June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.

2 238

On the water desorption of fruits: a case study of watermelon

(Citrullus Lanatus)

Réda Khama1,2*, Angélique Léonard2

(1. Department of Process Engineering, University of Batna 2, 5000 Batna, Algeria;

2. Department of Chemical Engineering, PEPs, University of Liège, 4000 Liège, Belgium)

Abstract: Studies on drying of the watermelon seeds or rinds are available, but not on drying of the fruit flesh. Data on
the sorption of watermelon fruit (Citrullus Lanatus), which are useful and important when identifying optimal food
drying and storage conditions, are also not available. Therefore, the main goal of this study is to investigate the water
desorption behavior of watermelon using the dynamic method requiring the automated dynamic vapor sorption (DVS)
technique based on a humidity-generating device. The moisture desorption isotherms are determined at three different
temperatures (20°C, 30°C and 40°C) over a relative humidity range of 0-80%. In addition, 11 different models are
applied to the experimental data in order to select the appropriate desorption curve equation. The models are compared
using the correlation coefficient, the coefficient of determination, the standard error, the reduced chi-squared, the mean
bias error and the root mean square error; they are predicted by a non-linear regression analysis using the Curve Expert
software and MS Excel computer program. The effect of the temperature on the constants of the found equation is also
examined. Besides, the desorption isosteric heat of watermelon is determined using the Clausius-Clapeyron relation.
Contrary to long times taken in the static method, the results show that 4.5 days are widely sufficient to reach the
equilibrium via the DVS device. In addition, the obtained curves are of the type III and the Henderson model best fits
-1
the experimental data. In the interval 0.05-0.45 kg water.kg dry matter of equilibrium moisture, the values of the isosteric
-1
heat vary from 65.02 to 113.25 kJ mol . Finally, the equation giving the isosteric heat of desorption v.s. the equilibrium
moisture content is obtained.
Keywords: isotherms, drying, DVS, isosteric heat, experimental methodology, statistical analysis

Citation: Khama, R. and A. Léonard. 2024. On the water desorption of fruits: a case study of watermelon (Citrullus
Lanatus). Agricultural Engineering International: CIGR Journal, 26(2): 238-252.

1 Introduction The thermodynamic concept of water activity is used

in the drying applications through these sorption
The water sorption properties of solid materials
isotherms, which represent the water activity
are recognised as critical factors in determining their
variation versus the moisture content variation of a
storage, stability, processing and application
body at equilibrium and at a defined constant
performance (Surface Measurement Systems
temperature (Arlabosse et al., 2003; Garbalińska et al.,
Company, 2023). These properties are often
2017).
represented by curves so-called sorption isotherms.
Knowing the moisture content of a product is
insufficient to predict its stability, making it
Received date: 2023-12-14 Accepted date: 2024-03-29 necessary to also know its water activity, a
*Corresponding author: Réda Khama, Ph.D Associate
thermodynamic property describing the interactions
Professor, Department of Process Engineering, University of
between water molecules and the food matrix
Batna 2, 5000 Batna, Algeria. Tel: +213 (0) 540640822,
Emails: [email protected]; [email protected].
(Caballero-Cerón et al., 2015).
June, 2024 On the water desorption of fruits: a case study of watermelon Vol. 26, No.2 239

Due to the complexity of the sorption phenomena, loss or gain and thus contributing to measurement
the isotherms of sorption cannot be predicted by the errors. (iii) The inability to measure the kinetics of
theory but must be experimentally established for water sorption/desorption uptake and release in real
every product. Accordinr to Djendoubi Mrad et al. time limiting the utility of these static measurement
(2013), the form of sorption isotherms is chiefly techniques. (iv) The risk of mould formation or other
shaped by the structure and composition of the food sample contamination resulting from lengthy
material. In this way, the experimental determination equilibration times and exposure to variable
is required because the methods of prediction are not laboratory temperatures and humidities. In addition,
able to simulate complex systems such as foods; the RH levels of the air around the samples, or even
temperature and pressure are also important factors the water activity of the samples, are limited by the
(Djendoubi Mrad et al., 2013). various available salt solutions (Yang et al., 2020).
In order to extend the life duration of fruits, their In the dynamic systems, the sorption curves can
water activity does not have to exceed 0.6 (Djendoubi be determined by dynamic vapor sorption (DVS),
Mrad et al., 2012). Then, it is essential to study the which is a modern technique producing sorption data
sorption characteristics of the dried products with with high accuracy under controlled conditions in a
water activities situated between 0 and 0.6 short period of time (Garbalińska et al., 2017;
corresponding to the interval 0-60% of the air relative Arlabosse et al., 2003; Argyropoulos et al., 2011).
humidity (RH) (Djendoubi Mrad et al., 2012). Such a system is a real alternative, indeed the DVS
Numerous methods are available for the Intrinsic is a highly sensitive, accurate and rapid
determination of water sorption isotherms (Basu et al., means for automated determination of moisture
2006). They can be divided into three categories, sorption properties of solids (Surface Measurement
namely gravimetric, manometric and hygrometric Systems Company, 2023).
(Deshmukh et al., 2017; Khama, 2016). The Haoua et al. (2015) examined the desorption
gravimetric method consists of fixing the pressure isotherms of unblanched and blanched samples of
above the sample and measuring mass changes potato exposed to a range of different temperatures
continuously or discontinuously in static or dynamic (30°C, 35°C and 40°C), using the DVS system. The
systems (Arlabosse et al., 2003; Basu et al., 2006; obtained curves were of the type II in BDDT
Khama, 2016). classification and the Oswin model best fitted the
In the static systems, the standardized saturated experimental data.
salt solution (SSS) method is the most common In the work of Goula et al. (2008), adsorption
technique using thermostat jars filled with isotherms of tomato pulp spray dried in dehumidified
supersaturated saline solutions to maintain the desired air were determined at six temperatures (20°C-70°C),
RH (Basu et al., 2006). Accoding to (Surface using a gravimetric technique. The experimental data
Measurement Systems Company, 2023), these were then fitted to six different models and the GAB
manual jar methods have disadvantages: (i) The model was found to be the most suitable to predict the
prolonged period of time (often many days and adsorption isotherms. The isosteric heat of sorption
commonly many weeks) taken for the samples to was determined from the equilibrium adsorption data
reach equilibrium, which necessitates the use of large using the Clausius–Clapeyron equation and the
samples sizes (typically 10-100 g) to compensate for isosteric heats of sorption were found to decrease
inherent analytical balance inaccuracies. (ii) exponentially with increasing moisture content. In
Measurement errors associated with the required addition, the enthalpy–entropy compensation theory
removal of samples from storage containers for was applied to the sorption isotherms and indicated
periodic weighing often results in extraneous mass an enthalpy controlled sorption process.
June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.2 240

The research of Bingol et al. (2012), focused on According to Maoto et al. (2019), snacking has
the sorption behavior of two popular medium rice become a habit among consumers in recent years.
varieties (M202 and M206), in California by a new These snacking trends together with unhealthy life
approach using DVS method for elucidating the styles may promote the formation of Reactive
differences in fissuring resistance. The moisture Oxygen Species (ROS) and free radicals which
sorption isotherms of rough, brown and white rice results in non-communicable diseases. As a result,
and husk were determined at 25°C over a water replacing energy-dense snacks with fruits is
activity range of 0-0.98. encouraged. As a functional food, watermelon is a
Although it was found that the sorption isotherms quintessential summer fruit loaded with
of different forms of M202 and M206 were similar, phytochemicals.
M202 absorbed and desorbed moisture faster than The phytochemicals have been reported for their
M206 which might make it more susceptible to pharmacological activities and therapeutic properties
fissuring. All obtained moisture sorption isotherms such as analgesic, laxative, antigiardial,
exhibited the sigmoid (Type II) shape and hysteresis gastroprotective, hepatoprotective, antibacterial,
was observed for all forms of rice. antifungal, antimicrobial, antiulcer, antioxidant, and
Absorption curves obtained using DVS method anti-inflammatory (Maoto et al., 2019).
were compared with the traditional SSS method. For Studies on the drying of watermelon seeds or
all forms of rice, the maximum difference for rinds are available, but on the drying of the fruit itself
equilibrium moisture content value between two are not. Data on water sorption of watermelon fruit is
methods was observed at 0 and 98% relative also not available. Therefore, in this work, the
humidities, which was approximately 8%-11% and desorption isotherms and isosteric heat of watermelon
7%-9%. Although at a single RH point equilibrium are determined by a gravimetric method requiring the
with DVS was attained much faster than SSS method, DVS. Indeed, regarding the previous published
obtaining a full isotherm with more than 10 data scientific works on watermelon, the theme of our
points might make SSS method more feasible for low research can be considered as the first one and the
diffusion materials. Among the select equations to results as a novel contribution.
describe sorption behavior of different forms of rice,
2 Materials and methods
Peleg equation gave the best fit for all forms of rice.
According to Garbalińska et al. (2017), the 2.1 Watermelon
Fresh watermelons are bought from a local market
investigation of Anderberg and Wadsö (2008)
in Liège, Belgium at their commercial maturity
contains a comparison of the sorption isotherms of
according to some indices namely skin colour, shape,
self-levelling flooring compounds with water to-
size and firmness. For each experiment, small
binder ratio of about 1.0 and about 0.7 obtained by
samples are used with a thickness of about 2.5 mm
using the traditional method and the DVS method.
and a mass varying between 33.49-36.23 mg.
Good compatibility of results was achieved for
2.2 Dynamic vapor sorption system (DVS)
relative air humidity ranges between 10% and 95%.
The experiments are conducted within the PEPs
Koocheki et al. (2007) mentioned that watermelon
(Products, Environment, and Processes) group at the
is a tropical fruit, which grows in almost all parts of
University of Liège, Belgium. The sorption analyzer
Africa and South East Asia and, according to Kerje
of water vapor (DVS) used is the last dynamic
and Grum (2000), watermelon is a very rich source of
analyzer of water sorption of the outer measurement
vitamins and often used as an appetizer or snack,
systems. The schematic of the main components of
depending on how it is prepared.
the DVS Intrinsic is shown in Figure 1, and the
June, 2024 On the water desorption of fruits: a case study of watermelon Vol. 26, No.2 241

specifications of this device are presented in Table 1. the DVS introduces another parameter – the time-out.
The cantilever of the microbalance is This parameter sets the maximum time spent at each
continuously held in horizontal position by an applied RH level before moving up or down.
moment. The electrical current used to uphold this
moment is therefore an indirect measure of the mass
loaded in the sample holder. The equipment is
capable of keeping a stable temperature in the range
20°C-40°C. The RH of the air surrounding the loaded
sample in the sample holder is controlled by mixing
dry and water saturated air to a mixed air with a
specified RH. Thus, in the range 0-98% RH all levels
of RH are possible. This is an advantage over the
traditional method with SSS, where the levels of RH
were limited by the different available salt solutions
(Engelund et al., 2010).
This type of DVS Advantage is capable of
handling up to 4 g sample mass; however, for a
highly hygroscopic material like fruits, a sample mass
of around 20-50 mg is enough. The smaller the
sample mass, the faster the data acquisition. However,
in order to secure accuracy of the obtained results,
Figure 1 Schematic of the main components of the DVS
sample masses below 1 g should be avoided. Using
Intrinsic (Surface Measurement Systems Company, 2023)
Table 1 DVS Intrinsic specifications (Surface Measurement Systems Company, 2023)
Specifications Values
Temperature range (°C) 20 - 40
Temperature stability (°C) +/-0.1
Maximum mass (low/high mass instrument) (g) 1 /4
Mass change (low/high mass instrument) (g) 0.150 - 1.0
Stability (24 h at 25 ºC and 0% RH) (low/high mass instrument) (µg) < 5 - < 20
Mass resolution (low/high mass instrument) (µg) +/- 0.1 /1.0
Humidity range 0 - 98%
Relative humidity accuracy +/-1%
Temperature stability (°C) +/- 0.1
Sample chamber (largeur ×profondeur × hauteur) (mm) 40 × 50 × 50
Reservoir volume (mL) 100 reservoir capacity
Mass (kg) 22
Size approximately (wide × deep ×high) (cm) 26 × 39 × 47
Air supply (bar) 3 to 4

2.3 Experimental procedure 0.001%-0.002% /min and based on previous findings.

The experiments are performed at 20°C, 30°C and For measuring the desorption isotherms with the DVS
40°C. In order for the software to automatically technique, samples of 2.5 mm thick are cut out from
change the surrounding RH when moisture inside the blocks for each temperature. They are
equilibrium for the sample is obtained, a dm/dt weighing from approx. 33.49 mg to approx. 36.23 mg,
stability criterion must be met. The dm/dt stability and subjected to desorption from water saturation.
criterion selected for the experiments is in the range The desorption experiments are run starting at 90%
June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.2 242

RH and ending at 0% RH in steps of 10% RH. The diagnosis in the same way. Some of them exhibit
same procedure was followed by Engelund et al. better diagnostic accuracy than others. In addition, the
(2010). use of many mathematical water desorption models is
2.4 Modeling and statistical analysis of the water common, and comparing them allows to select the
desorption curve of watermelon one that provides more insightful data. Choosing the
Eleven theoretical models are tested in order to most suitable model from a set of candidates can be
select the best model for describing the equation of guided by the model selection criteria, such as
the water desorption curve of watermelon. The Reduced Chi-Squared, which measures the goodness
experimental data are adjusted to the models and, to of fit of a regression model, is used to measure the
validate the goodness of the fit, six statistical criteria, goodness of fit of the trained regression model on
namely correlation coefficient (r), coefficient of validation data sets. As the value of the calculated
2
determination (R ) to estimate regression reliability, Reduced Chi-squared decreases, the observed data
2
standard error (SE), reduced chi-square (χ ) to check (experimental values) will be more similar to the
model compatibility test, mean bias error (MBE) and expected data (predicted values by the model). The
root mean square error (RMSE) are calculated using Experimental and predicted values are identical if the
Curve Expert Professional software (2.7.3) and MS Reduced Chi-Square value is null. However, the
Excel computer program. The relationship between experimental values differ from the predicted ones
the constants of the chosen model and the when the Reduced Chi-squared is higher.
temperature is also determined and the effect of Consequently, the model with the smallest values of
temperature on the constants of the equation is the Reduced Chi-Squared, is the most suitable.
examined. Models used in this investigation are reported in
In other words, if we ignore the mathematical Table 2. They express the equilibrium moisture
aspects of desorption curves, the study of the drying content (Xeq, in kg H20/kg dry matter) of the fruit
process can be significantly affected. Researchers in according to its water activity (aw). Xeq represents the
this context often repeat the use of mathematical moisture level at which a dried fruit reaches a balance
models. However, models do not present the with its surrounding environment.
Table 2 Mathematical models applied to the water desorption isotherms
Model Equation Reference
Langmuir Xeq = A aw / (1+B aw) (Langmuir, 1918) (1)
BET Xeq = Xm×C×aw / [(1-aw)+(C–1) (1– aw).aw] (Brunauer et al.,1938) (2)
Oswin Xeq = A×[aw / (1-aw)]B (Oswin, 1946) (3)
Smith Xeq = A – B ln(1-aw) (Smith, 1947) (4)
Halsey Xeq = [–exp(A)/ln(aw)](1/B) (Halsey, 1948) (5)
Henderson Xeq = [–ln(1– aw)/A](1/B) (Henderson, 1952) (6)
Kuhn Xeq = [C/ln(aw)]+k (Kühn, 1964) (7)
Chung-Pfost Xeq = 1/B×[ln(A) – ln(–ln(aw))] (Chung and Pfost, 1967) (8)
Iglessias & Chirifie Xeq = A+B×[aw / (1– aw)] (Iglesias and Chirife, 1978) (9)
GAB Xeq = Xm×C×K×aw / [(1–k×aw)×(1+(C–1) K aw)] (Van den Berg, 1984) (10)
Lewicki Xeq = A/(1– aw)B – B/(1– aw)C (Lewicki, 2000) (11)
The correlation coefficient (r) is calculated from from Equation 13 (Sacilik and Elicin, 2006) :
Equation 12 (Amini et al., 2021): 
N
(Oi  Ti ) 2
X N2 n  i 1
(13)
N n
 i1[oi  Ti ]2
N

r  1 (12)
 The mean bias error (MBE) is calculated from
N
i 1
[oi  Tm ]2
Equation 14 (Ukhurebor et al., 2020) :
The reduced chi-square (χ2) is calculated with the
1

N
degree of freedom which equals the number of MBE  i 1
(Oi  Ti ) (14)
N
observations N minus the number of fitted parameters The root mean square error (RMSE) is calculated
June, 2024 On the water desorption of fruits: a case study of watermelon Vol. 26, No.2 243

from Equation 15 (Sacilik and Elicin, 2006) : Equation 20 (Basu et al., 2006; Chung and Pfost,
1
1/2
 1967) is obtained:

N
RMSE   i 1
(Oi  Ti ) 2  (15)
N  qst 1
ln(aW )    k (20)
th
Where: Oi is the i actual value (experimental), Ti R T

is the ith predicted value, N is the number of data, n is Where: k is a constant and (-qst/R) is the slope of
the number of constants and Tm is given by Equation the curve ln (aw) = f (1/T) with constant water content
16 (Amini et al., 2021): from which the net desorption isosteric heat can be
calculated.

N
Oi
Tm  i 1
(16)
N
3 Results and discussion
2.5 Determination of the desorption isosteric heat
In order to evaluate the water binding force to 3.1 DVS isotherm analysis

watermelon and in order to determine the required The schema of desorption measurement using the

energy removing the water from the fruit during DVS is shown in Figure 2(a-c) and is prepared on the

desorption, the calculation of the isosteric heat (Qst) basis of the user's manual provided by the

in kJ mol-1, is essential (Koocheki et al., 2007). At manufacturer. Then the basic results obtained from

each water content, Qst can be calculated from the desorption isotherm experiment are shown. The

Equation 17 (Chung and Pfost, 1967; Van den Berg, mass as function of time is illustrated along with the

1984). RH surrounding the fruit sample (Engelund et al.,

Qst  qst  Lv (17) 2010). The graphs showing the courses of DVS,
registered directly during the measurements, are
Where : qst is the net isostetric heat of desorption
presented. In Figure 2(a–c), it is shown how the
in kJ mol-1 and Lv is the latent heat of vaporization of
courses of the researched desorption processes look
pure water in kJ mol-1, calculated from Equation 18
like for the selected samples for 20°C, 30°C and 40°C.
(Iglesias and Chirife, 1978) :
Due to the small sample amount used, i.e. only 35
Lv  R  (6887  5.31T ) (18)
mg dry mass, the entire experiment is completed
Where : R is the universal ideal gas constant =
within 2.5 days. The DVS isotherm analysis reports
8.315 J mol-1 K-1 and T is the average desorption
for three temperatures (20°C, 30°C and 40°C) are
temperature in K.
shown in Figure 2(a), and the mass of watermelon
The net isosteric heats (qst) with constant
sample and RH (dashed line) in the DVS equipment
equilibrium water contents are calculated from the
as function of time are shown in Figure 2(b). The
isosteric desorption curves. These curves are derived
change in mass over time can also be illustrated as the
from the desorption isotherms previously found and
dm/dt over time in Figure 2(c) for three temperatures.
illustrate the relationship between the water activity
As can be seen, the software automatically moves to
of the fruit and the constant water content
the next RH step upon meeting the dm/dt stability
temperature. To do this, the Clausius-Clapeyron
criterion set by the operator. From Figure 2(b) and
equation applied to the desorption curves determined
Figure 2(c) it is clear that the chosen criterion is
at three temperatures is used as follows (Basu et al.,
satisfactory in securing a sample mass sufficiently
2006; Chung and Pfost, 1967; Van den Berg, 1984) :
close to the true equilibrium mass at the different RH
 ln(aW ) qst
 (19) levels. Thus, in the following the sample mass at the
t RT 2

By integrating Equation 19, assuming that the end of each RH step is referred to as the mass in
water content is constant and the net desorption equilibrium with the given climate conditions. This
isosteric heat (qst) is independent of temperature, mass can be used for calculating the moisture content
of the sample on the basis of the acquired dry mass of
June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.2 244

the sample. This is found after completing the first standard settings and automatic humidity changes at
step of the test run at 0% RH (Figure 2(c)) (Engelund lower RH levels, mass equilibrium states have been
et al., 2010). obtained. However, for the highest humidities, it
The water desorption curves of the watermelon at would be advisable to have typical DVS settings with
20°C, 30°C and 40°C, representing the equilibrium a longer measurement time to ensure stabilization of
water content variation vs the water activity (RH of the moisture equilibrium state in the fruit sample. The
the air in equilibrium) are obtained via the DVS after same findings were made by Garbalińska et al. (2017).
an average of 4.5 days. According to the analysis, at

DVS Isotherm Analysis Report Sample: Watermelon_T20

University of Liège, Belgium
Watermelon_T20 Temp: 20.0 °C
Meth: Réda Khama H90-0 T20.sao
MRef: 20.3234
dm/dt (%/min) Actual RH Target RH

0.02 100
Meth: C:\DVS-Intrinsic 1\method\Réda Khama H90-0 T20.sao
Sample: Watermelon_T20
90
Temp: 20.0 °C
0.01
MRef: 20.3234 from Mass at end of first 0% RH stage
80
Target Change In Mass (%) - ref
RH (%) Sorption Desorption Hysteresis 0 70
Cycle 1 0.0 0.00 0 1000 2000 3000 4000 5000 6000 7000 8000
10.0 0.42
20.0 0.95 60
dm/dt (%/min)

30.0 1.69 -0.01

Axis Title
40.0 2.96
50
50.0 5.43
60.0 10.25 -0.02
70.0 19.42 40
80.0 38.43
90.0 70.13
-0.03 30

20
-0.04
10

DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008 -0.05 0
Time/mins
DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008

Sample: Watermelon_T20 DVS Isotherm Plot Temp: 20.0 °C

University of Liège, Belgium Meth: Réda Khama H90-0 T20.sao
MRef: 20.3234
Cycle 1 Desorp

80

70

60
Change In Mass (%) - Ref

50

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Target RH (%)
DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008

(a) DVS isotherm analysis report at 20℃

DVS Isotherm Analysis Report Sample: Watermelon_T30

University of Liège, Belgium
Watermelon_T30 Temp: 30.0 °C
Meth: Réda Khama H90-0 T30 2.sao
MRef: 19.2386
dm/dt (%/min)

0.02 100
Meth: C:\DVS-Intrinsic 1\method\Réda Khama H90-0 T30 2.sao
Sample: Watermelon_T30
90
Temp: 30.0 °C 0.015
MRef: 19.2386 from Mass at end of first 0% RH stage
80
Target Change In Mass (%) - ref 0.01 Actual RH
RH (%) Sorption Desorption Hysteresis
70
Cycle 1 0.0 0.00
10.0 0.34
0.005
20.0 0.85 60
dm/dt (%/min)

Target RH (%)

30.0 1.75
40.0 3.69
0 50
50.0 7.47
0 1000 2000Target RH 3000 4000 5000 6000 7000
60.0 13.34
70.0 22.54 40
-0.005
80.0 39.35
90.0 83.68
30
-0.01
20

-0.015
10

DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008 -0.02 0
Time/mins
DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008
June, 2024 On the water desorption of fruits: a case study of watermelon Vol. 26, No.2 245

Sample: Watermelon_T30 DVS Isotherm Plot Temp: 30.0 °C

University of Liège, Belgium Meth: Réda Khama H90-0 T30 2.sao
MRef: 19.2386
Cycle 1 Desorp

90

80

70

Change In Mass (%) - Ref

60

50

40

30

20

10

0
0 10 20 30 40 50 60 70 80 90 100
Target RH (%)
DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008

(b) Mass of watermelon sample and relative humidity (dashed line) in the DVS equipment as function of time

DVS Isotherm Analysis Report Sample: Watermelon_T40

University of Liège, Belgium
Watermelon_T40 Temp: 40.0 °C
Meth: Réda Khama H90-0 T40 2.sao
MRef: 23.247
dm/dt (%/min)

0.02 100
Meth: C:\DVS-Intrinsic 1\method\Réda Khama H90-0 T40 2.sao
Sample: Watermelon_T40
90
Temp: 40.0 °C
MRef: 23.247 from Mass at end of first 0% RH stage 0
0 1000 2000 3000 4000 5000 6000 7000 800080
Target Change In Mass (%) - ref Actual RH
RH (%) Sorption Desorption Hysteresis
70
Cycle 1 0.0 0.00 -0.02
10.0 0.45
20.0 1.19 60
dm/dt (%/min)

Target RH (%)
30.0 2.55
40.0 5.12
-0.04 50
50.0 9.43
Target RH
60.0 15.74
70.0 25.49 40
80.0 46.77
90.0 97.95 -0.06
30

20
-0.08

10

DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008 -0.1 0
Time/mins
DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008

Sample: Watermelon_T40 DVS Isotherm Plot Temp: 40.0 °C

University of Liège, Belgium Meth: Réda Khama H90-0 T40 2.sao
MRef: 23.247
Cycle 1 Desorp

120

100
Change In Mass (%) - Ref

80

60

40

20

0
0 10 20 30 40 50 60 70 80 90 100
Target RH (%)
DVS - The Sorption Solution © Surface Measurement Systems Ltd UK 1996-2008

c) Mass change over time, dm/dt along with the relative humidity illustrated as function of time for the watermelon sample
Figure 2 Graphs showing the courses of DVS measurements for samples of watermelon at 20°C, 30°C and 40°C
3.2 Influence of relative humidity and effect of Figure 3 shows the three desorption isotherms of
temperature on the equilibrium water content of watermelon at the temperature range of study. Since
watermelon water and sugars are the main components of the
June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.2 246

fruits (Basu et al., 2006), all the curves determined The effect of the temperature becomes perceptible
are of type III, according to the classification of from water activity values superior to 0.4.
Brunauer et al. Reported by (Basu et al., 2006; In the range 0-0.4 of water activity, the
Lewicki, 2000). Contrary to popular belief, glucose equilibrium water content decreases at constant RH if
and fructose are found in larger quantities in tomatoes, temperature increases.This corresponds to the
according to the USDA (USDA, 2015) nutrition increase in the kinetic energy of the water molecules
laboratory. As for watermelon, the amount of sugars of the fruit, which leads to the reduction of attraction
is even greater (USDA, 2015). forces (Uribe et al., 2011). According to Deshmukh et
The equilibrium moisture content varies from al. (2017), this tendency can be explained by the
-5 -5
1.20×10 to 0.46, from 1.14×10 to 0.39 and from consideration of the stirring states of the molecules.
1.67×10-5 to 0.35 kg water.kg dry matter
-1
for the Similar results were obtained by Basu et al. (2006),
temperatures of 20°C, 30°C and 40°C, respectively. and Uribe et al. (2011).

Figure 3 Isotherms desorption of watermelon at 20°C, 30°C and 40°C, determined by using the DVS technique
3.3 Modeling results temperature effect on the two parameters (A and B)
The models statistical analyzes applied to the of the selected Henderson model, the values are
water desorption data, expressing changes in examined for all the tests carried out. It appears that
equilibrium water content as a function of water the variations of A and B are negligible in the interval
activity, are compared using the correlation 20°C-40°C and therefore each parameter is taken as a
coefficient, the coefficient of determination, the constant. Indeed, the proposed values for A and B are
standard error, the reduced chi-squared, the mean bias 2.52 and 0.49, respectively (Table 4).
error and the root mean square error; they are The validation of the Henderson model is
predicted by a non-linear regression analysis using performed by comparing the calculated water
the Curve Expert software and MS Excel computer contents (predicted values) with experimentally
program. They show that the Henderson model is the determined values. This comparison is illustrated in
most appropriate model for the water desorption. The Figure 4. The model predictions and the desorption
results are then summarized in Table 3. Halsey, Kuhn data around a line show the relevance of the selected
and Chung-Pfost models have failed catastrophically, model in the description of the water desorption
and are not available. To take into account the behavior of the fruit.
June, 2024 On the water desorption of fruits: a case study of watermelon Vol. 26, No.2 247

Table 3 Statistics of mathematical models applied to the water desorption isotherms of watermelon
Models T (°C) R2 SE χ2 MBE RMSE
20 0.9964 0.0099 0.00009 -0.0037 0.0087
Langmuir 30 0.9919 0.0127 0.00016 -0.0045 0.0112
40 0.9923 0.0110 0.00012 -0.0033 0.0097
20 0.9985 0.0062 0.00004 -0.0019 0.0055
BET 30 0.9958 0.0091 0.00008 -0.0026 0.0081
40 0.9971 0.0067 0.00004 -0.0015 0.0060
20 0.9986 0.0061 0.00004 0.0017 0.0054
Oswin 30 0.9957 0.0092 0.00009 0.0023 0.0082
40 0.9971 0.0067 0.00005 0.0011 0.0060
20 0.9199 0.0471 0.00222 0.0336 0.0416
Smith 30 0.9194 0.0404 0.00163 0.0142 0.0357
40 0.8924 0.0416 0.00173 0.0040 0.0367
20 0.9993 0.0043 0.00001 0.0008 0.0038
Anderson 30 0.9997 0.0026 0.00001 -0.0003 0.0023
40 0.9993 0.0034 0.00001 0.0008 0.0030
20 0.9961 0.0102 0.00011 -0.00004 0.0091
Iglessias and Chirifie 30 0.9944 0.0106 0.00011 0.00005 0.0094
40 0.9881 0.0138 0.00020 -0.00001 0.0122
20 0.9991 0.0057 0.00002 -0.0014 0.0043
GAB 30 0.9973 0.0085 0.00005 -0.0025 0.0064
40 0.9979 0.0067 0.00003 -0.0013 0.0050
20 0.9989 0.0057 0.00003 -0.0002 0.0047
Lewicki 30 0.9968 0.0086 0.00006 -0.0003 0.0070
40 0.9971 0.0073 0.00005 -0.0001 0.0060

Table 4 Mathematical models parameters

Models T (°C) Parameters SE Range (95% confidence)

A = 0.0920 0.0050 0.0801 to 0.1038

20
B = -1.0543 0.0122 -1.0834 to -1.0253
A = 0.0784 0.0064 0.0630 to 0.0937
Langmuir 30
B = -1.0528 0.0188 -1.0973 to -1.0084
A = 0.0613 0.0050 0.0493 to 0.0734
40
B = -1.0785 0.0161 -1.1167 to -1.0402
Xm = 0.1428 0.0064 0.1275 to 0.1582
20
C = 0.4814 0.0530 0.3559 to 0.6069
Xm = 0.1227 0.0098 0.0993 to 0.1460
BET 30
C = 0.4655 0.0893 0.2543 to 0.6768
Xm = 0.1290 0.0115 0.1018 to 0.1563
40
C = 0.3121 0.0553 0.1812 to 0.4429
A = 0.0932 0.0029 0.0863 to 0.1001
20
B = 1.1681 0.0256 1.1074 to 1.2289
A = 0.0781 0.0043 0.0677 to 0.0885
Oswin 30
B = 1.1776 0.0459 1.0689 to 1.2863
A = 0.0614 0.0031 0.0539 to 0.0689
40
B = 1.2731 0.0414 1.1751 to 1.3711
A = -0.0553 0.0249 -0.1144 to 0.0036
20
B = 0.2786 0.0310 0.2051 to 0.3521
A = -0.0495 0.0214 -0,1001 to 0,0011
Smith 30
B = 0.2383 0.0266 0.1752 to 0.3013
A = -0.0455 0.0220 -0.0977 to 0.0065
40
B = 0.2092 0.0274 0.1443 to 0.2741
A = 2.3862 0.0227 2.3325 to 2.4399
20
B = 0.5135 0.0083 0.4937 to 0.5333
A = 2.5825 0.0197 2.5359 to 2.629129
Anderson 30
B = 0.5089 0.0059 0.4948 to 0.522985
A = 2.6004 0.0284 2.5331 to 2.6676
40
B = 0.4636 0.0079 0.4447 to 0.4825
A = -0.0176 0.0046 -0.0286 to -0.0065
20
Iglessias and Chirifie B = 0.1191 0.0027 0.1125 to 0.1257
June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.2 248

A = -0.0171 0.0048 -0.0286 to -0.0056

30
B = 0.1018 0.0028 0.0949 to 0.1086
A = -0.0183 0.0062 -0.0332 to -0.0034
40
B = 0.0904 0.0037 0.0815 to 0.0993
Xm= 0.2352 314096.8928 -807411.5325 to 807412.0030
C = 0.2341 239873.0900 -616613.1742 to 616613.6425
20
K = 1.1056 343464.1160 -882901.5126 to 882903.7237
k = 0.8467 10118.0574 -26008.4478 to 26010.1413
Xm = 0.1893 455410.8049 -1170670,554 to 1170670,932
C = 0.2435 441459.0074 -1134806,262 to 1134806,749
GAB 30
K = 1.1193 663470.6788 -1705504.556 to 1705506.795
k = 0.8467 8921.3035 -22932,0942 to 22933,7876
Xm = 1.0166 2772066.2358 -7125822.099 to 7125824.132
C = 0.0404 105744.5598 -271825.004 to 271825.085
40
K = 0.9229 102901.5448 -264515.919 to 264517.765
k = 0.8855 2097.21930 -5390.188 to 5391.959
A = 0.8044 0.0433 0.6983 to 0.9106
20 B = 0.8120 0.0408 0.7122 to 0.9118
C = 0.6995 0.0563 0.5615 to 0.8374
A = 0.7697 0.0731 0.5907 to 0.9486
Lewicki 30 B = 0.7794 0.0692 0.6099 to 0.9488
C = 0.6726 0.0942 0.4420 to 0.9031
A = 0.9216 0.0857 0.7117 to 1.1315
40 B = 0.9272 0.0824 0.7253 to 1.1290
C = 0.8667 0.0979 0.6270 to 1.1064
June, 2024 On the water desorption of fruits: a case study of watermelon Vol. 26, No.2 249

Figure 4 Experimental and predicted (Henderson model) desorption isotherms of watermelon

3.4 Determination of the desorption isosteric heat because the lines slopes are always negative
of watermelon (Equation 20). Similar results have been reported by
By plotting the curves ln (aw) = f (1/T), the affine other authors (Basu et al., 2006; Koocheki et al., 2007;
lines obtained in Figure 5 are the desorption isosteres Nur Farah Hani et al., 2014; Kiranoudis et al., 1993).
of watermelon. This figure already shows that the The isosteric equations obtained and used for the
isosteres slopes become weaker as the equilibrium determination of the net isosteric heats (qst) are
water contents decrease. In other words, the net summarized in Table 5. The determination
isosteric heats of desorption are inversely coefficients of all elaborated regressions are close to
proportional to the equilibrium water contents 1, thus showing an excellent correlation.

Figure 5 Desorption isosteres of watermelon

Table 5 Isosteric equations of watermelon
Xeq, kg H20 /kg dry matter Equation k R2
0.45 y = - 0.0022 x + 0.5297 0.5297 0.9686
0.40 y = - 0.0025 x + 0.5874 0.5874 0.9777
0.35 y = - 0.0028 x + 0.6556 0.6556 0.9855
0.30 y = - 0.0031 x + 0.7374 0.7374 0.9920
0.25 y = - 0.0035 x + 0.8379 0.8379 0.9968
0.20 y = - 0.0041 x + 0.9653 0.9653 0.9996
0.15 y = - 0.0049 x + 1.1346 1.1346 0.9996
0.10 y = - 0.0060 x + 1.3784 1.3784 0.9954
0.05 y = - 0.0080 x + 1.7940 1.7940 0.9831
June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.2 250

The isosteric heats (Qst) of the watermelon are The difference between the isosteric (Qst) and the
calculated using Equation 18 and plotted as a function net isosteric (qst) heats of desorption remains constant
of the equilibrium water content in Figure 6. In whatever the value of the water content, the
addition to the variation of (Qst) as a function of (Xeq), difference is the latent heat of vaporization. The
this figure also shows the variation of (qst) as a difference between the isosteric heat (Qst) and the
function of the same explanatory variable and gives latent heat of vaporization (Lv) decreases as the
the possibility of comparing the two variables moisture content increases to zero at high water
explained with the latent heat of vaporization of the content values. Therefore, the value of the isosteric
pure water (Lv). Figure 6 shows that the isosteric heat heat tends towards the latent heat value. This is due to
is more important for low moistures and, for high the decrease in the net desorption isosteric heat which
humidities, it becomes negligible in front of latent tends to zero at the high water content values. In
heat of evaporation. Going from 0.05 to 0.45 kg order to express the isosteric heat (Qst) as a function
-1
water.kg dry matter of equilibrium moisture content, the of the equilibrium water content, the following
values of isosteric heat vary from 65.02 to 113.25 kJ equation is proposed:
mol-1. Qst  110 exp ( 1.29 X eq ) (21)

Figure 6 Temperature effect on the isosteric heat of watermelon

4 Conclusion equilibrium moisture content average of watermelon
varies from 1.20×10-5 to 0.46, from 1.14×10-5 to 0.39
The study of the water desorption of watermelon
and from 1.67×10-5 to 0.35 kg water.kg dry matter
-1
at the
by a gravimetric method using DVS has shown the
temperatures of 20°C, 30°C and 40°C, respectively.
success of such a method for the experimental
At constant temperature, the equilibrium moisture
determination of the equilibrium water contents of
content increases with water activity and there is no
fruits. Contrary to the long durations taken for the
influence of the temperature on desorption in the
determination of the water desorption isotherms of
interval 0-0.4 of water activity.
fruits by the static method, four and a half days are
At constant RH, the equilibrium moisture content
largely sufficient to reach equilibrium via DVS. The
is inversely proportional to the temperature in the
main results can be summarized as follows:
interval 0-0.4 of water activity.
The results show that the Henderson model is the
The net isosteric heat of desorption is inversely
most appropriate model to represent the water
proportional to the equilibrium moisture content and
desorption of watermelon fruit and the obtained
becomes negligible in front of the evaporation latent
curves are of the type III.
heat for the high humidities. In the interval 0.05 -
In the range 0-0.8 of the water activity, the -1
0.45 kg water.kg dry matter of equilibrium moisture, the
June, 2024 On the water desorption of fruits: a case study of watermelon Vol. 26, No.2 251

isosteric heat values vary from 65.02 to 113.25 kJ Djendoubi Mrad, N., C. Bonazzi, N. Boudhrioua, N. Kechaou,
-1
mol . and F. Courtois. 2012. Influence of sugar composition
on water sorption isotherms and on glass transition in
apricots. Journal of Food Engineering, 111(2): 403-411.
References Djendoubi Mrad, N., C. Bonazzi, F. Courtois, N. Kechaou, and
Amini, G., F. Salehi, and M. Rasouli. 2021. Drying kinetics of N. Boudhrioua Mihoubi. 2013. Moisture desorption
basil seed mucilage in an infrared dryer: Application of isotherms and glass transition temperatures of osmo-
GA-ANN and ANFIS for the prediction of drying time dehydrated apple and pear. Food and Bioproducts
and moisture ratio. Journal of Food Processing and Processing, 91(2): 121-128.
Preservation, 45(3): e15258. Engelund, E. T., M. Klamer, and T. M. Venås. 2010
Anderberg, A., and L. Wadsö. 2008. Method for simultaneous Acquisition of sorption isotherms for modified woods
determination of sorption isotherms and diffusivity of by the use of dynamic vapour sorption instrumentation.
cement-based materials. Cement and Concrete Research, Principles and Practice. In the International Reasearch
38(1): 89-94. Group on Wood Protection, 41st Annual Meeting,
Argyropoulos, D., R. Alex, and J. Müller. 2011. Equilibrium IRG/WP 10-40518. Biarritz, France, 9-13 May.
moisture contents of a medicinal herb (Melissa Garbalińska, H., M. Bochenek, W. Malorny, and J. van
officinalis) and a medicinal mushroom (Lentinula Werder. 2017. Comparative analysis of the dynamic
edodes) determined by dynamic vapour sorption. vapor sorption (DVS) technique and the traditional
Procedia Food Science, 1: 165-172. method for sorption isotherms determination -
Arlabosse, P., E. Rodier, J. H. Ferrasse, S. Chavez, and D. Exemplified at autoclaved aerated concrete samples of
Lecomte. 2003. Comparison between static and four density classes. Cement and Concrete Research, 91:
dynamic methods for sorption isotherm measurements. 97-105.
Drying Technology, 21(3): 479-497. Goula, A. M., T. D. Karapantsios, D. S. Achilias, and K. G.
Basu, S., U. S. Shivhare, and A. S. Mujumdar. 2006. Models Adamopoulos. 2008. Water sorption isotherms and
for sorption isotherms for foods: A review. Drying glass transition temperature of spray dried tomato pulp.
Technology, 24(8): 917-930. Journal of Food Engineering, 85(1): 73-83.
Bingol, G., B. Prakash, and Z. Pan. 2012. Dynamic vapor Halsey, G. 1948. Physical adsorption on non-uniform surfaces.
sorption isotherms of medium grain rice varieties. LWT The Journal of Chemical Physics, 16(10): 931-937.
- Food Science and Technology, 48(2): 156-163. Haoua, A., A. Leonard, and B. Bouchekima. 2015.
Brunauer, S., P. H. Emmett, and E. Teller. 1938. Adsorption of Determination of desorption isotherms using the DVS
gases in multi molecular layers. Journal of the apparatus (dynamic vapor sorption system) at different
American Chemical Society, 60(2): 309-319. temperatures on unblanched and blanched samples of
Caballero-Cerón, C., J. A. Guerrero-Beltrán, H. Mújica-Paz, J. potato. Journal of Agroalimentary Processes and
A. Torres, and J. Welti-Chanes. 2015. Moisture Technologies, 21(2): 142-147.
Sorption Isotherms of Foods: Experimental Henderson, S. M. 1952. A basic concept of equilibrium
Methodology, Mathematical Analysis, and Practical moisture. Agricultural Engineering, 33: 29-32.
Applications. In Water Stress in Biological, Chemical, Iglesias, H. A., and J. Chirife. 1978. An empirical equation for
Pharmaceutical and Food Systems, eds. Gutiérrez- fitting water sorption isotherms of fruits and related
López, G., L. Alamilla-Beltrán, M. del Pilar Buera, J. products. Canadian Institute of Food Science and
Welti-Chanes, E. Parada-Arias, and G. Barbosa- Technology Journal, 11(1): 12-15.
Cánovas, ch. 15, 187-214. New York, NY, USA: Kerje, T., and M. Grum. 2000. The origin of melon, Cucumis
Springer. melo: A review of the literature. Acta Horticulturae,
Chung, D. S., and H. B. Pfost. 1967. Adsorption and 510: 37-44.
desorption of water vapour by cereal grains and their Khama, R. 2016. Conception, réalisation et simulation d’un
products. Part I. Heat and free energy changes of séchoir solaire opérant en mode convectif utilisépour le
adsorption and desorption. Transactions of the ASAE, séchage de produits agroalimentaires. Ph.D. diss.,
10(4): 549-551. Département de Génie des procédés. Université
Deshmukh, G., P. Birwal, S. Patel, R. Datir, and A. Abderrahmane Mira Béjaia (UAMB), Algérie.
Surywanshi. 2017. Sorption isotherms of traditional Kiranoudis, C. T., Z. B. Maroulis, E. Tsami, and D. Marinos-
Indian dairy products: A review. Journal of Food, Kouris. 1993. Equilibrium moisture content and heat of
Nutrition and Population Health, 1(3): 25. desorption of some vegetables. Journal of Food
June, 2024 AgricEngInt: CIGR Journal Open access at http://www.cigrjournal.org Vol. 26, No.2 252

Engineering, 20(1): 55-74. Smith, S. E. 1947. The sorption of water vapour by high
Koocheki, A., S. M. A. Razavi, E. Milani, T. M. Moghadam, polymers. Journal of the American Chemical Society,
M. Abedini, S. Alamatiyan, and S. Izadkhah. 2007. 69(3): 646-651.
Physical properties of watermelon seed as a function of Surface Measurement Systems Company. 2023. Advanced
moisture content and variety. International Agrophysics, Sorption Analyzers: DVS Intrinsic PLUS Compact
21(4): 349-359. DVS instrument for complex water sorption challenges.
Kühn, I. 1964. A new theoretical analysis of adsorption Available at:
phenomena: Introductory part: The characteristic https://surfacemeasurementsystems.com/product/dvs-
expression of the main regular types of adsorption intrinsic-plus/. Accessed on 19-06-2023 and 19-11-2023.
isotherms by a simple equation. Journal of Colloid Ukhurebor, K. E., S. O. Azi, U. O. Aigbe, R. B. Onyancha,
Science, 19(8): 685-698. and J. O. Emegha. 2020. Analyzing the uncertainties
Langmuir, I. 1918. The adsorption of gases on plane surfaces between reanalysis meteorological data and ground
of glass, mica and platinum. Journal of the American measured meteorological data. Measurement, 165:
Chemical Society, 40(9): 1361-1403. 108110.
Lewicki, P. P. 2000. Raoult’s law-based food water sorption United State Department of Agriculture (USDA). 2015. Food
isotherm. Journal of Food Engineering, 43(1): 31-40. and Nutrition. Available at: https://www.usda.gov.
Maoto, M. M., D. Beswa, and A. I. O. Jideani. 2019. Accessed on 19-09-2015 and 05-03-2021.
Watermelon as a potential fruit snack. International Uribe, E., A. Vega-Gálvez, K. Di Scala, R. Oyanadel, J. S.
Journal of Food Properties, 22(1): 355-370. Torrico, and M. Miranda. 2011. Characteristics of
Nur Farah Hani, M., W. Z. Wan Nur Zahidah, K. Saniah, and Convective Drying of Pepino Fruit (Solanum muricatum
H. S. M. Irwani. 2014. Effects of drying on the physical Ait.): Application of Weibull Distribution. Food and
characteristics of dehydrated watermelon rind candies. Bioprocess Technology, 4: 1349-1356.
Journal of Tropical Agriculture and Food Science, Van den Berg, C. 1984. Desorption of water activity of foods
42(2): 115-123. for engineering purposes by means of the GAB model
Oswin, C. R. 1946. The kinetics of package life. III. Isotherm. of sorption. In Engineering and Foods (Vol. 1), ed. B.
Journal of the Society of Chemical Industry, 65(12): M. Mckenna, 311-321. London, U.K.: Elsevier.
419-421. Yang, R., J. Guan, S. Sun, S. S. Sablani, and J. Tang. 2020.
Sacilik, K., and A. K. Elicin. 2006. The thin layer drying Understanding water activity change in oil with
characteristics of organic apple slices. Journal of Food temperature. Current Research in Food Science, 3: 158-
Engineering, 73(3): 281-289. 165.