution (AAPH). The uorescein
solution was made in three steps. The FL stock #1 solution was
made by diluting 22.5 mg of uorescein (sodium salt) in 50 mL of
0.075 M phosphate buffer solution (pH 7.0) (PBS). FL stock #2
was made by diluting 50 uL of stock #1 in 10 mL of PBS. The working FL was made by diluting 320 uL of FL stock #2 in 20 mL of PBS.
The AAPH solution was made by dissolving AAPH in PBS at a concentration of 17.2 mg/mL.

40

20

100

Outlet Temperature (C)

Fig. 1. Antioxidant capacity (lmol TE/100 g) as a function of outlet gas temperature
from the spray dryer.

0
0

50

100

Outlet Temperature (C)

Fig. 3. Yield as a function of outlet gas temperature (liquid feed rate 1.9 mL/min)
from the spray dryer, with the latest batch of extract.

774

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

60
1.9mL/min
50
Yield (%)

9.3mL/min
40

30
20
10
0
0

50
100
Outlet Temperature (C)

150

Fig. 4. Yield as a function of inlet gas temperature to the spray dryer (liquid feed
rate 1.9 mL/min), with the latest batch of extract.

80

Yield (%)

60

40

20

0
0

50

100
150
Inlet Temperature (C)

200

Fig. 5. Yield as a function of outlet gas temperature from the spray dryer, with a
previous batch of extract.

When analysing the results, it may (at rst) appear that there is
a signicant effect of temperature, since there is approximately
40% variation in the results seen in Fig. 1. However, if it is considered that an Arrhenius based temperature effect might be expected
[22], with the rate of (degradation) reactions doubling with every
10 C temperature rise, then over the 40 C change in outlet temperature that is experienced in this experimental matrix, it would
be expected that a 24 change in antioxidant content would occur.
Even with a short contact time, the temperature should affect the
extent of antioxidant retention. Hence, when the Arrhenius-based
temperature effect is considered, the 40% variation in antioxidant
content is not so signicant when a much larger change might
have been expected. In addition, the variation in the antioxidant
content is not systematic, as demonstrated by the low R2 value
for the correlation between antioxidant content and outlet temperature and the 95% condence interval for the slope of the relationship between the two parameters (0.1, 0.1), which suggests that
no signicant correlation exists between them. Therefore, these
experimental results suggest that the effect of temperature on
the antioxidant content during spray drying is not signicant. This
may be due to the relationship between the rate of removal of
moisture and the exposure of the thermally sensitive antioxidant
with respect to the rate and amount of antioxidant degradation.
In particular, increasing the temperature increases the degradation
rate, but it also decreases the moisture content more quickly,
which decreases the degradation rate, so the overall effect may
not be very strong or signicant, as seen in this study. This relationship (increasing temperature increases the degradation rate
but reduces the moisture content quickly, reducing the rate) can

be seen in the work of Goula et al. [5] for the loss of vitamin C in
the spray drying of tomato pulp.
The effect of outlet air drying temperatures on the yield is significant (Figs. 3 and 4). A peak-in-yield trend is usually expected when
drying materials containing low molecular-weight sugars [3234].
The peak-in-yield trend over this range of temperatures may be explained by the following physical processes. At the lower temperatures, the droplets are not dried sufciently and the particles leave
the dryer wet and hence sticky. As the temperature increases, the
particles leave the dryer with lower moisture contents and hence
are less sticky, resulting in a higher yield. In particular, the lower
moisture content results in an increase in the glass-transition temperature of the powder, as predicted by the Gordon and Taylor
equation. Therefore, the overall difference between the particle
and glass-transition temperature decreases, and the resulting powder is less sticky, which results in a greater product yield. Initially,
the glass-transition temperature (Tg) increases at a greater rate than
the particle temperature (T), which causes the stickiness to decrease
and hence the yield to increase. The point at which the temperature
of the powder increases at a faster rate than the glass-transition
temperature is the point when the yield starts to decrease. When
the overall difference between the particle and glass-transition temperatures starts to increase, so too does the stickiness of the powder
and hence the yield decreases. The lack of a distinct peak-in-yield
trend in this study emphasises the importance of verifying the
source of the sample solution, particularly when it is a natural product, as is the case here.
At rst sight, it might appear that yields of 60% are rather low.
For an industrial spray dryer producing milk powder, the commonly-accepted industry practice would be to achieve a yield of
over 99.9%. However, yields of around 60% are considered to be
good for small-scale spray dryers, such as these Buchi designs
[26,30,3234]. The fundamental reason for these low yields with
small-scale spray dryers can be found from the work of Hanus
and Langrish [35], who found high wall deposition rates of particles and hence low yields, in this scale of equipment due to the
high level of inertial deposition for (even) small droplets, where
the atomizer is very close to the dryer walls.
Figs. 3 and 4 come from different batches. In addition, there are
two different liquid feed rates used for the results in Fig. 4. Despite
all these differences, the maximum yield is always found when the
outlet gas temperature is around 6065 C. This outlet gas temperature, giving a maximum yield for this rosella/bre mixture, compares
closely with an outlet gas temperature of 65 C giving the maximum
yield in the work of Bhandari et al. [30] for sucrose. This similar result
suggests that the combination of sugars, food acids and other biomaterials in the rosella/bre extract may have a similar combined glasstransition temperature to sucrose (63 C). This nding is also supported by the work of Chiou and Langrish [27] that measured the
glass transition temperature of a similar rosella/bre mixture to be
in the range of 5768 C, which is in the same range as that of sucrose.
These results suggest that this rosella antioxidant-rich extract
can be dried in a spray dryer with the same levels of degradation
across this temperature range, and hence the focus should be
obtaining the maximum yields from the spray-drying process.
Therefore, the technology is available to industry to create an antioxidant-rich powder to be used as a nutraceutical product or as an
additive for food products.
The results obtained from these experiments suggest that there
is no signicant effect of spray-drying temperatures (inlet or outlet) on the antioxidant content of this rosellabre mixture. However, the drying temperature (inlet air temperature) does have a
signicant effect on the yield (recovery of inlet solids). To give
the maximum yield (recovery), which was 67% of the solids fed
into the dryer, the optimum inlet air temperature is 110 C, with
a liquid feed rate of 1.9 mL/min at an atomising air ow rate of

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

455 L/h, a main air ow rate of 38 m3/h and a solids concentration in the liquid of 10% (w/v). The peak yield was found at an outlet air temperature of 6065 C, regardless of the batch of material
or the liquid feed rate to the dryer.
To demonstrate the effects of high inlet and outlet air temperatures and fast drying rates on possible degradation reactions, a
plug-ow simulation has been used, as will be described next.
3. Simulation equations for plug-ow dryers
In this approach, the solids and the gas are assumed to ow parallel to each other in plug ow. The approach is better suited to
tall-form dryers than to short-form ones, due to the dryer shape.
It was reported rst by Keey and Pham [36], and it has been used
by Zbicinski [37], Truong et al. [38] (who give a particularly clear
description), Pearce [39] and Chiou et al. [3]. The droplet, air and
solids have been represented by the subscripts p, a, s, respectively,
in the following equations.

drying kinetics. The particle or droplet drying kinetics has been

modelled here using the concept of a characteristic drying curve
[41], as follows:

dmp
Ap K p
n
p  pv p
dh
U px v s

dU px

dh

"

1

qa
3 qa C D U R U px  U ax 1
g
4
U px
qp
qp dp

"
#
dU pr
3 qa C D U R U pr  U ar 1

4
U px
dh
qp dp
"

dU pt
3 qa C D U R U pt  U at 1

4
U px
dh
qp dp

Ap pdP

The droplet diameter is affected by shrinkage. The assumption

of balloon shrinkage without skin or crust formation is reasonable
for salt [42], allowing the droplet diameter, dp, to be estimated by
the equation:

UR
CD

q
U px  U ax 2 U pr  U ar 2 U pt  U at 2
24
1 0:15Re0:687

p
Rep

qa U R dp
la

!13
10

11

Here dpi (m) is the initial droplet diameter, X is the moisture content
(dry-basis) and qpi is the initial droplet density (kg m3). The gasphase mass-transfer coefcient is dened by the following
equations:

Mw K m
Ma P

Kp

12

qa Dv Sh

13

dp

The mass-transfer coefcient based on a partial-pressure driving force is Kp (kg m2 s1 Pa1), Km (kg m2 s1) is the mass-transfer coefcient based on a concentration driving force, Dv (m2 s1) is
the diffusivity of water through air, Sh () is the Sherwood number,
Mw (g mol1) is the molecular weight of water, Ma (g mol1) is the
molecular weight of air, and P (Pa) is the total pressure. The diffusivity may be estimated from the following equation [43]:

1:17564  109  T 1:75

abs  101325
P

Dv

The Schmidt number (Sc) is given by Eq. (15), so that the Sherwood number may be estimated by a RanzMarshall Eq. (16) [44]:

Sc

la
qa Dv

14

15

0:33
Sh 2:0 0:6Re0:5
p Sc

16

Here the absolute temperature of the droplet or particle is Tabs (K).

Here l is the viscosity (kg m1 s1). The radial distance, r, of particles/droplets has been calculated as a function of axial distance (h)
from the atomizer, as follows:

dr U pr

dh U px

qpi  1000
qp  1000

1X
q
1 X qqs s

qp

The following equation denes the particle Reynolds number

(Rep):

Rep

dp dpi

Km

The parameters have the following meanings. The density is q

(kg m3), the droplet diameter is dp (m), the relative velocity between the droplet and the air is UR (m s1), and the drag coefcient
is CD. The subscripts a and p refer to the air and the particle or
droplet, respectively. The parameters UR and CD may be estimated
by equations given by Rhodes [40], as follows:

Here, the parameters include mp, the mass of the particle or droplet
(kg), pvs (Pa), the partial pressure of the surface of the droplet, pvb
(Pa), the partial pressure of water vapour in the bulk air, n (), the
relative drying rate, Ap (m2), the surface area of the droplet, and
Kp (kg m2 s1 Pa1), the mass-transfer coefcient based on a partial-pressure driving force. The surface area of the droplet (Ap) is
simply given by the following equation:

3.1. Droplet trajectory equations

The droplet trajectory equations essentially arise from droplet
axial, radial and tangential momentum balances. The parameters
Up and Ua, respectively, represent the particle and air velocities
(m s1). The axial, radial and tangential components are represented by the subscripts x, r and t, and the symbol h represents
the axial distance from the atomizer.

775

3.2. Droplet mass-balance equations

There are a number of approaches for estimating the unsteadystate mass balance for the droplets, which may be described as

3.3. Droplet energy-balance equations

The unsteady-state thermal-energy (heat) balance for the droplet or particle follows as the following equation [38]:
dm

dT p pdp ka NuT a  T p dhp U px Hfg

dh
ms C ps XC pw U px

17

The Prandtl number (Pr) is given by Eq. (18), allowing the

Nusselt number (Nu) to be estimated by a RanzMarshall
Eq. (19) [44]:

776

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

C pa la
ka

Pr

18

0:33
Nu 2:0 0:6Re0:5
p Pr

19

Other product and particle properties are needed, as follows

[38]:

ms

Xmp
100

20

Hfg 2:792  106  160T abs  3:43T 2abs

21
1

1

The thermal conductivity of humid air is ka (W m K ), the latent heat of water vaporisation is Hfg (J kg1), the mass of solids in
the droplet is ms (kg) and the specic heat capacity is Cp (J kg1 K1).
3.4. Drying medium mass and energy balance equations
The mass-balance equation for the drying air connects the mass
of moisture gained by the gas with the mass of moisture lost by the
droplets or particles [38]:

X 
dY b droplets

dh

dmp
dh


ndroplets

22

The ow rate of droplets is ndroplets (number s1), and the mass

ow rate of the dry air is G (kg s1). For the thermal energy balance,
the gain in humid heat capacity and heat loss from the dryer equals
the energy lost by the droplets or particles, to give:

dHh
1

G
dh

!

X 
dT p
UAT a  T amb
ndroplets

ms C ps XC pw
L
dh
droplets
23

Parameters here include the length of the spray-drying chamber

(L, m), the heat-transfer coefcient for heat loss from the dryer (UA,
W K1), and the enthalpy of the humid air is Hh (J kg1).
3.5. Drying kinetics
The concept of a characteristic drying curve has been used here
to predict the drying kinetics for hindered drying. This approach
has behaved reasonably in the work of Keey and Pham [36] and
Langrish and Kockel [45]. The concept of the characteristic drying
curve [41] suggests that the relative drying rate (n), which is the actual drying rate of the solids relative to the unhindered drying rate,
is some function of the characteristic moisture content, as follows:

n f U if X 6 X cr
n 1 if X P X cr

X  Xe
X cr  X e

24

where U is the characteristic moisture content (), X is the actual average moisture content of the particle (kg kg1), Xcr is the
critical moisture content (kg kg1) and Xe is the equilibrium moisture content (kg kg1). Below the critical moisture content of the
droplet, hindered drying occurs. The equilibrium moisture content
is the moisture content of the particle in equilibrium with the gas,
where no further reduction in moisture content occurs. A linear
falling rate curve (f = U) has been used in this work [38].
3.6. Degradation kinetics
Degradation reactions can be modelled with a reaction rate
integrated over the particle residence times through the dryer.

The approach to modelling the degradation rate constant was similar to that for modelling diffusion in solids [46], so the implicit
assumption was made that the degradation reaction was limited
by the rate of diffusion of reactants in the degradation process to
and from reaction sites. The equations used were similar to those
used by Goula et al. [4] in that the effects of temperature and moisture content were separately identied:

kdegrade A expBX exp

C
1
Co



EA
RT

25

tf

kdegrade Ctdt

26

Here kdegrade is the degradation rate constant, A and B are constants,

X is the moisture content of the particle (kg kg1), T is the absolute
temperature of the particle (K), R is the universal gas constant
(8.314 J mol1 K1), EA is the activation energy for the degradation
process, C is the concentration of the chemical being degraded, Co
is the initial concentration, t is time and tf is the residence time of
the particle in the dryer. Assuming that the degradation reaction
is limited by diffusional processes of mass transfer to and from reaction sites, the activation energy (EA) divided by the gas constant (R)
was taken to be around 5000 K [46], giving an activation energy of
around 40 kJ mol1. For diffusion of moisture, the constant B in Eq.
(25) has been suggested to be around 20 kg kg1, so this value has
been used here, again assuming that the degradation reaction is
limited by diffusional processes.
4. Model solution
The model, simulating co-current spray drying, and incorporating the experimentally-estimated heat loss from the Buchi B-290
spray dryer, is a set of at least eight ordinary differential equations,
with additional equations for each particle size class, representing
the mass-transfer rates and mass and energy balances for each size
class. The equations were implemented and solved in Mathworks
Matlab package using the inbuilt Matlab stiff ordinary differential
equation solver function; ode23s. The stiff equation solver, ode23s
uses the second and third order RungeKutta methods to solve the
ordinary differential equations.
The spray dryer simulated in this work has been the Buchi B290
design used in the experiments, with a height of 0.48 m and a
diameter of 0.155 m [3]. The initial and boundary value conditions
of some parameters must be entered. The initial values of the radial
and tangential components of the inlet air velocity were both set to
0 ms1, because the air forced into the dryer is introduced axially.
The initial values of other parameters, such as axial components of
the inlet air velocity and inlet air temperatures, were based on the
experimental conditions used and described in the following sections. The product of the overall heat-transfer coefcient for heat
loss from the outside of the dryer (U) and the outside surface area
of the dryer (A) has been found to be (UA) 20 W K1 [34], and this
value has been used here.
The model assumes plug ow through the spray dryer from the
inlet to the outlet. It is assumed that no back-mixing of particles
occurs. While this assumption is not strictly true, the plug-ow
assumption was adequate to predict the wall deposition behaviour
in the work of Hanus and Langrish [35]. For exploring the effects of
different operating conditions, the use of such a simple model for
the drying behaviour to predict the relative effect of different operating conditions has been found by workers in the past, such as
Keey and Pham [36], to be adequate, particularly for the tall-form
design of spray dryer used here, where tall-form spray dryers have
lengths that are of the order of ve times their diameters [47]. This
plug-ow reaction approach has known weaknesses in terms of

777

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

5. Dryer dimensions, inlet and boundary conditions

To illustrate the consequences of the application of the concept
of a characteristic drying curve (with a linear falling rate curve, as
in Chiou et al. [3]), the model has been simulated with the following inputs (initial and boundary conditions and process parameters), as described in the experimental section above.
Liquid solution pump rate of 1.9 mL/min (5% of the maximum),
an aspirator (main drying air ow) rate of 0.0127 kg/s, an atomising air ow rate of 455 L/h, a solids concentration in solution of 5%
w/v, no constant-rate period (critical moisture content = initial
moisture content), and inlet gas temperatures between 70 C and
150 C.

The simulated heat and mass-transfer results, for the changes in

the moisture contents of the largest particles (Fig. 6) and for the
changes in the particle temperatures (Fig. 7) as a function of distance from the top of the spray dryer, show many features that
are common to all cocurrent ow dryers. As would be naturally expected, moisture contents reduce more rapidly at higher inlet gas
temperatures. Less intuitive is the observation, common to many
cocurrent dryers, that the peak particle temperature occur not at
one of the end points of the dryer (inlet or outlet) but in the middle
of the dryer. This result is due to the differing rates of heat and
mass transfer, with heat transfer being more rapid than mass
transfer, leading to the particle being initially heated very quickly,
followed by a period where both the particle and the gas temperatures decrease due to energy being used in evaporation. The sizes
given in Fig. 7 are the droplet size classes from the atomizer, with
ten size classes as shown in Table 1, as reported by Hanus and
Langrish [35]. Each size class has ten percent of the mass from
the atomizer. In Fig. 7, sizes 1, 8, 9 and 10 are therefore sizes of
2.4 lm, 22.0 lm, 31.1 lm and 55.3 lm.

Moisture content (kg kg-1)

6.0
5.0
4.0

70 C
3.0

150 C
2.0
1.0

0.20

140
120
Size 1, 70 C
Size 8, 70 C
Size 9, 70 C
Size 10, 70 C
Size 1, 150 C
Size 8, 150 C
Size 9, 150 C
Size 10, 150 C

100
80
60
40
20
0
0.0

0.2

0.4

0.6

Distance from top of dryer (m)

Fig. 7. Predicted particle temperatures as a function of distance from the top of a
Buchi B290 spray dryer at two inlet gas temperatures (70 C and 150 C). Sizes 1, 8,
9 and 10 correspond to initial droplet sizes of 2.4 lm, 22.0 lm, 31.1 lm and
55.3 lm.

Table 1
Droplet size distribution from the atomizer in this dryer (Hanus and Langrish [35]).

6. Results and discussion

0.0
0.00

160

Particle temperature (oC)

the effects of different designs on the ow patterns in the dryers

[46], but these plug-ow approaches are computationally very
tractable compared with Computational Fluid Dynamics simulations (CFD). Time-dependent CFD calculations, when properly performed, take days, weeks and months [48], while this plug-ow
reactor approach can be solved in minutes.

0.40

0.60

Distance from top of dryer (m)

Fig. 6. Predicted particle moisture contents for the largest particle size class
(30 lm) as a function of distance from the top of a Buchi B290 spray dryer at two
inlet gas temperatures (70 C and 150 C).

Particle size (lm)

2.4, 4.4, 6.1, 8.1, 10.3, 13.1, 16.8, 22.0, 31.1, 55.3

For crystallization-in-drying, this approach has already been

shown to be useful in predicting trends in crystallization during
drying by Chiou et al. [3] and Woo et al. [49]. For the degradation
process, the simulations have focussed on the experimental data
described earlier.
Considering Eq. (25) again, there are potentially competing effects of temperature and moisture content. In this equation, high
material temperatures lead to high degradation rates, but lower
material moisture contents lead to lower reaction rates due to
the effect of low moisture content in restricting the diffusional ow
of moisture and other reactants.
Fig. 8 shows the experimental and predicted relationships between the antioxidant content of the rosella bre mixture as a function of the outlet air temperature. This gure shows how much the
antioxidant level in the resulting powder decreased experimentally
relative to the initial level (100%), as a function of the outlet air temperature. The graph shows a minimum value of 28% and a maximum
value of 51%, but there is no signicant trend. Fig. 9, for the effect of
inlet air temperature on the predicted and measured amounts of
degradation, also shows no signicant effect of temperature. Hence
the trade-off between temperature and moisture content appears to
be fundamental to the behaviour of this material in a spray dryer. By
the words trade-off, we mean that increasing the temperature increases the rate of degradation but reduces the moisture content
quickly reduces this rate. This trade-off is suggested to apply, regardless of whether or not the inlet temperature or the outlet temperature is considered in either the predicted results from the simulation
or the actual experimental results.
The plug-ow simulation approach, and a value for the constant
A in Eq. (25) of 800 s1, shows that the inlet air temperature, over a
range from 70 C to 150 C, has very little effect on the nal antioxidant concentration due to these competing effects, with the predicted nal percentages of the initial antioxidant levels in Figs. 8
and 9 being 43% at an inlet air temperature of 150 C and 38% at
an inlet air temperature of 70 C. These results, and their insensitivity to changing inlet air temperature, suggest that a basic assumption in the degradation kinetics, that the reactions are limited by
diffusional processes for water, reactants and products within the
particles, may be reasonable for many spray-dried products.

778

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

Percentage of initial
antioxidants (%)

60

content are the highest. However, for larger particles that carry
much of the mass of liquid, degradation occurs throughout a part
(0.04 m out of a total height of 0.48 m) of the dryer that is signicantly more than the active height for the smaller particles.

50
40
30

7. Conclusions
Water Extraction
Prior Extraction

20
10

Simulation

0
0

20

40

60

80

100

Outlet Temperature (C)

Fig. 8. Actual experimental and predicted percentages of initial antioxidants as a
function of outlet gas temperature to the spray dryer.

Percentage of initial
antioxidants (%)

60
50

In the experiments, an outlet gas temperature of 6065 C has

been found to give a peak yield of between 65% and 70%, for both
batches of material and two liquid feed rates. The antioxidant
retention of the sample was not signicantly affected by the inlet
or outlet air temperatures. The simulations suggest that the competing effects of increasing air temperature and decreasing water
activity (at higher inlet air temperatures) on the degradation kinetics have been responsible for this result. It may be more important
to obtain higher product yields compared with minimising the
degradation by changing the operating conditions for these
extracts.
Acknowledgements

40

Financial support from Lang Technologies P/L and the Australian Research Council is gratefully acknowledged under the Linkage Grants program. Thanks are also due to Vic Cherikoff Foods
P/L for the bioactive extract and KFSU P/L for the sugar cane Fibacel
bre.

30
20

Water Extraction

10

Prior Extraction
Simulation

0
0

50

100

150

200

Inlet Temperature (C)

Fig. 9. Actual experimental and predicted percentages of initial antioxidants as a
function of inlet gas temperature from the spray dryer.

Fig. 10. Predicted degradation rates as a function of distance from the top of a Buchi
B290 spray dryer at an inlet gas temperature of 150 C.

Fig. 10, for an inlet air temperature of 150 C, shows that product degradation occurs throughout the dryer. For small particles,
most of the degradation is predicted to occur very close to the
top of the dryer, where both particle temperature and moisture

References
[1] K. Masters, Spray Drying Handbook, Halsted Press, New York, 1996. pp. 165
256.
[2] M. Kwapinska, I. Zbicinski, Prediction of nal product properties after
cocurrent spray drying, Drying Technology 23 (8) (2005) 16531665.
[3] D. Chiou, T.A.G. Langrish, R. Braham, Partial crystallisation behaviour during
spray drying: simulations and experiments, Drying Technology 26 (1) (2008)
2738.
[4] A. Goula, K.G. Adamopoulos, N.A. Kazakis, Inuence of spray drying conditions
on tomato powder properties, Drying Technology 22 (5) (2004) 11291151.
[5] A. Goula, K.G. Adamopoulos, Retention of ascorbic acid during drying of tomato
halves and tomato pulp, Drying Technology 24 (1) (2006) 5764.
[6] T.A.G. Langrish, Multi-scale mathematical modelling of spray dryers, Journal of
Food Engineering 93 (2) (2009) 218228.
[7] J. Yan, X.D. Chen, A fundamental model of particle deposition incorporated in
CFD simulations of an industrial milk spray dryer, Drying Technology 28 (8)
(2010) 960971.
[8] K.L. Wolfe, R.H. Liu, Cellular antioxidant activity (Caa) assay for assessing
antioxidants, foods, and dietary supplements, Journal of Agricultural and Food
Chemistry 55 (22) (2007) 88968907.
[9] S.P. Wong, L.P. Leong, J.H.W. Koh, Antioxidant activities of aqueous extracts of
selected plants, Food Chemistry 99 (4) (2006) 775783.
[10] Q. Chen, H. Shi, C.-T. Ho, Effects of rosemary extracts and major constituents on
lipid oxidation and soybean lipoxygenase activity, Journal of the American Oil
Chemists Society 69 (10) (1992) 9991002.
[11] H.C. Grice, Safety evaluation of butylated hydroxyanisole from the perspective
of effects on forestomach and oesophageal squamous epithelium, Food and
Chemical Toxicology 26 (8) (1988) 717723.
[12] J. Kubola, S. Siriamornpun, Phenolic contents and antioxidant activities of
bitter gourd (Momordica charantia L.) leaf, stem and fruit fraction extracts
in vitro, Food Chemistry 110 (4) (2008) 881890.
[13] M. Senevirathne, S.-H. Kim, N. Siriwardhana, J.-H. Ha, K.-W. Lee, Y.-J. Jeon,
Antioxidant potential of Ecklonia cava on reactive oxygen species scavenging,
metal chelating, reducing power and lipid peroxidation inhibition, Food
Science and Technology International 12 (1) (2006) 2738.
[14] G. Gradinaru, C.G. Biliaderis, S. Kallithraka, P. Kefalas, C. Garcia-Viguera,
Thermal stability of Hibiscus sabdariffa L. anthocyanins in solution and in solid
state: Effects of copigmentation and glass transition, Food Chemistry 83 (3)
(2003) 423436.
[15] A. Vega-Glvez, K. Di Scala, K. Rodrguez, R. Lemus-Mondaca, M. Miranda, J.
Lpez, M. Perez-Won, Effect of air-drying temperature on physico-chemical
properties, antioxidant capacity, colour and total phenolic content of red
pepper (Capsicum annuum, L. Var. Hungarian), Food Chemistry 117 (4) (2009)
647653.
[16] R.L. Prior, G. Cao, A. Martin, E. Soc, J. McEwan, C. OBrian, N. Lischner, M.
Ehlenfeldt, W. Kalt, G. Krewer, C.M. Mainland, Antioxidant capacity as
inuenced by total phenolics and anthocyanin content, maturity, and variety
of Vaccinium species, Journal of Agricultural and Food Chemistry 46 (7) (1998)
26862693.

T.A.G. Langrish, R. Premarajah / Advanced Powder Technology 24 (2013) 771779

[17] H. Wang, G. Cao, R.L. Prior, Total antioxidant capacity of fruits, Journal of
Agricultural and Food Chemistry 44 (3) (1996) 701705.
[18] F. Que, L. Mao, X. Fang, T. Wu, Comparison of hot air-drying and freeze-drying
on the physicochemical properties and antioxidant activities of pumpkin
(Cucurbita moschate Duch.) ours, International Journal of Food Science and
Technology 43 (7) (2008) 11951201.
[19] M.C. Nicoli, M. Anese, L. Manzocco, C.R. Lerici, Antioxidant properties of coffee
brews in relation to the roasting degree, LWT Food Science and Technology
30 (3) (1997) 292297.
[20] A. Piga, A. Del Caro, G. Corda, From plums to prunes: inuence of drying
parameters on polyphenols and antioxidant activity, Journal of Agricultural
and Food Chemistry 51 (12) (2003) 36753681.
[21] R.L. Prior, J.A. Joseph, G. Cao, B. Shukitt-Hale, Can foods forestall aging?,
Agricultural Research 47 (2) (1999) 1517
[22] R.I.M. Chan, R.H.C. San, H.F. Stich, Mechanism of inhibition of N-methyl-Nnitroso-guanidine induced mutagenesis by phenolic compounds, Cancer
Letters 31 (1) (1986) 2734.
[23] J.C. Brand-Miller, Importance of glycemic index in diabetes, American Journal
of Clinical Nutrition 59 (3 supplement) (1994) 747752.
[24] H. Hemmerle, H.J. Burger, P. Below, G. Schubert, R. Rippel, P.W. Schindler, E.
Paulus, A.W. Herling, Chlorogenic acid and synthesis chlorogenic acid
derivatives: novel inhibitors of hepatic glucose-6-phosphate translocase,
Journal of Medicinal Chemistry 40 (2) (1997) 137145.
[25] J.L. Donovan, A.S. Meyer, A.L. Waterhouse, Phenolic composition and
antioxidant activity of prunes and prune juice (Prunus domestica), Journal of
Agricultural and Food Chemistry 46 (4) (1998) 12471252.
[26] Z. Fang, B.R. Bhandari, Effect of spray drying and storage on the stability of
bayberry polyphenols, Food Chemistry 129 (3) (2011) 11391147.
[27] D. Chiou, T.A.G. Langrish, Crystallization of amorphous components in spraydried powders, Drying Technology 25 (9) (2007) 14271435.
[28] R.L. Prior, H. Hoang, L. Gu, X. Wu, M. Bacchiocca, L. Howard, M. HampschWoodill, D. Huang, B. Ou, R. Jacob, Assays for hydrophilic and lipophilic
antioxidant capacity (Oxygen Radical Absorbance Capacity (Orac)) of plasma
and other biological and food samples, Journal of Agricultural and Food
Chemistry 51 (11) (2003) 32733279.
[29] US Department of Agriculture (USDA), Nutrient Data Laboratory, Beltsville
Human Nutrition Research Centre (BHNRC), Agricultural Research Service
(ARS), Oxygen Radical Absorbance Capacity (ORAC) of Selected Foods 2007.
Beltsville, Maryland, 2007.
[30] B.R. Bhandari, N. Datta, R. Crooks, T. Howes, S. Rigby, A semi-empirical
approach to optimise the quantity of drying aids required to spray dry sugarrich foods, Drying Technology 15 (10) (1997) 25092525.
[31] D.C. Montgomery, Design and Analysis of Experiments, seventh ed., John
Wiley, Hoboken, NJ, 2008. 25, 511, 522.
[32] B.R. Bhandari, A. Senoussi, E.D. Dumoulin, A. Lebert, Spray drying of
concentrated fruit juices, Drying Technology 11 (5) (1993) 10811092.

779

[33] M. Imtiaz-Ul-Islam, T.A.G. Langrish, Comparing the crystallization of sucrose

and lactose in spray dryers, Food and Bioproducts Processing 87 (2) (2009) 87
95.
[34] R. Premarajah, T.A.G. Langrish, The effect of spray drying conditions on yields
of bioactive extracts, in Proceedings of Chemeca 2009, Perth, Australia, 2009.
[35] M. Hanus, T.A.G. Langrish, Re-entrainment of wall deposits from a laboratoryscale spray dryer, Asia-Pacic Journal of Chemical Engineering 2 (2) (2007)
90107.
[36] R.B. Keey, Q.T. Pham, Behaviour of spray dryers with nozzle atomizers,
Chemical Engineer (London) 311 (1976) 516521.
[37] I. Zbicinski, Development and experimental verication of momentum, heat
and mass transfer model in spray drying, Chemical Engineering Journal 58 (2)
(1995) 123133.
[38] V. Truong, B.R. Bhandari, T. Howes, Optimization of co-current spray drying
process of sugar-rich foods. Part I moisture and glass transition temperature
prole during drying, Journal of Food Engineering 71 (1) (2005) 5565.
[39] D.L. Pearce, A novel way to measure the concentration of a spray in a spray
dryer, Drying Technology 24 (6) (2006) 777781.
[40] M. Rhodes, Introduction to Particle Technology, John Wiley, New York,
Chicester, 1998. pp. 24.
[41] R.B. Keey, Introduction to Industrial Drying Operations;, Pergamon Press,
Oxford, UK, 1978. 376.
[42] T.A.G. Langrish, I. Zbicinski, The effects of air inlet geometry and spray cone
angle on the wall deposition rate in spray dryers, Transactions of the
Institution of Chemical Engineers 72 (A) (1994) 420430.
[43] R.H. Perry, D.W. Green, J.O. Maloney, Perrys Chemical Engineers Handbook,
McGraw-Hill, New York, 1997. 2-166.
[44] W.E. Ranz, W.R. Marshall, Evaporation from drops, Chemical Engineering
Progress 148 (4) (1952) 141146.
[45] T.A.G. Langrish, T.K. Kockel, Implementation of a characteristic drying curve
for milk powder using a computational uid dynamics simulation, Chemical
Engineering Journal 84 (1) (2001) 6974.
[46] R.B. Keey, T.A.G. Langrish, J.C.F. Walker, Kiln Drying of Lumber, SpringerVerlag, Heidelberg, Germany, 2000. pp. 114115.
[47] D. Reay, Fluid ow, residence time simulation and energy efciency in
industrial dryers, in: M. Roques (Ed.), Proceedings of 6th International Drying
Symposium (IDS88), Versailles, France, 1988, KL1KL8.
[48] D.F. Fletcher, B. Guo, D. Harvie, T.A.G. Langrish, J.J. Nijdam, J. Williams, What is
important in the simulation of spray dryer performance and how do current
CFD models perform?, Applied Mathematical Modelling 30 (11) (2006) 1281
1292
[49] M.W. Woo, S. Rogers, C. Selomulya, X.D. Chen, Particle drying and
crystallization characteristics in a low velocity concurrent pilot scale spray
drying tower, Powder Technology 223 (2012) 3945.