Received: 27 April 2020 | Revised: 6 July 2020 | Accepted: 7 July 2020

DOI: 10.1111/jfpp.14782

ORIGINAL ARTICLE

Optimization of process variables on physicochemical

properties of milk during an innovative refractance window
concentration

Asaad R. Al-Hilphy1 | Haider I. Ali1 | Sajedah A. Al-IEssa1 | José M. Lorenzo2 |

Francisco J. Barba3 | Mohsen Gavahian4

1
Department of Food Science, College of
Agriculture, University of Basrah, Basrah, Iraq Abstract
2
Centro Tecnológico de la Carne de Galicia, An innovative refractance window (RW) concentrator was developed and used to
Ourense, Spain
concentrate milk samples at temperatures of 50–70°C and pressures of 0.4–0.8 bar.
3
Nutrition and Food Science Area,
Preventive Medicine and Public Health,
Optimum process conditions were found through response surface methodology to
Food Science, Toxicology and Forensic compare RW at optimal conditions with the conventional concentration (CC). Also,
Medicine Department, Faculty of Pharmacy,
Universitat de València, Burjassot, Spain
the effects of process parameters and their interactions on the RW process and
4
Department of Food Science, National product specifications were analyzed. According to the results, optimal RW condi-
Pingtung University of Science and tions were a pressure of 0.4 bar and a temperature of 69.1°C. Besides, the energy ef-
Technology, Pingtung, Taiwan, ROC
ficiency and overall heat transfer coefficient of RW at optimal conditions were higher
Correspondence than those of CC by 125.2% and 15.3%, respectively. Compared to CC, RW reduced
Mohsen Gavahian, Department of Food
Science, National Pingtung University of the concentration-time by 60.8% and minimized the changes in the chemical compo-
Science and Technology, Pingtung 91201, sition and color of the milk. RW process was found to be a time- and energy-saving
Taiwan, ROC.
Email: [email protected] concentration technique that can produce concentrated milk with improved quality.
Practical applications
Alternative concentration techniques that can reduce energy consumption and pro-
cessing time while guaranteeing the quality retention of the product might be con-
sidered by the food industry. The innovative RW milk concentration method that was
developed in the present study showed to be a promising time- and energy- saving
concentration technique. This method was able to retain the quality characteristics
of milk during the concentration process. Also, the optimization information provided
in the present study can be used for developing upscaled units that can be used for
the commercial concentration process of milk.

1 | I NTRO D U C TI O N shelf-life or to produce several dairy products (Musina et al., 2018;

Stratakos et al., 2019). For example, the shelf-life of the milk can be
Cow milk is a nutritionally valuable product containing many of enhanced by producing condensed milk which involves evaporative
the essential macro and micronutrients including proteins, carbo- operations (Guimarães, Martins, Flauzino, Basso, & Telis Romero,
hydrates, fats, minerals, vitamins, and bioactive components (Al- 2020). Also, milk and whey protein powders can be produced
HilphyShirkole, Ali, & Mohsin, 2019; Stratakos et al., 2019). Milk through the evaporation process to be used in dairy product for-
produced in the farm is usually processed to enhance its safety and mulation (Musina et al., 2018). In either case, retaining the quality

J Food Process Preserv. 2020;44:e14782. wileyonlinelibrary.com/journal/jfpp © 2020 Wiley Periodicals LLC. | 1 of 13

https://doi.org/10.1111/jfpp.14782
2 of 13 | AL-HILPHY et al.

parameters, such as product color, is among the technical consider- concentration processes, the milk temperature was increased from 5
ations in the concentration unit operation as it can affect the accep- to 40°C using a laboratory water bath (GFL 1008, German) and then
tance of the product (Faion, Becker, Fernandes, Steffens, & Valduga, the milk was standardized by the adding milk powder (Alsabah, Iran)
2019; Fernández-Vázquez et al., 2018). and dairy cream (Barmezan, Iran) to adjust the TSS and fat content of
As the traditional concentration methods may negatively affect the the raw materials using Pearson square (Figure 1). The standardized
physicochemical properties of milk, researchers are exploring the appli- milk samples were then concentrated using either RW or conven-
cability of emerging technologies for milk concentration to produce a tional concentrator.
high-quality product (Faion et al., 2019; Moejes, vanWonderen, Bitter,
& vanBoxtel, 2020; Parmar, Singh, Meena, Borad, & Raju, 2018). The
refractance window (RW) is a novel food processing technique that is 2.2 | Milk concentration processes
believed to produce high-quality products and has been used for dry-
ing food material at both laboratory and industrial scales (Bernaert, An RW system, designed and developed at the Department of Food
VanDroogenbroeck, VanPamel, & DeRuyck, 2019; Jafari, Azizi, Mirzaei, Science of the University of Basrah, was used to concentrate milk.
& Dehnad, 2016; Raghavi, Moses, & Anandharamakrishnan, 2018). This device consists of a concentration unit made of Pyrex glass
However, there is limited information about the applicability of this tech- (with the inner diameter of 6 cm, the height of 84 cm, and a thick-
nology for the concentration of food materials such as milk. Hence, the ness of 0.2 cm), a 2 kW heating unit (which heats the water as a heat
present study aims to investigate the potential use of RW technology for transfer medium for circulation), and a control unit. A schematic rep-
milk concentration. Moreover, it focused on understanding the effects resentation of the RW system was prepared using Actrix Technical
of RW process parameters, including process temperature and pressure, 2000 Software (Autodesk Inc., USA) (Figure 2). For the RW concen-
on the performance of RW milk concentrator and chemical composition tration process, 5L of the standardized milk sample was placed in the
of the product, as well as the overall color change of the product. The RW system and heated to 95°C. Afterward, partial vacuums (0.4–0.8
suitability of the proposed data modeling approach was also verified by bars, depending on the treatment) were applied, corresponding to
comparing the predicted data with those of experimental analysis. the milk boiling point of 50–70°C. The sample was held at these con-
ditions to the time that the desired concentration, i.e., 26% TSS, was
achieved.
2 | M ATE R I A L S A N D M E TH O DS Besides, 5 L of milk samples were conventionally concentrated
(CC) to 26%TSS using an electric heater (Orbon, India) at a power
2.1 | Raw materials of 2 kW. All the concentration processes were repeated for three
times. Besides, the temperature of the milk during the process
Fresh cow milk was obtained from the Agricultural Researches was monitored using a digital temperature controller (LTR5/LEA
Station of the University of Basrah and was stored at 5°C. Before Electronic, Italy).

F I G U R E 1 The standardization process

of milk in the present study. SNF, solid
non-fat; TSS, total soluble solid
AL-HILPHY et al. | 3 of 13

F I G U R E 2 Schematic representation of the refractance window (RW) system used in the present study ((1) The milk inlet hole, (2) Milk
inlet valve, (3) Plastic cap, (4) Tank cap, (5) Milk distribution tube, (6) Water tank, (7) Hot water circulation tube, (8) Insulator, (9) Glass tube
(Pyrex), (10) Milk layer inside tube, (11) Vent valve, (12) Plastic cylinder, (13) Milk transfer tube, (14) Centrifugal pump, (15) Iron base, (16)
Tire, (17) Electric valve, (18) Manual valve, (19) Electric heater, (20) Iron centrifuge pump, (22). Manual valve, 24&23. Plunger, (25) A vacuum
pump, 27&26. Plastic tanks, (28). Washing water circulation tube, (29). The return water pipe from the heat exchanger, (30). Cold water
transfer tube to the heat exchanger, (31). Iron shaft, (32). Control panel, (33). Water circulation pump switch, (34). Washing pump operation
switch, (35) Milk Recycle Pump Operation switch, (36). Heat exchanger water circulation operation switch, (37). Vacuum pump operation
switch, (38). Milk exit electric valve operation, (39). Digital temperature gauge, (40). Heater operation switch, (41). Signal lamps, (42). Tap
for condensed water out, (43). Valve, (44). Thermocouple for temperature measurement, (45). Steam trap, (46). Valve, (47). Discharge tube,
(48). Partition, (49). Valve, (50). Vacuum pressure gauge, (51). Heat exchanger, (52). Valve, (53). Steam trap, (54). Tube, (55). Vacuum tube, (56)
Valve)

2.3 | Process evaluation Qin

SEC (kJ∕kg) =
Mw (1)

2.3.1 | Specific energy consumption and

energy efficiency Q0
𝜂= × 100 (2)
Qin
Specific energy consumption (SEC) and energy efficiency (η) were
calculated to analyses the energy consumption for RW and CC milk where, Qin, Qo and Mw are input, output energy (kJ) and mass of evap-
concentration (Equations 1-2) (Chamberland et al., 2020). orated water (kg).
4 of 13 | AL-HILPHY et al.

2.3.2 | Overall heat transfer coefficient where ΔE is overall color change, Lo ∗ , ao ∗ , andbo ∗are lightness, red-
ness-greenness, and yellowness-blueness of raw milk (before the con-
Overall heat transfer coefficient was calculated using Equation (3) centration process) and L*, a*, and b* are those of the concentrated
(Pehlivan & Özdemir, 2012). samples, respectively.

( )
MCP Th0 − Tm0 + mw 𝜆w
U= (3)
A ΔTm 2.6 | Experimental design

where, U is the overall heat transfer coefficient (W/m2°C), A is The preliminary trials were carried out to select the range of inde-
2
the area (m ), ΔTm is the logarithmic mean temperature difference pendent variables, i.e., pressure (0.4, 0.6, and 0.8 bar) and tempera-
(LMTD). M is the mass flow rate of milk (kg/s), Cp is the specific heat ture (50, 60, and 70°C). Furthermore, the central composite design
(kJ/kg°C), (Th0‒Tm0) is the temperature difference between the tem- (CCD) (3 × 3) of response surface methodology (RSM) (Design-Expert
perature of hot water (°C) and temperature of hot milk (°C) respec- Software, version 7, Stat-Ease Inc., United States) was employed to
tively,mw mass of evaporated water from milk (kg/s) and λw is the assess the effects of independent variables and their interactions on
latent heat (kJ/kg). ΔTm can be calculated according to Equation (4) the U, concentration-time (CT), SEC, η, moisture content, ash pH, fat,
as follows: protein, lactose, TSS, ∆E. In this regards, thirteen experiments were
used in CCD for each of the pressure values with five replications at
( ) ( )
Thi − Tmi − Th0 − Tm0 the center point with two axial points. The independent variables
ΔTm = (4)
ranges are presented in Table 1. The codes in the CCD were −1, 0,
[ ]
T −T
ln T hi −Tmi
h0 m0
+1 epitomized the lowest, medium, and highest values, respectively.
where, (Thi‒Tmi) is the difference between cold water temperature (°C),
and cold milk temperature (°C) respectively.
2.7 | Process optimization and data modeling

2.4 | Chemical analyses Graphical and numerical methods were applied to optimize milk
concentration conditions (independent variables), i.e., pressure
All the chemical assays in the present study, including determination (0.4–0.8 bar) and the temperature (50–70°C) of the holding phase,
of moisture content, protein, fat, pH, ash, lactose and TSS, were per- to obtain a milk concentrate with good thermal performance and
formed according to the Official Methods of Analysis (AOAC, 2016). chemical properties. The minimum values of CT, moisture content,
SEC, and ∆E as well as the maximum values of U, energy efficiency,
ash, pH, fat, protein, lactose, and TSS were targeted in the optimiza-
2.5 | Color measurement using computer vision tion process using version 7 of Design-Expert software (Stat-Ease
Inc., United States) at various conditions of milk concentration pro-
A 720p HD camera (IP67 Endoscope, Mileseey, China), and four cess (0.4-0.8 bar pressure, 50–70°C temperature). After that, func-
light-emitting diodes (LED) lamps (LB13W, Konnice Co., China) that tions of the desirability were developed for the dependent variables
are located on the top of a wooden-black color box. Two lamps have (U, CT, SEC, η, moisture content, ash, pH, fat, protein, lactose, TSS,
a 45º angle with the camera lens and two are perpendicular to the and ∆E). Besides, the optimum combinations of pressure and tem-
sample. The image processing technique was employed to analyze perature were selected via response surface plots of the dependent
the color characteristics of milk samples in triplicate similar ac- variables, and the desirability function method was used to optimize
cording to that described in the literature with some modifications the multiple responses according to Equation 6. In the present study,
(Gavahian, Farahnaky, Javidnia, & Majzoobi, 2012; Yam & Papadakis, the desirability (D) values range between 0 and 1 and the impotence
2004). In this regard, the digital images obtained and the ImageJ term (ri ) ranges between 1–5 considered as appropriate values for
software (Version 1.52q, National Institutes of Health, United the optimization purposes (Eren & Kaymak-Ertekin, 2007).
States) was utilized to obtain CIE L*a*b* values from digital images.
These parameters were used to calculate overall color changes (∆E).
� n
�1∕ ∑ ri
(6)
�1∕ ∑
r r r ri � r
d1i × d2i × d3i × ………… × drni
�
D (x) = = di i
This parameter represents the effects of concentration processes on
i=1
the sample color, that is, a higher ∆E value indicates more changes
in the milk color after the concentration process. In this regard, the where di is the desirability for each response, ri is a number refers to the
color values of the concentrated milk samples were compared to relative importance of the i response, where 5 is the greatest impor-
those of raw milk (Equation 5) (Gavahian, Sheu, Tsai, & Chu, 2020; tance and 1 is the minimum importance.
Wasnik et al., 2019). To predict the dependent variables, the quadratic polyno-
√
( ∗ ∗ )2 ( )2 ( )2 mial regression model was used as given in Equation 7 (Khuri &
ΔE = Lo − L + a∗o − a∗ + b∗o − b∗ (5) Cornell, 2019).
 AL-HILPHY et al.

TA B L E 1 Central composite design matrix for the effect of pressure and temperature on the dependent variables of the milk concentrated by the refractance window along with p-value,
R 2 and Lack of Fit of model.

RW run P (bar) T (°C) U (W/m2 .°C) CT (h) SEC (kJ/kg) η (%) MC (%) Pr. (%) Ash (%) Fat (%) Lact. (%) pH TSS (%) ∆E

h c c c c b c b c b a
1 0.4 50 70.140 ± 0.14 1.12 ± 0.01 2,752.49 ± 6.11 62.20 ± 1.09 74.00 ± 0.32 7.72 ± 0.02 1.20 ± 0.01 7.50 ± 0.12 9.58 ± 0.12 6.50 ± 0.02 26.00 ± 0.22 10.77c ± 0.12
c a a j a b e d f c c
2 0.8 50 168.33 ± 1.91 3.30 ± 0.02 6,605.00 ± 5.34 34.30 ± 0.98 76.01 ± 0.93 7.10 ± 0.02 1.10 ± 0.2 7.30 ± 0.03 8.45 ± 0.23 6.40 ± 0.01 24.00 ± 0.76 7.63d ± 0.23

3 0.8 60 217.56 ± 2.43b 3.22 ± 0.03a 6,606.01a ± 3.89 43.72h ± 1.04 76.00 a ± 0.21 7.45b ± 0.01 1.00 d ± 0.01 7.30 d ± 0.06 8.25g ± 0.42 6.51b ± 0.02 24.00 c ± 0.98 11.52b ± 0.11

4 0.8 70 252.49 ± 2.71a 3.00 ± 0.07a 6,607.03a ± 8.67 49.68e ± 2.10 75.00 b ± 0.24 7.32b ± 0.02 1.08d ± 0.03 7.40 c ± 0.07 9.20 ± 0.45 6.60 a ± 0.04 25.01b ± 0.03 9.52c ± 0.22

5 0.6 60 142.00 ± 0.98 d 2.00 ± 0.06 b 4,403.98 b ± 9.65 46.65g ± 0.89 74.02c ± 0.16 7.13b ± 0.03 1.42a ± 0.02 7.66a ± 0.05 9.79a ± 0.11 6.60 a ± 0.02 26.00 a ± 0.97 11.46 b ± 0.21
e d c a c a c d d b a
6 0.4 70 105.20 ± 1.30 1.00 ± 0.01 2,752.50 ± 5.99 71.80 ± 0.56 74.01 ± 0.52 8.12 ± 0.02 1.20 ± 0.04 7.30 ± 0.08 9.38 ± 0.23 6.54 ± 0.02 26.01 ± 0.93 12.48a ± 0.22
g c c b c b a c c b a
7 0.4 60 90.650 ± 1.60 1.10 ± 0.02 2,752.50 ± 6.98 66.88 ± 0.96 74.00 ± 0.43 7.51 ± 0.04 1.43 ± 0.03 7.40 ± 0.04 9.66 ± 0.09 6.50 ± 0.01 26.00 ± 0.44 9.97c ± 0.34

8 0.6 60 145.04 ± 2.02d 2 ± . 010.09b 4,403.99b ± 9.11 47.44f ± 1.20 74.02c ± 0.82 7.27b ± 0.03 1.38 b ± 0.04 7.50 b ± 0.07 9.85a ± 0.14 6.53b ± 0.04 26.00 a ± 0.06 11.45b ± 0.33
d b b e b c c c d c b
9 0.6 60 145.04 ± 2.11 2.05 ± 0.08 4,404.0 ± 11.10 48.85 ± 0.99 75.01 ± 0.47 6.96 ± 0.11 1.24 ± 0.05 7.40 ± 0.05 9.40 ± 0.15 6.41 ± 0.03 25.02 ± 0.04 10.50 c ± 0.22
d b b e b b d d e b b
10 0.6 60 145.04 ± 2.09 2.00 ± 0.07 4,403.60 ± 12.34 48.85 ± 1.09 75.00 ± 0.17 7.80 ± 0.14 1.00 ± 0.04 7.30 ± 0.09 8.90 ± 0.22 6.50 ± 0.01 25.00 ± 0.11 11.46b ± 0.13
c b b d b c c b d b b
11 0.6 70 168.33 ± 1.89 1.98 ± 0.02 4,403.90 ± 8.43 55.24 ± 2.03 75.00 ± 0.11 6.95 ± 0.06 1.25 ± 0.01 7.50 ± 0.10 9.30 ± 0.34 6.51 ± 0.05 25.00 ± 0.01 11.12b ± 0.43

12 0.6 50 112.22 ± 2.30 e 2.10 ± 0.01b 4,404.01b ± 5.67 39.64i ± 0.96 74.01c ± 0.0.05 7.21b ± 0.05 1.42a ± 0.03 7.68a ± 0.08 9.69c ± 0.33 6.60 a ± 0.02 26.00 a ± 0.83 11.51b ± 0.45

13 0.6 60 145.04 ± 2.55d 2.04 ± 0.02b 4,404.30 b ± 6.13 48.85e ± 3.01 75.00 b ± 1.24 7.35b ± 0.01 1.25c ± 0.02 7.40 c ± 0.04 9.00 e ± 0.13 6.63a ± 0.04 25.00 b ± 1.01 11.45b ± 0.56

p-value of <.0001 <.0001 .0001 <.0001 .0188 .3394 .4868 .2415 .1446 .8127 .0084 .009
model

R2 0.9994 0.9986 0.9787 0.9936 0.6155 0.4946 0.4138 0.5557 0.6296 0.2364 0.6525 0.9376

LoF 0.2726 0.110 0.1246 0.3701 0.6270 0.4613 0.5946 0.8415 0.6489 0.6126 0.6270 0.083

The different l letters in the columns indicate a significant effect at level of 0.05.
Abbreviations: CT, concentration-time; LoF, lack of fit; MC, moisture content; Pr, protein; P, pressure; Lact., lactose; SEC, specific energy consumption; SS, sum of square; T, temperature; TSS, total soluble
solids; U, heat transfer coeffecient; η, energy effeciency; ∆E, total color change.
|
5 of 13
6 of 13 | AL-HILPHY et al.

Abbreviations: CT, concentration-time; Lact., lactose; MC, moisture content; Pr., protein; P, pressure; SEC, specific energy consumption; SS, sum of square; T, temperature; TSS, total soluble solids; U, heat
k k k
k−1

−154.859
−225.516
∑ ∑ ∑ ∑
Y = 𝛽0 + 𝛽 i Xi + 𝛽 ii X2i + 𝛽 ij Xi Xj (7)

196.831

11.1426
−7.305

0.067
i=1 i=1 i=1 i<j=1

∆E
where Y is the responses (U, CT, SEC, η, moisture content, ash pH, fat,
protein, lactose, TSS, and ∆E), X is the independent variables, k is the

−11.666
number of factors, i and j are factors numbers, β0 is a constant, and βi,

TSS (%)

32.307

−0.075
0.125
βii and βij are coefficients of linear, quadratic, and interaction terms,

‒
‒
respectively.

1.32 × 10–4
−0.9181
−0.025
−0.020
−0.123
7.268
pH
2.8 | Statistical analysis

Statistical analysis was carried out using Design-Expert software

Lact. (%)

−0.249

1.5061
16.771

−0.118
−9.732
(version 7, Stat-Ease Inc., United States). Analysis of Variance

2.287
(ANOVA) was exploited to evaluate the differences between the
means. All components in the polynomial models were evaluated

8.34 × 10–4
−0.03750
−3.91379
statistically using p-value at .05 level. The results in the current

−0.127
10.792
Fat (%)

2.279
study were represented as means ± standard deviations and all ex-

transfer coeffecient; η, energy effeciency; ∆E, total color change. P2T and PT2 for ∆E were calculated as +2.229 and -0.114, respectively.
periments were carried out in triplicates.

Regression coefficient for response surface linear, quadratic and reduced cubic models for dependent variables

2.500 × 10–3
−8.90 × 10–3

6.03 × 10–5
−2.84914
Ash (%)
3 | R E S U LT S A N D D I S CU S S I O N

3.027
0.894
3.1 | Performance of RW concentrator

−7.37 × 10–4
−0.022
−9.669
3.1.1 | Overall heat transfer coefficient Pr. (%)

0.108

8.155
7.108
The results showed that the U values ranged between 70.140 W/m2°C

−0.12500
MC (%)

at a pressure of 0.40 bar and a temperature of 50°C to 252.49 W/m2°C

11.666
67.692

0.075
at a pressure of 0.80 bar and a temperature of 70°C (Table 1). This may

‒
‒
be due to an increase in temperature. Jebson and lyer (1991) found

−7.44 × 10–3
that the U values lie between 0.3–3.2 kW/m2°C for concentrated skim
−317.663
107.342

177.797
1.1367

milk in a 5-stage multi-effect evaporator. On the other hand, they

0.722
η (%)

mentioned that the values of U ranged between 0.8–3.08 kW/m2°C

for whole milk (Jebson & Iyer, 1991). In another study, the total heat
transfer coefficient of milk ranged from 477 to 939 W/m2°C when the
−1.14 × 10–4
SEC (kJ/kg)

scraper thin film evaporator was used (Sangrame, Bhagavathi, Thakare,

1,360.222

6,882.153
1,107.952

−0.104

Ali, & Das, 2000). Silveira et al. (2013) stated that the total heat trans-
0.249

fer coefficient for the concentration of skimmed milk under vacuum

ranged between 1.24–1.93 kWh/m2°C in the first stage and ranged
−1.62 × 10–4

between 1.25–2.00 kW/m2°C in the second stage. They emphasized

−0.022
−1.038

that the heat transfer coefficient was not significantly affected by the
CT (h)

3.486
0.023

2.594

conditions of concentration (Silveira et al., 2013).

The results of the statistical analysis in Table 1 showed that there
was a significant effect (p < .05) of the mathematical model. Also,
U (W/m2 °C)

−334.957

the Lack of Fit (LoF) was not significant (p > .05). It has been noted
230.887
−78.405

−0.045
−6.136
4.756

that R-Squared was 0.9994. These indicators showed that the mul-
tiple nonlinear correlation equation can be used to predict U values.
In addition, the regression coefficients of models are presented in
Table 2. Furthermore, Equation (8) is used to calculate U as a func-
TA B L E 2

Intercept

tion of temperature and pressure.

Source

P×T

U = −78.405 − 334.957P + 4.756T − 6.1369PT + 230.887P2 − 0.045966T2

2

T2
P

P
T

(8)
AL-HILPHY et al. | 7 of 13

Concerning the interference between pressure and temperature, RW process. The results showed that the concentration of milk by
Figure 3a, which was drawn using the response surface methodol- the RW milk concentrator required less time which reached 1 hr
ogy, the value of the overall heat transfer coefficient was 70.18 W/ when using a pressure of 0.40 bar and a temperature of 70°C. The
m2°C when using a pressure of 0.40 bar and a temperature of 50°C. longest CT was 3.3 hr at a pressure of 0.80 bar and a temperature of
it was observed that it increased to 196 W/m2°C when using a pres- 50°C. The results indicated that the effect of pressure was greater
sure of 0.80 bar and a temperature of 50°C. This is because the than the effect of temperature, as pressure was the main factor af-
increase of pressure led to a decrease in evaporation and thus the fecting the CT.
temperature remains more stable. It was also observed that U rose The statistical analysis (Table 1) showed a significant effect
from 104 W/m2°C to 252 W/m2°C when using a temperature of (p < .05) for the mathematical model. In addition, LoF was not signifi-
70°C and a pressure of 0.80 bar. This is due to the high temperature, cant (p > .05). It has been noted that R-Squared was 0.9986. These in-
which increases the overall heat transfer coefficient. As shown in dicators showed that the mathematical model applies to data and can
Figure 3a, the interference between pressure and temperature can be used to predict the CT and the coefficient of models are depicted
increase the overall heat transfer coefficient better, especially when in Table 2 and Equation(9) which used to calculate CT as follows:
using their maximum values.
CT = −1.03897 + 3.48621P + 0.023948T − 0.022500PT
(9)
+2.59483P2 − 1.6206910−4 T2
3.1.2 | Concentration-time
Regarding the interference between pressure and temperature,
It is observed from Table 1 the central composite design matrix for the three-dimensional Figure 3b that was drawn by the response
the effect of pressure (bar) and temperature (°C) on the CT of the surface methodology. The CT value was 1.11 hr when using pressure

F I G U R E 3 Response surface plot of (a) heat transfer coefficient, (b) concentration-time (c) specific energy consumption, (d) energy
efficiency as a function of process temperature and pressure
8 of 13 | AL-HILPHY et al.

0.40 bar and temperature 50°C, and the CT increased to 3.30 hr surface methodology. The η value was 61.38% when using a pressure
when using pressure 0.80 bar and a temperature of 50°C. In general, of 0.40 bar and a temperature of 50°C and decreased to 34.10% at
lowering the pressure and increasing the temperature can reduce a pressure of 0.80 bar and a temperature of 50°C. η reached 72.03%
the CT due to withdrawal of the largest amount of moisture per unit when using a pressure of 0.40 bar and a temperature of 70°C and
of time and rapid evaporation, respectively (Silveira et al., 2013; decreased to 50.52% at a pressure of 0.80 bar and a temperature of
Yanniotis, 2007). 70°C. This may be due to the decrease in pressure and the increase
in the temperature led to an increase in η. Lowering the pressure
Specific energy consumption and increasing the temperature were increased the amount of evap-
The results showed that the values of the SEC ranged between orated water and reducing CT (Silveira et al., 2015).
2,752.49 kJ/kg at a pressure of 0.4 bar and temperature of 50°C to
6,607 kJ/kg at a pressure of 0.80 bar and a temperature of 50°C
(Table 1). Silveira et al. (2013) found that the specific energy consump- 3.2 | Chemical composition
tion of the skim milk concentration was 3,024 kJ/kg vaporized water in
the first stage, and 2,889 kJ/kg evaporated water in the second stage 3.2.1 | Moisture content
in the two-stage evaporator. Yanniotis (2007) showed that multi-effect
evaporators reuse the latent heat of the steam and thus save energy. According to Table 1, the values of moisture content ranged be-
The statistical analysis in Table 1 showed a significant effect tween 74% at a pressure of 0.40 bar and a temperature of 50°C to
(p < .05) of the mathematical model. Besides, the LoF was not sig- 76.01% at a pressure of 0.80 bar and a temperature of 50°C. Also,
nificant (p > .05). Besides, high R-Squared values indicated that the there was a significant effect (p < .05) of the mathematical model.
mathematical model well-predicted the values of U, SEC, CT, and η. Besides, the LoF was not significant (p > .05). A linear model was
Moreover, according to the coefficients of models in Table 2, and proposed to predict the moisture content (MC) of concentrated milk
Equation (10) was proposed to calculate the SEC. as a function of process temperature (T) and pressure (P) (Table 2

SEC = 1107.9523 + 1360.2225P − 0.1041T + 0.2492PT and Equation 12).

(10)
+6882.1534P2 − 1.1473 × 104 T2 MC = 67.69231 + 11.66667P + 0.075000T − 0.12500TP (12)

Regarding the interference between pressure and temperature, For the interference between pressure and temperature as
the three-dimensional Figure 3c that was drawn by the response sur- shown in the three-dimensional Figure 4a that was drawn by the
face methodology. The value of SEC was 2,752.5 kJ/kg when using response surface methodology, the value of moisture content was
a pressure of 0.40 bar and a temperature of 70°C and decreased to 73.60% when using the pressure 0.40 bar and a temperature of 50°C
2,752.49 kJ/kg at a pressure of 0.40 bar and temperature of 50°C. and increased to 75.28% at a pressure of 0.80 bar and a temperature
of 70°C. The moisture content increased from 74.10% at a pressure
of 0.40 bar and a temperature of 70°C to 75.27% at 0.80 bar and a
3.1.3 | Energy efficiency temperature of 70°C. This is because the increase in pressure leads
to a decrease in the vacuum inside the cylinder, which leads to a
The results showed that the η ranged between 34.30% at a pressure reduction of the amount of water evaporated from the milk (Silveira
of 0.80 bar and a temperature of 50°C to 71.80% at a pressure of et al., 2013; Tanguy et al., 2016). Yanniotis (2007) showed that con-
0.40 bar and a temperature of 70°C as presented in Table 1. Silveira centration is used to reduce the weight and volume of liquid prod-
et al. (2013) found that η was 79% and 81% in the first and second ucts as the concentration reduces the water activity in the nutrient.
stages of the concentration process for skim milk, respectively.
The results of the statistical analysis in Table 1 showed that there
was a significant effect (p < .05) of the mathematical model in η. The 3.2.2 | Protein
results of the statistical analysis showed that the LoF was insignifi-
cant (p > .05). R 2 was 0.9936. These indicators showed the possibil- According to Table 1, the protein values ranged between 6.95% at a
ity of using the nonlinear mathematical model in predicting η values pressure of 0.60 bar and a temperature of 70°C to 8.12% at a pres-
and the model coefficients are illustrated in Table 2, and Equation sure of 0.40 bar and a temperature of 70°C (Table 1). These results
(11) that used to calculate η as follows: were close to the results of a research team that stated the percent-
age of protein in concentrated milk was 8.6% (Madoumier, Azzaro-
𝜂 = 107.34270 − 317.66383P + 1.13674T + 0.72208PT Pantel, Tanguy, & Gésan-Guiziou, 2015).
(11)
2
+177.79707P − 7.44567 × 10 T −3 2 The not significance of the quadratic model and the LoF on the
protein content is presented in Table 1. The concentration process
Regarding the interference between pressure and temperature, affected the physical and chemical conditions of the product. Also,
the three-dimensional Figure 3d that was drawn by the response the reaction of Maillard occurs during the concentration process, due
AL-HILPHY et al. | 9 of 13

F I G U R E 4 Response surface plot of (a) moisture content, (b) protein (c) fat, (d) pH, (e) ash (f) lactose, (i) TSS, (j) ∆E as a function of process
temperature and pressure
10 of 13 | AL-HILPHY et al.

to the increase in the amount of evaporated water due to the high of pH also decreased at a pressure of 0.40 bar and a temperature
temperature that led to the increase in protein (Tanguy et al., 2016). of 60°C from 6.57 to 6.50 at 0.80 bar and a temperature of 70°C.
Table 2 shows the regression coefficients and Equation (13) was Tanguy et al. (2019) clarified that can be expected to decrease pH
used to calculate the protein (Pr) of concentrated milk. during the thermal processing of milk such as concentration and
sterilization processes (Tanguy et al., 2019).
Pr = 7.10 − 0.9699P + 0.108T − 0.022PT + 8.1551P2 − 7.379 × 10−4 T2
(13)

Regarding the interference between pressure and tempera- 3.2.5 | Ash

ture, as shown in the three-dimensional Figure 4b that was drawn
by the response surface methodology. To obtain the highest per- The results showed that the ash values ranged between 1.00% at a
centage of protein, the pressure of 0.40 bar and the temperature pressure of 0.80 bar and a temperature of 60°C to 1.43% at a pressure
of 70 ˚C was used to reach 7.86%, then followed by the pressure of of 0.40 bar and a temperature of 60°C (Table 1). Probably, the pressure
0.40 bar and temperature 50°C, where it amounted to 7.65%. The drop resulted in the withdrawal of the largest amount of moisture and,
increase in pressure causes a decrease in protein content at differ- consequently, the ash increased with the decrease in pressure. These
ent temperatures. For example, the protein decreased insignificantly results are close to the findings of Madoumier el al. (2015) who found
(p > .05) from 7.86% to 7.28% when the pressure increased from 0.4 the ash content of a concentrated milk sample was 2.6%.
to 0.8 bar and a temperature of 70°C. The quadratic model and the LoF have not significant effect
(p > .05) on the ash (Table 1). The regression coefficients of models
are shown in Table 2 and Equation (16) that used to calculate the ash
3.2.3 | Fat of concentrated milk.

Ash = 0.89483 + 3.0273P − 8.90805 × 10−3 T − 2.5 × 10−3 PT

The results showed that the fat values ranged between 7.3% at pres- (16)
sure 0.80 bar and a temperature of 50°C to 7.68% at a pressure of −2.84914P2 + 6.03448 × 10−5 T2

0.60 bar and a temperature of 50°C (Table 1). These results are closed
to the findings of Madoumier el al. (2015) who indicated that the per- According to Figure 4e, the ash value was decreased by increas-
centage of fat for concentrated milk was 7.0%. Also, Table 1 showed no ing pressure. This may be due to an increase in temperature with
significant effect (p > .05) for the quadratic model. For the interaction an increase in the vacuum (decrease in pressure) as it led to an in-
between pressure and temperature as shown in the three-dimensional crease in evaporation and the withdrawal of moisture to the out-
Figure 4c, the fat content was 7.5% when using pressure 0.60 bar and side through the vacuum pump. Tanguy et al. (2019) disclosed that
a temperature of 50°C, and decreased insignificantly (p > .05) to 7.3% evaporation under vacuum depends on some factors such as ther-
at a pressure of 0.80 bar and a temperature of 50°C. Equation(14) was mal treatment before concentration, operation temperature, holding
used to calculate the fat content of concentrated milk as a function of time in the evaporator, and storage time after concentration.
process pressure (P) and temperature (T) (Table 2).

Fat content = 10.79276 − 2.279P + 0.127T − 0.03755PT + 3.913P2 − 8.344 × 10−4 T2 3.2.6 | Lactose
(14)
3.2.4 | pH As presented in Table 1, the lactose values ranged between 8.25%
at a pressure of 0.80 bar and a temperature of 60°C to 9.85% at a
It was observed that the pH values ranged from 6.4 at a pressure of pressure of 0.60 bar and temperature of 60°C. These results were
0.80 bar and a temperature of 50°C to 6.63 at a pressure of 0.60 bar closed to the results of Madoumier el al. (2015) who found that the
and a temperature of 60°C (Table 1). The quadratic model, inde- percentage of lactose in concentrated milk was 12%.
pendent variables, interactions between them, and the LoF have About the interference between pressure and temperature,
not significant effect (p > .05) on pH. The regression coefficients of as shown in the three-dimensional Figure 4f, the value of lactose
models (Table 2). Also, Equation (15) was used to calculate the pH of reached 9.80% when pressure and temperature were 0.40 bar and
concentrated milk samples. 50°C, respectively. The lactose content was decreased to 8.41% at
a pressure of 0.80 bar and temperature 50°C. Also, it decreased at a
pH = 7.268 − 0.123P + 0.025T − 0.020PT − 0.9187P2 + 1.327 × 10−4 T2 pressure of 0.40 bar and a temperature of 70°C from 9.38% to 8.94%
(15) at 0.80 bar and a temperature of 70°C. The regression coefficient
For the interference between pressure and temperature as illustrated in Table 2 and Equation (17) was proposed to calculate the
shown in the three-dimensional Figure 4d that was drawn by the re- lactose of concentrated milk.
sponse surface methodology. The value of pH was 6.53 when using Lact. = 16.77155 + 2.28764P − 0.24941T + 0.11875PT
pressure 0.40 bar and temperature 50°C. The value of pH decreased (17)
−9.73276P2 + 1.507 × 10−3 T2
to 6.44 at a pressure of 0.80 bar and temperature 50°C. The value
AL-HILPHY et al. | 11 of 13

3.2.7 | Total solids soluble at higher temperatures due to the relationship between Maillard re-
action and process temperature (Kareb, Champagne, & Aïder, 2016).
The results presented in Table 1 indicated that the TSS values ranged Also, a higher concentration of dry matter can be responsible for the
between 24% at a pressure of 0.80 bar and a temperature of 60°C color change. Furthermore, the color changes could be related to
and 50°C to 26% at a pressure and temperature of 0.40 bar and the final concentration of the product. Besides, it was observed that
50°C; 0.6 bar and 60°C; 0.4 bar and 70°C; 0.4 bar and 60°C and the effects of model were significant. Also, R-Squared was 0.9376.
0.6 bar and 50°C. This may be due to a decrease in pressure, as the Therefore, the reduced cubic model (RCM) can be suggested to predict
vacuum can remove a large amount of moisture. Also, there was a the ∆E for this RW concentration process. ∆E is given in Equation 19:
significant effect (p < .05) of the linear model on the TSS. In addition, ΔE = 196.832 − 225.517P − 7.306T + 11.143PT
the LoF was not significant (p > .05). R 2 was 0.6525. These indicators (19)
−154.859P2 + 0.0675T2 + 2.23P2 T − 0.115PT2
showed that the linear model can be used to predict TSS values an
illustrated in Equation (18):
For the interaction between pressure and temperature as shown
TSS = +32.30769 − 11.66667P − 0.075000T + 0.12500PT (18) in the three-dimensional Figure 4j which depicted by the response
surface methodology, the highest value of ∆E was 12.63 at pressure
Regarding the interference between pressure and temperature as and temperature of 0.40 bar and 70°C, respectively. Also, ∆E was de-
shown in the three-dimensional Figure 4i that was drawn by the re- creased to 7.79 when process pressure and temperature were 0.80 bar
sponse surface methodology, the value of TSS was 26.3 when using a pressure and 50°C, respectively. Furthermore, in general, ∆E was re-
pressure of 0.40 bar and a temperature of 50°C. The value of TSS has duced by decreasing the temperature (Figure 4j). It was previously ex-
decreased to 24.2 when using a pressure of 0.80 bar and a temperature plained that reducing the milk concentration temperature can limit the
of 50°C. This may be due to the increase in pressure that reduced the changes in the overall changes of the product color due to the reduced
amount of water evaporated from the milk and raised the boiling point. chemical degradations and reactions such as the Maillard reaction.

3.3 | Overall color change 3.4 | Process optimization and models validation

Table 1 shows that the ∆E values were ranged from 7.63 to 12.48 at The results of the optimization process of the thermal perfor-
pressure and temperature ranges of 0.40–0.80 bar and 50–70°C, re- mance of the RW concentrator, chemical properties, and color
spectively. At constant pressure, greater values of ∆E were observed components are illustrated in Table 3. The result revealed that

TA B L E 3 Results of the optimization process at optimum and central conditions for milk concentration using RW and CC

RW

Optimum level (p = 0.40 bar; T = 69.07°C) Central levels (p = 0.6 bar; T = 60°C)

Dependent Variables Experimental Predicted Experimental Predicted Experimental CC

U (W/m2°C) 113.21b ± 1.79 113.20 b 142.00a ± 3.07 144.56 a 98.24c ± 3.12

c c b b
CT (h) 1.18 ± 0.07 1.10 2.00 ± 0.12 2.03 3.00a ± 0.051
SEC (kJ/kg) 2,958.39c ± 15.34 2,959.27 c 4,403.98a ± 9.65 4,406.67a 3,952.80 b ± 6.32
a a b b
η% 68.14 ± 1.21 68.15 46.65 ± 1.98 48.15 30.25c ± 3.86
MC% 74.61b ± 0.30 74.23 b 74.00 c ± 3.56 74.69 c 75.00a ± 0.01
b b c c
Pr% 7.43 ± 0.04 7.59 7.13 ± 1.02 7.26 7.70a ± 0.26
Fat% 7.45a ± 0.07 7.43 a 7.66a ± 0.09 7.47 a 7.20 b ± 0.05
b b a a
Ash% 1.25 ± 0.09 1.30 1.42 ± 0.01 1.28 1.11c ± 0.06
Lact.% 9.24b ± 0.17 9.50 b 9.79a ± 0.96 9.38 a 9.00 c ± 0.15
a a a a
TSS% 25.97 ± 0.14 26.00 26.00 ± 1.08 25.97 25.00 b ± 0.42
pH 6.51a ± 0.078 6.48 a 6.60a ± 0.12 6.54 a 6.47b ± 0.03
c c a a
∆E 9.49 ± 0.64 9.58 11.44 ± 0.15 11.26 10.77b ± 1.23

The different l letters in the rows indicate a significant effect at level of 0.05.
Abbreviations: CC, conventional concentration; CT, concentration-time; MC, moisture content; Lact., lactose; P, pressure; Pr., protein; RW,
refractance window; SEC, specific energy consumption; SS, sum of square; T, temperature; TSS, total soluble solids; U, heat transfer coeffecient; η,
energy effeciency; ∆E, total color change.
12 of 13 | AL-HILPHY et al.

the optimum milk concentration process conditions were a tem- CVS Computer vision system
perature of 69.07°C and a pressure of 0.43 bar. These conditions CC Conventional concentration
provide a concentrated milk with predicted U (113.20 W/m2 °C), CIE Commission international d’eclairage
CT (1.1 hr), SEC (2,959.27 kJ/kg), η (68.15%), moisture content CP Specific heat (kJ/kg.°C)
(74.23%), protein (7.59%), fat (7.43) ash (1.30%), lactose (9.50%), Cpi Specific heats of milk composition (kJ/kg.°C)
TSS (26.00%), pH (6.48) and ∆E (9.58). CT Concentration-time
To check the veracity of the predicted values by models at opti- D Desirability
mized conditions, validation experiments were performed based on E Overall color
the central and optimum concentration process with 6 replications. F Fat (%)
As illustrated in Table 3, when milk concentrated by optimum or cen- L* Lightness
tral conditions, the experimental results were similar to the predicted M Mass flow rate of milk (kg/s)
results by using optimized models, which verified that the validity of LoF Lack of Fit
response regression models. This reflects the actual results of the LMTD Logarithmic mean temperature difference
independent variables. Moreover, there is a significant difference r Impotence
(p < .05) between the experimental data for the optimum level and R Correlation coefficient
the experimental data for the central level. The U, η, fat, ash, lactose RCM Reduced cubic model
content, TSS, and pH for milk concentrated by CC were lower than RM Raw milk
those of optimized and central levels of RW. Furthermore, the CT, RSM Response surface methodology
and MC of milk concentrated by CC were higher than both optimized RW Refractance window
and central levels of RW. This study demonstrated that the RW con- SEC Specific energy consumption
centrator system possesses higher efficiency and thermal perfor- SS Sum of squares
mance could significantly improve the concentrated milk quality. Std. Dev. Standard deviation
T Temperature (°C)
TSS Total soluble solid (%)
4 | CO N C LU S I O N S U Overall heat transfer (W/m2°C)
x Value
RW was found to be superior to the conventional concentration Greek symbols
method of milk concentration in terms of processing time, energy
consumption, energy efficiency, and some quality parameters of β Constant
the product. This innovative technique saved about 61% of the pro- γ Constant
cessing time and reduced the changes in the product color during ∆ Differences
concentration. Also, it was revealed that selecting the appropriate η Energy efficiency
pressure and temperature can enhance the performance of the RW λw Latent heat (kJ/kg)
system and product characteristics. The food industry may benefit Subscripts
from such emerging approaches in the future after further evalua-
tions including up-scaling studies. i Linear
i Response
AC K N OW L E D G M E N T ii Quadratic
The authors thank the Department of Food Science, College of h0 Hot water
Agriculture, University of Basrah for providing the laboratory of m0 Hot milk
food engineering and facilities. m Milk
m Mean temperature
C O N FL I C T O F I N T E R E S T mi Cold milk
The authors have declared no conflicts of interest for this article. hi Cold water
º Interceptpre.Predicted
N O M E N C L AT U R E
a* Redness-greenness ORCID
Adeq Adequate precision Asaad R. Al-Hilphy https://orcid.org/0000-0001-5850-1519
Adj Adjusted José M. Lorenzo https://orcid.org/0000-0002-7725-9294
ANOVA Analysis of variance Francisco J. Barba https://orcid.org/0000-0002-5630-3989
b* Yellowness-blueness Mohsen Gavahian https://orcid.org/0000-0002-4904-0519
AL-HILPHY et al. | 13 of 13

REFERENCES Musina, O., Rashidinejad, A., Putnik, P., Barba, F. J., Abbaspourrad,
A., Greiner, R., & Roohinejad, S. (2018). The use of whey pro-
Al-HilphyShirkole, A. R. S., Ali, H. I., & Mohsin, G. F. (2019). Technology
tein extract for manufacture of a whipped frozen dairy dessert.
of ohmic heating for the pasteurization of milk for the pasteurization
Mljekarstvo, 68(4), 254–271. https://doi.org/10.15567/​m ljek​
of milk. Novel Dairy Processing Technologies (Vol. 17, 3–46). New York,
arstvo.2018.0402
NY: Taylor & Francis . https://doi.org/10.1201/97813​15167​121-1
Parmar, P., Singh, A. K., Meena, G. S., Borad, S., & Raju, P. N. (2018).
AOAC. (2016). Official methods of anaylsis. Retrieved from http://www.
Application of ohmic heating for concentration of milk. Journal
eoma.aoac.org/metho​ds/info.asp?ID=16264.
of Food Science and Technology, 55(12), 4956–4963. https://doi.
Bernaert, N., VanDroogenbroeck, B., VanPamel, E., & DeRuyck, H. (2019).
org/10.1007/s1319​7-018-3431-4
Innovative refractance window drying technology to keep nutrient
Pehlivan, H., & Özdemir, M. (2012). Experimental and theoretical inves-
value during processing. Trends in Food Science and Technology, 84,
tigations of falling film evaporation. Heat and Mass Transfer/Waerme-
22–24. https://doi.org/10.1016/j.tifs.2018.07.029
Und Stoffuebertragung, 48(6), 1071–1079. https://doi.org/10.1007/
Chamberland, J., Bouyer, A., Benoit, S., Provault, C., Bérubé, A., Doyen,
s0023​1-011-0962-x
A., & Pouliot, Y. (2020). Efficiency assessment of water reclamation
Raghavi, L. M., Moses, J. A., & Anandharamakrishnan, C. (2018). Refractance
processes in milk protein concentrate manufacturing plants: A pre-
window drying of foods: A review. Journal of Food Engineering, 222,
dictive analysis. Journal of Food Engineering, 272, 109811. https://doi.
267–275. https://doi.org/10.1016/j.jfood​eng.2017.11.032
org/10.1016/j.jfood​eng.2019.109811
Sangrame, G., Bhagavathi, D., Thakare, H., Ali, S., & Das, H. (2000).
Eren, İ., & Kaymak-Ertekin, F. (2007). Optimization of osmotic dehydration of
Performance evaluation of a thin film scraped surface evaporator
potato using response surface methodology. Journal of Food Engineering,
for concentration of tomato pulp. Journal of Food Engineering, 43(4),
79(1), 344–352. https://doi.org/10.1016/j.jfood​eng.2006.01.069
205–211. https://doi.org/10.1016/S0260​-8774(99)00150​-8
Faion, A. M., Becker, J., Fernandes, I. A., Steffens, J., & Valduga, E. (2019). Sheep’s
Silveira, A. C. P., de Carvalho, A. F., Perrone, Í. T., Fromont, L., Méjean, S.,
milk concentration by ultrafiltration and cheese elaboration. Journal of
Tanguy, G., … Schuck, P. (2013). Pilot-scale investigation of effective-
Food Process Engineering, 42(4), https://doi.org/10.1111/jfpe.13058
ness of evaporation of skim milk compared to water. Dairy Science
Fernández-Vázquez, R., Stinco, C. M., Hernanz Vila, D., Heredia, F. J., Chaya,
and Technology, 93(4–5), 537–549. https://doi.org/10.1007/s1359​
C., & Vicario, I. M. (2018). Internal preference mapping of milk-fruit
4-013-0138-1
beverages: Influence of color and appearance on its acceptability. Food
Silveira, A. C. P., Tanguy, G., Perrone, Í. T., Jeantet, R., Ducept, F., de
Science & Nutrition, 6(1), 27–35. https://doi.org/10.1002/fsn3.494
Carvalho, A. F., & Schuck, P. (2015). Flow regime assessment in falling
Gavahian, M., Farahnaky, A., Javidnia, K., & Majzoobi, M. (2012).
film evaporators using residence time distribution functions. Journal
Comparison of ohmic-assisted hydrodistillation with traditional hy-
of Food Engineering, 160, 65–76.
drodistillation for the extraction of essential oils from Thymus vul-
Stratakos, A. C., Inguglia, E. S., Linton, M., Tollerton, J., Murphy, L.,
garis L. Innovative Food Science and Emerging Technologies, 14, 85–91.
Corcionivoschi, N., … Tiwari, B. K. (2019). Effect of high pressure
https://doi.org/10.1016/j.ifset.2012.01.002
processing on the safety, shelf life and quality of raw milk. Innovative
Gavahian, M., Sheu, F. H., Tsai, M. J., & Chu, Y. H. (2020). The effects of
Food Science & Emerging Technologies, 52, 325–333. https://doi.
dielectric barrier discharge plasma gas and plasma-activated water
org/10.1016/j.ifset.2019.01.009
on texture, color, and bacterial characteristics of shiitake mushroom.
Tanguy, G., Siddique, F., Beaucher, E., Santellani, A. C., Schuck, P., &
Journal of Food Processing and Preservation, 44(1), e14316. https://doi.
Gaucheron, F. (2016). Calcium phosphate precipitation during con-
org/10.1111/jfpp.14316
centration by vacuum evaporation of milk ultrafiltrate and microfil-
Guimarães, B., Martins, M. J. N., Flauzino, R. D., Basso, R. C., & Telis
trate. LWT - Food Science and Technology, 69, 554–562. https://doi.
Romero, J. (2020). Thixotropy of sweetened condensed milk applied to
org/10.1016/j.lwt.2016.02.023
flow fluid dynamics analysis of cylindrical pipes. Journal of Food Process
Tanguy, G., Tuler-Perrone, I., Dolivet, A., Santellani, A. C., Leduc, A.,
Engineering, 43(6), e13397. https://doi.org/10.1111/jfpe.13397
Jeantet, R., … Gaucheron, F. (2019). Calcium citrate insolubilization
Jafari, S.-M., Azizi, D., Mirzaei, H., & Dehnad, D. (2016). Comparing qual-
drives the fouling of falling film evaporators during the concentration
ity characteristics of oven-dried and refractance window-dried ki-
of hydrochloric acid whey. Food Research International, 116, 175–183.
wifruits. Journal of Food Processing and Preservation, 40(3), 362–372.
https://doi.org/10.1016/j.foodr​es.2018.08.009
https://doi.org/10.1111/jfpp.12613
Wasnik, P. G., Menon, R. R., Sivaram, M., Nath, B. S., Balasubramanyam,
Jebson, R. S., & Iyer, M. (1991). Performances of falling film evaporators.
B. V., & Manjunatha, M. (2019). Development of mathematical model
Journal of Dairy Research, 58(1), 29–38. https://doi.org/10.1017/
for prediction of adulteration levels of cow ghee with vegetable fat
S0022​02990 ​0 033483
using image analysis. Journal of Food Science and Technology, 56(4),
Kareb, O., Champagne, C. P., & Aïder, M. (2016). Contribution to the
2320–2325. https://doi.org/10.1007/s1319​7-019-03677​-x
production of lactulose-rich whey by in situ electro-isomerization
Yam, K. L., & Papadakis, S. E. (2004). A simple digital imaging method
of lactose and effect on whey proteins after electro-activation as
for measuring and analyzing color of food surfaces. Journal of
confirmed by matrix-assisted laser desorption/ionization time-of-
Food Engineering, 61(1), 137–142. https://doi.org/10.1016/S0260​-
flight-mass spectrometry and sodium. Journal of Dairy Science, 99(4),
8774(03)00195​-X
2552–2570. https://doi.org/10.3168/jds.2015-10037
Yanniotis, S. (2007). Solving problems in food engineering. Athens, Greece:
Khuri, A. I., & Cornell, J. A. (2019). Response surfaces: Designs and analyses
Springer Science and Business Media.
(2nd ed.). Retrieved from https://www.bookd​eposi​tory.com/Respo​
nse-Surfa​ces-Desig​ns-Analy​ses-Andre​-I-Khuri/​97803​67401252
Madoumier, M., Azzaro-Pantel, C., Tanguy, G., & Gésan-Guiziou, G. (2015).
Modelling the properties of liquid foods for use of process flowsheeting How to cite this article: Al-Hilphy AR, Ali HI, Al-IEssa SA,
simulators: Application to milk concentration. Journal of Food Engineering, Lorenzo JM, Barba FJ, Gavahian M. Optimization of process
164, 70–89. https://doi.org/10.1016/j.jfood​eng.2015.04.023 variables on physicochemical properties of milk during an
Moejes, S. N., vanWonderen, G. J., Bitter, J. H., & vanBoxtel, A. J. B.
innovative refractance window concentration. J Food Process
(2020). Assessment of air gap membrane distillation for milk con-
centration. Journal of Membrane Science, 594, 117403. https://doi. Preserv. 2020;44:e14782. https://doi.org/10.1111/jfpp.14782
org/10.1016/j.memsci.2019.117403