Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

Contents lists available at ScienceDirect

Food and Bioproducts Processing

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

Drying kinetic, quality, energy and exergy

performance of hot air-rotary drum drying of green
peas using adaptive neuro-fuzzy inference system

Mohammad Kaveh a , Yousef Abbaspour-Gilandeh a,∗ , Guangnan Chen b,∗

a Department of Biosystems Engineering, College of Agriculture and Natural Resources, University of Mohaghegh
Ardabili, Ardabil, Iran
b Faculty of Health, Engineering and Sciences, University of Southern Queensland, Toowoomba, QLD 4350, Australia

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

Article history: The purpose of this paper was to study the drying kinetic, quality, energy and exergy perfor-
Received 3 July 2020 mance of green peas in a hot air-rotary drum dryer using the Adaptive Neuro-Fuzzy Inference
Received in revised form 15 August System (ANFIS) model. The effect of air temperature and rotary rotation speed was exam-
2020 ined. The drying tests were performed at three air temperature levels of 40, 55 and 70 ◦ C and
Accepted 23 August 2020 three rotation speeds of 5, 10 and 15 rpm. The modeling of kinetic, quality (color, shrink-
Available online 1 September 2020 age, and rehydration ratio), energy and exergy indices of the green pea drying was also
investigated using the adaptive neuro-fuzzy inference system (ANFIS). The results showed
Keywords: that the energy utilization rate of the process varied in the range of 0.0121–0.1556 kJ/s. The
Green pea minimum and maximum values of energy utilization ratio were respectively obtained in the
Hot air-rotary drum dryer drying conditions at 70 ◦ C and rotation speed of 15 rpm, and at 40 ◦ C and rotation speed of 15
Energy rpm. Reducing the drying temperature and speed of rotation reduced the quality properties.
Color Minimum values of color (E) and shrinkage was 64.95 ± 1.67 and 39.19 ± 1.05, respectively.
Exergy improvement potential The exergy loss rate and average exergy efficiency varied from 0.02 to 0.11 kJ/s and from
0.5434 to 0.8382, respectively. The average rate of exergy improvement potential increased
with increasing air temperature and rotary rotation speed. The highest R2 value for the pre-
diction of moisture ratio, energy utilization, energy utilization ratio, exergy loss and exergy
efficiency with the ANFIS model was 0.9996, 0.9999, 0.9995, 0.9989 and 0.9996, respectively.
© 2020 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.

1. Introduction Drying of crops is often a very energy-intensive process (Bundschuh

and Chen, 2014; Sivakumar et al., 2016). Therefore, it is very important
Drying of agricultural products is one of the oldest preservation meth- to be able to design, manufacture new dryers that can significantly
ods in the post-harvest stage (Chen, 2015). Drying is usually referred to reduce the energy uses. In recent years, due to the decreased energy
as the process of moisture reduction due to the presence of simultane- resources in the world, several innovative methods have been devel-
ous heat and mass transfer between the environment and the product oped to reduce the time and energy consumption (Soysal et al., 2006;
surface (Zohrabi et al., 2020a). In this process, the reduction of moisture Onwude et al., 2016). In the hot-air drying process, due to the low ther-
content leads to the decreased or ceased microbial activity. Further- mal conductivity of the material, the rate of heat transfer to the internal
more, the dried crop will also achieve a lower weight and volume, which parts is usually low, which, as a result, increases the time and energy
greatly facilitates the rest of the steps until the consumption, including required to perform the drying process. On the other hand, due to the
the handling and transport (Jahanbakhshi et al., 2020). lack of proper time contact between the material and the hot air, the air

∗
Corresponding authors.
E-mail addresses: [email protected] (Y. Abbaspour-Gilandeh), [email protected] (G. Chen).
https://doi.org/10.1016/j.fbp.2020.08.011
0960-3085/© 2020 Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 169

Nomenclature wt+t Weight of product at time t + t (kg)

Wout Final dry weight (g)
A1 , A2 B1 , B2 , C1 , C2 , D1 and D2 Language indicators Win Initial wet weight (g)
Adc Cross-sectional area of the drying chamber (m2 ) Xn Normalized value of the parameter
Cv Specific heat of vapor(kj/kg K) Xi Real value of each parameter
Ca Specific heat of air (kj/kg K) Xmin Minimum value of the desired parameter
Cpout Specific heat of outlet air (kj/kg ◦ C) Yi Predicted output
Cpin Specific heat of inlet air (kj/kg ◦ C) Yo Actual output
Ėxd Destruction exergy (kw or kj/s) a Density (kg/m3 )
Ėxloss Exergy Loss (kw or kj/s) ϕ Relative air humidity (%)
EU Energy utilization (kj/s) Ex Exergy efficiency
EUR Energy utilization ratio (kwh/kg) ω∞ Absolute humidity of drying (kg water/kg dry
Ėin Inlet energy rate, kw air)
Ėout Outlet energy rate, kw ωout Absolute humidity of outlet air (kg water/kg dry
Ėxin Inlet energy rate (kw or kj/s) air)
Ėxout Outlet Exergy rate (kw or kj/s) ωa Absolute humidity of inlet air (kg water/kg dry
Ėxevap Exergy evaporation (kw or kj/s) air)
f1 and f2 First-degree polynomials ω Absolute humidity of air (kg water/kg dry air)
h Specific enthalpy (kj/kg) P1 , P2 , q1 , q2 , r1 , r2 , u1 , u2 Linear coefficients of the output
hfg Latent heat of vaporization of water (kj/kg) function
hin Inlet enthalpy (kj/kg) Ai , Bj , Cj , Di Membership functions of the language
hout Outlet enthalpy (kj/kg) variables a, b, c and d
hpe Outlet enthalpy (kj/kg) t Time between two sample weighing (s)
IP Improvement potential rate (kj/s) f Final volume (cm2 )
ṁin Inlet mass flow rate (kg/s) i Initial volume (cm2 )B
Mb Initial moisture content (g water/g d.m.)
MC Moisture content (% d.b.)
Me Equilibrium moisture content (g water/g dry
leaving the dryer chamber is often not fully saturated. This means the
matter)
removal of energy from the dryer (loss of exergy) and further reduction
MR Moisture ratio in the efficiency of the dryer (Motevali et al., 2014). One of the new meth-
Mt Moisture content (g water/g dry matter) ods presented by researchers is the use of rotary dryer. In this type of
ṁout Outlet mass flow rate (kg/s) dryers, the product rotates throughout the process, and its contact with
ṁwout Humidity ratio of outlet air (kg water/kg drying the inlet air causes the air heat to completely cover the whole product.
Air) As a result, less energy is required to dry the products (Ghasemkhani
ṁwin Humidity ratio of inlet air (kg water/kg drying et al., 2016).
air) To increase the energy utilization, the analysis of energy and exergy
(maximum effective work resulting from a certain amount of available
ṁv Mass transfer rate (kg water/s)
energy or from a flow of materials) has received significant attention.
ṁp Product mass (kg)
In the exergy analysis, the main goal is to determine the location
ṁw Water mass (kg)
and amount of irreversible production during various processes of the
ṁa Air mass (kg) thermodynamic cycle and the factors affecting the production of this
Wr Weight after rehydration irreversibility (Yu et al., 2019; Johnson and Langrish, 2018). In this way,
Wd Initial weight of the dry sample in addition to evaluating the efficiency of different components of the
n Number of data patterns thermodynamic cycle, the ways to increase the cycle efficiency are also
P Atmospheric pressure (kpa) identified (Mokhtarian et al., 2016; Aghbashlo et al., 2013; Taghinezhad
Pa Saturated vapor pressure inlet air (kpa) et al., 2020). The decreasing trend of exergy losses and the increase
Pout Saturated vapor pressure outlet air (kpa) in the exergy efficiency from the drying engineering point of view is
P∞ Saturated vapor pressure ambient pressure important, because the moisture removal rate is inversely related to
exergy loss and directly related to exergy efficiency (Azadbakht et al.,
(kpa)
2017).
Pvs Saturated pressure (kpa)
As mentioned, the analysis of energy and exergy is an effective
Q̇ev Evaporation hear transfer rate (kj/s)
tool in optimizing the drying process. Therefore, the study in this
Rv Gas constant (kj/kg/C) for water vapor field is of special importance in various industrial processes. Darvishi
R∞ Gas constant (kj/kg/C) for ambient air et al. (2016), for example, analyzed the energy and exergy efficiency
t Drying time (min) in a microwave dryer at different power levels and thicknesses of kiwi
T Temperature (◦ C or K) slices. They showed that increasing the microwave power increases
Tin Temperature of inlet air (◦ C) the energy and exergy efficiency. Corzo et al. (2008); Castro et al. (2018)
Tref Temperature of ambient air (◦ C) and Akpinar (2004) examined the energy and exergy of the drying pro-
Tout Temperature of outlet air (◦ C) cess in the hot-air dryers for coroba slices, onion, and red pepper,
respectively. Based on the results, they found that increasing the air
Tp Product average temperature (C or K) during
temperature can have a great impact on the increased energy and
drying
exergy efficiency. Karthikeyan and Murugavelh (2018) also analyzed
A Rotary rotation speed (rpm)
the energy and exergy of turmeric in a mixed-mode forced convection
Va Volumetric flow rate (m3 /s) solar tunnel dryer. Vijayan et al. (2020) further examined the energy
wt Weight of product at time t and exergy of bitter gourd slices in an indirect forced convection solar
dryer. Aktas et al. (2017) also performed the energy and exergy analysis
170 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

of carrot in two heat pump and infrared–heat pump dryers at differ- drum dryer (Fig. 1) was designed and manufactured in the
ent temperatures. They employed two different temperatures for the Department of Biosystem Engineering, Mohaghegh Ardabili
inlet air and observed that the energy and exergy efficiency increases University, Ardabil, Iran. The dryer includes: (1) air blower
with the increase in the inlet air temperature. Also, in this research, the (centrifugal fan (1 hp/3000 rpm)), (2) electric heater (3 ele-
use of infrared power in the whole drying process increases the exergy
ments) (3) dryer cylinder, (4) cylinder rotation system, and
efficiency.
(5) electrical panel. To control the velocity of the inlet air, an
Aghbashlo et al. (2008) dried the potato slices using the semi-
inverter (LS, Korea) was used. The drying of green pea seeds
industrial continuous band dryer method. They analyzed the energy
and exergy of the drying process. They reported that the highest energy was done inside the dryer cylinder. The dryer cylinder has 2
and exergy efficiency of the dryer was achieved at the highest temper- ball bearings on the central pipe, and the gearbox (VF 861/100)
ature and velocity of inlet air and the lowest feed rate. Ghasemkhani was used for the rotation. The cylinder speed can be adjusted
et al. (2016) also studied the drying energy and exergy of apple in a and changed to different speeds. For this purpose, the inverter
rotating-tray dryer and demonstrated that the temperature and rotary (LS, Korea) was used. The temperature of the air entering the
rotation speed had a very important influence in reducing the drying dryer chamber was measured by a K-type thermocouple. Two
time and increasing the exergy efficiency. thermostats (Atbin Mega, Iran) were also used to control the air
Today, with the rapid development of computer processing
temperature in the circuit. To prevent the heat loss, the entire
technologies and software, the advantages of artificial intelligence
dryer was insulated with glass wool. The dryer was turned on
technology are also used to solve the problems related to the modeling
half an hour before the start of each experiment to create sta-
of processes (Taghinezhad et al., 2020; Aghbashlo et al., 2015; Sun et al.,
2019). A statistical model that has recently attracted great attention is
ble and uniform conditions at the onset of drying. To perform
the adaptive neuro-fuzzy inference system (ANFIS), which is a combi- the experiments, the drying of green pea was set at three tem-
nation of fuzzy structures and artificial neural networks. This model perature levels (40, 55 and 70 ◦ C), one inlet air level (1 m/s) and
has its significant advantages, such as the ability to simulate nonlin- three drum rotation speeds (5, 10 and 15 rpm). The green pea
ear systems, high accuracy and less time to build the model and its drying experiments were performed in three replications.
calculations (Kaveh et al., 2018). In most studies, the drying kinetics has been reported
Using the artificial intelligence, a number of studies have been based on the moisture ratio (MR). The moisture ratio was
reported on the prediction of energy and exergy efficiency of drying calculated using Eq. (2) according to the initial moisture, equi-
process for various agricultural products such as quince (Abbaspour-
librium moisture and mass moisture (Ghasemkhani et al.,
Gilandeh et al., 2020), potato cubes (Azadbakht et al., 2017), mushroom
2016):
(Liu et al., 2019), carrot (Nazghelichi et al., 2011), orange slices
(Azadbakht et al., 2018), pomegranate arils (Nikbakht et al., 2014), and
Mt − Me
banana slices (Taheri-Garavand et al., 2018) by different dryers. How- MR = (2)
Mb − Me
ever, there are still very few reported data on the kinetic, quality, energy
and exergy analysis of the green pea drying process using the new hot
2.3. Quality properties
air-rotary dryer. Therefore, this study aimed to study the drying kinet-
ics, quality, and energy and exergy performance of green pea drying by
a new hot air-rotary drum dryer at different temperatures and rotary 2.3.1. Color
rotation speeds. The use of ANFIS to predict the moisture ratio, energy To investigate the color indices, color space was used. L* color
and exergy parameters was also explored. index indicates the brightness of the sample whose range
varies from zero (absolute black) to 100 (absolute white) (Xiao
2. Materials and methods et al., 2014; Zhang et al., 2020). a* index ranges in −60 to +60
and positive values are corresponding to red color while the
2.1. Preparation of samples negative ones are associated with green color (Zhang et al.,
2020). Similar to a*, b* also varies between −60 to +60 in which
Green pea is one of the most common and popular leguminous the positive and negative values are indicative of yellow and
vegetable crops grown in the world. It has a delicious taste, and blue color, respectively (Xiao et al., 2014). The color indices
high nutritive value. In this project, the fresh green peas were were evaluated by an HP scanner (G3110, China). Total color
purchased from a local farm in Sardasht in West Azarbaijan difference (E) was considered for the overall evaluation of
province of Iran and stored in a refrigerator at 3–5 ◦ C. In order the color in fresh and dried green pea as expressed in Eq. (3)
to determine the moisture content of fresh and dried green (Ghasemkhani et al., 2016).
peas, three samples of at least 10 g were placed in a dish and 
2 2 2
placed in an oven (Memmert UFB500, Germany) with the tem- E = (L∗ ) + (a∗ ) + (b∗ ) (3)
perature of 103 ◦ C for 24 h until reaching a constant weight
(Zielinska et al., 2013a; Marvasti et al., 2015). After removing 2.3.2. Shrinkage
from the oven and weighing, the moisture content (MC) was To measure the shrinkage in the green peas, the volume vari-
calculated based on the wet weight by Eq. (1) (Zohrabi et al., ation of the samples was assessed by the fluid displacement
2020b). The obtained moisture content of green pea was 73.23 method through the use of toluene. Shrinkage percentage (Sb )
± 0.5 (%, w.b). was calculated by determination of the initial (i ) and final (f )
volumes through Eq. (4) (Kaveh et al., 2018).
(Win − Wout )
MC = (1)
Win (i − f )
Sb = × 100 (4)
i
2.2. Dryer
2.3.3. Rehydration capacity
A variety of drying methodologies are available for drying peas To assess the rehydration capacity (RC) of the samples, green
including rotary drying, fluid-bed drying, microwave drying peas were immersed in water at 25 ◦ C after drying and weight
and infrared drying etc. To conduct this study, a hot air-rotary measurement (Wd ). After 20 min, samples were removed
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 171

Fig. 1 – Schematic diagram of hot air- rotary drum dryer chamber. (1) Air blow system (1 hp/3000 rpm), (2) Airflow control
system, (3) Air heating chamber. (4) Out let temperature sensor (k- tyoe thermocouple), (5) Airflow transfer tube, (6) Control
panel, (7) Motor for rotating the drum, (8) Anemometer for air velocity measurement, (9) Drying chamber, (10) Feed hopper,
(11) Air exit opening, (12) Rotary drum, (13) Digital Balance, (14) Thermometer, (15) Computer and (16) Chassis.
from the water and their surface water was dried. Sam- h3 = (ha )3 + ω(hg )3 (7d)
ples were then weighed again (Wr ). The rehydration capacity
was estimated through the following equation (Kaveh and According to the first law of thermodynamics, the rate of
Abbaspour-Gilandeh, 2020): energy utilization (J/s) is expressed as Eq. (8) (Azadbakht et al.,
2017).
Wr
RC = (5)
Wd Eu = ṁa (hin − hout ) (8)

2.4. Thermodynamic analysis of dryer •

The equations related to the air mass flow rate mp (J/s), dry
air density (kg/m3 ), air enthalpy at the inlet and outlet of the
2.4.1. Energy analysis
dryer chamber, specific inlet and outlet heat (Cpin and Cpout ),
In order to perform the energy and exergy analyses, the green •
pea drying was considered as a steady flow process and the heat transfer rate due to evaporation (Q ev ), relative humidity,
information obtained from the drying experiments was used. heat transfer rate (kg water/s), and saturated vapor pressure
In general, the mass balance equations can be shown as (kPa) are shown in Table 1.
follows (Akpinar, 2005; Aktas et al., 2017): The ratio of energy consumed during the process to the
energy supplied to the system is defined as the energy utiliza-
 •  • tion ratio and is calculated by Eq. (21) (Abbaspour-Gilandeh
min = mout (6)
et al., 2020).

• • • •
For product(mp )2 = (mp )4 = mp (6a) ma (hin − hout )
EUR = • (21)
• • • •
ma (hin − hpe )
For water(ω1 ma )2 + (mw )2 = ω3 ma + (mw )4 (6b)
2.4.2. Exergy analysis
• • •
For air(ma ) = (ma )3 = ma (6c) Exergy analysis is performed based on the second law of ther-
modynamics. This law states that in addition to quantity,
Also, the energy balance equation is obtained using the energy also has quality, and a real thermodynamic process
following formula (Dinçer and Zamfirescu, 2015): occurs through the reduction of energy quality. Fig. 2 shows
a schematic diagram of the drying process in a hot air-rotary
• • drum dryer chamber where the input and output parameters
Ein = Eout (7a)
are specified. The exergy balance equation for the hot air-
rotary drum dryer is based on Eq. (22) (Dinçer and Zamfirescu,
The energy balance equation for the whole system can be 2015):
calculated as follows (Erbay and Icier, 2011):
• •
• • • • • Exin = Exout (22)
ma h1 + mp (hp )2 + (mw )2 (hw )2 = ma h3 + mp (hp )4
• The exergy balance equation for the inlet and outlet con-
+ (mw )4 (hw )4 + Qloss (7b)
ditions can be written as Eq. (23) (Erbay and Icier, 2011):
• • • • •
h1 = (ha )1 + ω1 (hv )1 =(h
˜ a )1 + ω(hg )1 (7c) Exin = Exout + Exd + Exloss − Exevap (23)
172 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

Table 1 – Formulas used for determining energy utilization of hot air-rotary drum dryer.
Equation Eq. number Reference

ṁa = a Va Adc (9) Taheri-Garavand et al. (2018)

101.325
a = 0.287(T +273.16)
(10) Abbaspour-Gilandeh et al. (2020)
a
hin = Cpin (Tin − Tref ) + hfg ṁwin (11) Akpinar et al. (2005)
hout = Cpout (Tout − Tref ) + hfg ṁwout (12) Manalu et al. (2016)
Cpin = 1.004 + 1.88ṁwin (13) Nazghelichi et al. (2011)
Cpout = 1.004 + 1.88ṁwout (14) Nazghelichi et al. (2011)
Q̇ev = hfg × ṁw (15) Zohrabi et al. (2020a)
hfg = 2.503 × 106 - 2. 386 × 103 (T- 273. 16) (16) Ghasemkhani et al. (2016)
273.16 ≤ T(K) ≤ 338.72
0.5
hfg = (7, .33, ×, 1012 , −, 1.60, ×, 107 , T 2 )
338.72 ≤ T(K) ≤ 533.16
×Pvs
ṁwin = 0.622 × P−P vs 6887
 T
 (17) Nazghelichi et al. (2011)
Pvs = 0.1 exp 27.014 − T − 5.3 ln 273.16 (18) Mondal et al. (2020)
ṁv
ṁwout = ṁwin + ṁa
(19) Nazghelichi et al. (2010)
wt −wt+t
ṁv = t (20) Liu et al. (2019)

Fig. 2 – Schematic diagram of hot air- rotary drum drying chamber with input and output parameters.

Using the second law of thermodynamics, the values of the In this study, the relative humidity of the environment and
•
exergy rate of the air entering the dryer chamber Exin (J/s), temperature of fresh air was considered as = 20% and T = 20
• ◦ C.
exergy rate of the air leaving the dryer chamber Exout (J/s),
• The average changes in inlet air temperature to the dryer
evaporation exergy Exeva (kJ/kg) and exergy loss rate in the
• and the relative humidity during the drying process are shown
system Exloss (J/s) are calculated by Eqs. (24–27), respectively in Table 2.
(Ghasemkhani et al., 2016; Manalu et al., 2016). The maximum improvement in the exergy efficiency of a
• •
∞
system or process is achieved when the exergy loss or irre-
Exin = ma [Ca + ωa Cv ] (Ta − T )
versibility is minimal. The exergy improvement potential for
     the process of drying the samples was calculated using Eq. (29)
Ta Pa
−T∞ [Ca + ωa Cv ] ln − (Ra + ωa Rv ) ln (24) (Sheikhshoaei et al., 2019; Afzali et al., 2019).
T∞ P∞
    
1 + 1.6078ω∞ ωa • •
+T∞ [R∞ + ωa Rv ] ln + 1.6078ωi R∞ ln IP = (1 − − Exout )
1 + 1.6078ωa ω∞ Ex )( Exin (29)

• •
∞ 2.5. ANFIS
Exo = ma [Ca + ωin Cv ] (Tout − T )

    
Tout Pout By combining the fuzzy inference systems and artificial neu-
−T∞ [Ca + ωout Cv ] ln − (Ra + ωo Rv ) ln (25)
T∞ P∞
ral network learning algorithms, the ANFIS system is able to
     effectively model and analyze the relationship between the
1 + 1.6078ω∞ ωout
+T∞ [R∞ + ωout Rv ] ln + 1.6078ωo R∞ ln
1 + 1.6078ωout ω∞ input and output variables from a given dataset. The rela-
tionship between inputs and outputs was constructed using
• T∞ • the both input-output data and fuzzy if-then rules. ANFIS is
Exeva = 1− Q ev (26)
Tp a neuro-fuzzy technique which is the result of the integra-
tion between neural network and fuzzy inference system (FIS)
• • •
(Jahanbakhshi et al., 2020).
Exloss = Exin − Exout (27)
The structure of ANFIS model is shown in Fig. 3 with three
Exergy efficiency ( Ex ) was also defined as the ratio of the inputs (inlet air temperature, rotary rotation speed and dry-
exergy leaving to the exergy entering the drying chamber and ing time) and one output (moisture ratio, energy utilization,
was calculated using Eq. (28) (Darvishi et al., 2018; Peinado energy utilization ratio, exergy loss, or exergy efficiency). The
et al., 2011). main limitation in the development of an adaptive neuro-
fuzzy inference system is related to the number of inputs. If
• • • the number of inputs is more than five, then the calculation
Exin − Exloss Exloss
Ex = • =1− • (28) time and the number of rules will increase and the model will
Exin Exin possibly fail. In a standard structure, two, three, four, and at
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 173

Table 2 – Average values of inlet air temperature and relative humidity under different drying conditions.
Rotation speed (rpm) Air temperature (◦ C) Inlet air temperature (◦ C) Relative humidity (g water vapor/kg dry air)

5 40 36.39 ± 0.75 33.21 ± 0.11

55 50.47 ± 1.03 33.24 ± 0.14
70 64.48 ± 1.28 33.30 ± 0.16
10 40 36.61 ± 0.72 33.18 ± 0.15
55 50.75 ± 1.09 33.23 ± 0.18
70 64.63 ± 1.32 33.28 ± 0.12
15 40 36.73 ± 0.68 33.17 ± 0.13
55 50.73 ± 1.33 33.22 ± 0.17
70 64.77 ± 1.41 33.27 ± 0.19

Fig. 3 – ANFIS structure.

most five input variables enter into the network and a suitable where w is the rate of normalization.
output is derived (Khoshnevisan et al., 2015). Fourth layer (Defuzzification): Each wi is the values related
For example, the if-then fuzzy rules can be expressed in to the result of linear regression fi in this function layer and
the following forms (Kaveh et al., 2018): finally, the output result is generated in Eq. (38) (Ojediran et al.,
Rule 1 : ifT isA1 , Ub isB1 , andt isC1 thenf1 = p1 T + q1 Ub + r1 t + u1 (30) 2020).

Rule 1 : ifT isA2 , Ub isB2 , andt isC2 thenf2 = p2 T + q2 Ub + r2 t + u2 (31)

O4i = wi fi = wi (pi T + qi Ub + ri t + ui )i = 1, 2 (38)
First layer (Fuzzification): converts the input variables into
a fuzzy set through the member functions. Each node in layer Fifth layer (Summation): Each output rule is added to the
1 is an adaptive node, and the function of the corresponding output layer. The total output can be calculated by Eq. (36) as
node is written as follows (Ojediran et al., 2020): follows (Kaveh et al., 2018):

O1Ai = Ai (T)i = 1, 2 (32)  wi fi

i
O5i = wi fi = i = 1, 2 (39)
i
wi
O1Bj = Bj (Ub )j = 1, 2 (33) i

In this paper, one model of Adaptive Neuro-Fuzzy Infer-

O1Ai = Ci (t)i = 1, 2 (34)
ence System (ANFIS) was developed using 471 datasets to
predict the moisture content, energy utilization, energy uti-
O1Bj = Dj (PT)j = 1, 2 (35)
lization ratio, exergy loss and exergy efficiency. In order to
Second layer (Multiplication): Each rule is a node in ANFIS model the ANFIS, the data were randomly assigned to two
using the soft-min or product to determine the binding factor groups with 70% (330 data) for the training and 30% (141 data)
of the rule. The input signal in Eq. (36) is multiplied by this for the model evaluation. Neural Network Toolbox MATLAB
layer and removes the product (Armaghani et al., 2015). (ver4.01) software was also used for the analyses.
Basically, the network speed and accuracy are reduced if
O2i = wi = Ai (T) × Bi (Ub ) × Ci (t)j = 1, 2 (36) the data are not pre-processed. To avoid such situation and to
unify the data values for the network, the normalization oper-
Third layer (Normalization): Each wi is a small scale of wi in ation is performed, which will prevent the too small amounts
this layer. The ith ratio of the weight rules governing the sum of the weights. Eq. (40) was utilized to normalize the used data
of the forces of each base can be calculated by Eq. (37): (Armaghani et al., 2015):

wi Xi − Xmin
O3i = w = i = 1, 2 (37) Xn = (40)
w1 + w2 Xmax − Xmin
174 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

Many quantitative indicators were used to evaluate the per- lower the product moisture (Motevali et al., 2014). In drying the
formance of the networks used and the power of the model to green pea in a combined infrared–hot air-rotary drum dryer,
predict the output parameters (moisture ratio, energy utiliza- Kaveh and Abbaspour-Gilandeh (2020) showed that the drying
tion, energy utilization ratio, exergy loss and exergy efficiency). speed increases with the increasing rotary rotation speed. In
The criteria such as the coefficient of determination (correla- this way, the lowest and highest drying times were obtained
tion) (R2 ), mean squared error (MSE) and mean absolute error at the rotation speeds of 5 and 15 rpm, respectively. Similar
(MAE) were used in this study to determine the quality of net- results have been reported by Parizi et al., 2016 for the drying
work performance (Khanali and Rafiee, 2014). of pistachio and by Kaensup et al. (2002) for the drying of chilli.


N
2
3.2. Quality properties
(Yi − YO )

R2 = 1 −
i=1
 (41) 3.2.1. Analysis of variance results

N
2
The results of the analysis of variance (ANOVA) obtained by
(Yi − Ymean )
Duncan test (using SAS software. 9.1) are presented in Table 3
i=1
for thermal properties, quality and nutritional features of the
green pea dried by the proposed dryer.
1
N
2
MSE = (Yi − YO ) (42)
N 3.2.2. Color
i=1
Table 3 provides a comparison of the means values of the color
change ( E). As it can be seen, the lowest E was obtained at
1
N
MAE = |Yi − YO | (43) the inlet air temperature of 70 ◦ C and drum rotation speed of 15
N
i=1 rpm (p < 0.01). By enhancing the temperature and drum rota-
tion speed, the E of the sample was declined (from 41.81 to
2.6. Statistical analysis 64.95 for all the experiments) (Table 4). Variation of E showed
that by increasing temperature and drum rotation speed, the
SAS 9.1 software was used for statistical analysis of the data. color variation of the samples declined in comparison with
Duncan’s multifactorial test was also employed for comparing the initial samples.
the mean of main effect levels and the interaction between the Zielinska et al. (2013b) investigated the quality indices
variables. of green peas dried by different methods. They also found
that as drying time by hot air dryer is very long, chloro-
3. Results and discussion phyll may be degraded to undesirable gray compounds such
as pheophorbide or pheophytin and tend to convert to color-
3.1. Drying kinetic less compounds. In a study by Kaveh and Abbaspour-Gilandeh
(2020) on the color of green pea dried by a hybrid hot air-
Fig. 4 shows the variations in the moisture ratio of green pea convective- infrared dryer, the color variation of the samples
against the drying time during the combined hot air-rotary decreased by the increase of the infrared power, air tempera-
drum drying process under the test conditions. As it can be ture and drum rotation. Horuz and Maskan (2015) examined
seen, increasing the temperature of the air entering the dryer the color variation of pomegranate dried by hot air dryer, they
reduced the process time. At higher temperatures, the rate showed that the color variation will be declined by increasing
of heat transfer from the heat source to the crop increased, the air temperature.
which increased the rate of moisture evaporation and reduced
the process time. In addition, increasing the temperature also 3.2.3. Shrinkage
increased the rate of moisture transfer from the tissue to The study of the shrinkage phenomenon of agricultural prod-
the surface of the crop and increased the evaporation, thus ucts is necessary to better understand the drying process and
increasing the drying speed at high temperatures. Similar control quality. Shrinkage occurs during the drying process at
results have been reported about the effect of temperature on the same time as the removal of moisture and therefore can
the drying time of other agricultural and food products such as affect the drying speed. Moisture is trapped in dense cells and
apple (Ghasemkhani et al., 2016), quince (Abbaspour-Gilandeh is prevented from moving to the outer surface for subsequent
et al., 2020), poplar wood chips (Zohrabi et al., 2020b), maize removal (Wang et al., 2018). Table 3 reveals the ANOVA results
grain (Mondal et al., 2020) and Pistacia Atlantica (Kaveh et al., about the impact of inlet air temperature and drum rotation
2020). on the shrinkage (Sb ) level in the dried green peas. According
From Fig. 4, it can be seen that the maximum slope of the to this table, the effects of air temperature and drum rota-
moisture ratio line was related to the use of 70 ◦ C temperature tion speed on the shrinkage were significant at the level of 1%.
and rotation speed of 15 rpm and the lowest slope change was According to the Table 4, the highest Sb (61.83%) was recorded
related to the temperature of 40 ◦ C and rotation speed of 5 at inlet air temperature of (40 ◦ C) and drum rotation speed of 5
rpm. The reason for the results was that in the dry samples rpm; while the lowest Sb (39.19%) was observed at air tempera-
with higher rotary speeds, the change in moisture transfer rate ture and drum rotations peed of 70 ◦ C and 15 rpm, respectively
from the center of the sample to the surface increased com- (p < 0.01).
pared to the samples dried with lower rotary speeds. Fig. 4 also Table 4 shows the Sb for the green peas dried by the pro-
showed that with an increasing rotary rotation speed, the crop posed hot air- rotary drum dryer. As the results suggest, Sb was
underwent more friction under the heat, and thus, the rate decreased at higher temperatures and rotary speed. Green pea
of moisture removal from the green pea was faster and the has high initial moisture. During the drying process, signifi-
drying time was lower. In other words, the higher the rotary cant Sb can alter its initial structure. Green pea texture has
rotation speed, the more the mass and heat transfer and the turgor pressure (i.e. the liquid content imposes pressure on
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 175

Fig. 4 – Variation of moisture ratio with temperature at different rotary rotation speeds.

Table 3 – Variance analyses for the means of variables and their interaction for green pea drying in hot air-rotary drum
dryer.
S.O.V DF Sb RC E Exergy efficiency Exergy improvement potential

Temperature 2 534.87** 0.714** 637.39** 0.090** 4.67e-5**

Rotation speed 2 115.34** 0.087** 126.65** 0.012** 1.31e-5**
Temperature* Rotation speed 4 0.508 ns 0.004* 2.049** 0.0004* 9.85e-5 ns
Error 18 7.30 0.003 1.46 0.0002 9.85e-5

∗∗
Significant at 1% probability level.
∗
Significant at 5% probability level.
ns
not significant.

Table 4 – Quality parameters of green pea dried at different drying conditions.

Rotation speed (rpm) Temperature (◦ C) E RC Shrinkage (%)

5 40 64.95 ± 1.67g 1.10 ± 0.03a 61.83 ± 1.87h

5 55 57.34 ± 0.99e 1.42 ± 0.05d 52.42 ± 1.44ef
5 70 48.86 ± 1.06c 1.62 ± 0.03f 46.02 ± 1.56c
10 40 62.57 ± 1.24f 1.19 ± 0.04b 57.43 ± 1.23g
10 55 52.64 ± 1.02d 1.49 ± 0.05de 50.72 ± 1.71de
10 70 45.19 ± 1.12b 1.77 ± 0.07g 42.61 ± 1.49b
15 40 58.73 ± 1.33e 1.30 ± 0.04c 54.58 ± 1.44f
15 55 48.11 ± 1.24c 1.54 ± 0.07ef 48.44 ± 1.22cd
15 70 41.81 ± 1.08a 1.89 ± 0.06h 39.19 ± 1.05a
176 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

the cell wall and the liquid inside the cell is under pressure) rate. With an increasing air temperature and rotary rotation
(Nguyen et al., 2018). During the drying process, water elim- speed of the dryer, the energy utilization increased, as the
ination from the cell will reduce the stress imposed by the higher temperatures and rotation speeds further reduced the
liquid on the cell wall which in turn may create Sb on the moisture content. In other words, the higher temperature and
product texture. High temperature and drum rotation speed, rotary rotation speeds caused more heat and mass trans-
however, will accelerate the water evaporation leaving no time fer, resulting in more moisture removal from the green pea
for the deformation (Dehghannya et al., 2018a). Their results (Abbaspour-Gilandeh et al., 2020; Nazghelichi et al., 2010).
indicated that an increase in the inlet air temperature will
decrease the shrinkage. Aydogdu et al. (2015); Dai et al. (2019), 3.4. Energy utilization ratio
and Dehghannya et al. (2018b) reported similar results for dry-
ing eggplant, apple, and apple, respectively. The energy utilization ratio for drying the green pea was cal-
culated using Eq. (21) and its values were found to be in the
3.2.4. Rehydration range of 0.035–0.5843. Fig. 6 shows the variations in the energy
According to Table 3, ANOVA results of RC at different temper- utilization ratio with temperature and rotary rotation speed.
atures and drum rotation speeds show that these parameters According to the results, at each rotary speed, the energy uti-
had significantly impacts on the RC at 1%. lization ratio increased when the temperature increased from
As shown in Table 4, RC will be increased by the increase of 40 to 70 ◦ C. This finding was consistent with the observations
the air temperature and drum rotations speed. This could be reported by Motevali and Minaei (2012). Azadbakht et al. (2017)
attributed to the destruction of the cell tissue by higher tem- also reported that using a fluidized bed dryer, the energy uti-
peratures and rotation speeds that can enhance the RC and lization ratio in the process of drying the potato slices varied
hence entrapment of the water in the spaces created by the from 0.37 to 0.716. They stated that the energy utilization ratio
damaged cells (Vega-Gálvez et al., 2012). Also, high tempera- increased with increasing the temperature. Erbay and Icier
ture increases the molecular diffusion rate, hence enhancing (2011) dried the olive leaves in a tray dryer under different
the imbibe capacity of green peas. In addition, high rehydra- conditions. The highest and lowest values of energy utiliza-
tion temperature contributes to the expansion of green pea tion ratio were 0.2621 and 0.1405, respectively. The increase in
tissue and space density, which also improves the expansion air temperature led to the increase in energy utilization ratio.
of capillaries with enhanced hydrophilic properties (Wang As the air temperature rises, the air enthalpy increases,
et al., 2018). According to the results obtained by Aydogdu et al. and as a result, the drying rate increases. Increasing the tem-
(2015); Dehghannya et al. (2018b) and Wang et al. (2018) the ele- perature increases the energy utilization and enthalpy of the
vation of the drying temperature will enhance the RC which inlet air. Increasing the enthalpy of the inlet air also increases
could be due to accelerated moisture evaporation and hence the transfer of mass and heat. As a result, the energy uti-
decline of samples shrinkage. lization and energy utilization ratio increase (Akpinar, 2005;
Akpinar et al., 2005). The energy utilization ratio was at its
3.3. Energy utilization highest value at the beginning of the process. Over time, the
energy utilization ratio steadily decreased and reached its
Using the first law of thermodynamics and the information lowest value at the end of the process. This was because
obtained from the experiments, the energy analysis of the at the beginning of the process, the moisture content of
green pea drying process was performed. The values of energy the crop and also the energy absorption were high. As the
utilization rate for different drying conditions were calculated process continued and the moisture content decreased, the
using Eq. (8) and the obtained results are shown in Fig. 5. energy utilization ratio decreased. The surface moisture also
The energy utilization rate in the process of drying the green evaporated faster due to the higher mass and heat transfer
pea in different temperature conditions and rotary rotation coefficients (Corzo et al., 2008; Dinçer and Zamfirescu, 2015).
speeds varied from 0.012 to 0.1556 kJ/s. These obtained values Nazghelichi et al. (2010) confirmed these findings in the car-
were comparable to the results reported by other researchers. rot drying process, while similar conclusions were obtained by
Corzo et al. (2008) reported the energy utilization rate in the Abbaspour-Gilandeh et al. (2020) in the drying of quince slices,
process of drying a thin layer of coroba slices at different and Motevali and Minaei (2012) in the drying of pomegranate
inlet air velocities and temperatures of 71, 82 and 93 ◦ C to arils and Yogendrasasidhar and Setty (2018) in the drying of
be between 0.009 and 0.65 kJ/s. The energy utilization rate kodo millet grains and Fenugreek seeds.
of drying the zedoary slices at different moisture contents According to the results of this study, it was observed that
and temperatures of 40, 50, 60 and 70 ◦ C was reported to at constant air temperature, the increase in rotary rotation
range from 0.001 to 0.121 kJ/s (Manalu et al., 2016). Abbaspour- speed led to the increase in energy utilization ratio. With an
Gilandeh et al. (2020) also obtained the energy utilization rates increasing rotary rotation speed, less time was required for
in the hot-air drying of quince slices under different condi- the green pea to dry, as the difference in the partial pressure
tions in the range of 0.009 to 0.0694 kJ/s. The results of this of water vapor between the crop and the air passing through
study showed that with increasing the temperature, the rate the dryer increased, resulting in a faster transfer of moisture
of energy utilization increased, which was consistent with the from the crop to the dryer and the earlier drying of the crop.
findings reported by other researchers, including Nazghelichi Eventually, the energy utilization ratio increased as the rotary
et al. (2010); Yogendrasasidhar and Setty (2018) and Aghbashlo rotation speed increased and, in other words, as the mass and
et al. (2009) Also, with increasing the rotary rotation speed, the heat transfer increased.
energy utilization rate increased from 5 to 15 rpm.
The energy utilization was high at the beginning of the 3.5. Exergy loss
experiment due to the high rate of moisture removal from
the green pea, and it decreased rapidly due to the complete Fig. 7 shows the results obtained from the analysis of exergy
removal of the remaining moisture which occurred at a slower loss in the green pea drying process in different temperature
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 177

Fig. 5 – The changes of energy utilization during green pea drying process.

and rotary rotation speed conditions under the constant veloc- chamber was reduced, resulting in less evaporation (less mois-
ity of the air entering the dryer chamber. As it can be seen, ture leads to less absorption of exergy) and consequently, less
the amount of lost exergy varied from 0.02 to 0.11 kJ/s. Sim- exergy utilization and less exergy loss (Nazghelichi et al., 2010;
ilar values were also obtained for coroba slices in the range Darvishi et al., 2018).
of 0.005–0.010 kJ/s (Corzo et al., 2008), for kodo millet in the A small amount of the exergy provided in the drying cham-
range of 0.002–0.015 kJ/s, for fenugreek seeds in the range ber was dissipated and the energy was available in the outlet
of 0.002–0.013 kJ/s (Yogendrasasidhar and Setty, 2018), and air. This indicated that the exergy loss in the dryer chamber
for rough rice in the range of 0.062–0.175 kJ/s (Khanali et al., was higher at higher temperatures and rotary rotation speeds,
2013). Thus, the findings of this study were consistent with while the exergy loss in the dryer chamber was lower at lower
their results. The results showed that the rate of exergy loss temperatures and rotary rotation speeds.
increased with increasing the temperature and rotary rota-
tion speed. The higher the drying temperature, the higher the
3.6. Exergy efficiency
exergy, which increased the water evaporation or exergy con-
sumption, thus increasing the exergy loss (Nazghelichi et al.,
The average exergy efficiency of the dried green peas with dif-
2010). Liu et al. (2019); Darvishi et al. (2018), and Aktas et al.
ferent temperatures is shown in Fig. 8, with the rotation speeds
(2017) also found that with increasing the temperature, the
of 5–15 rpm. The average values of exergy efficiency ranged
exergy loss increased.
from 0.5434 at the temperature of 40 ◦ C and rotation speed
As shown in Fig. 7, as the drying time elapsed, the exergy
of 5 rpm to 0.8382 at the temperature of 70 ◦ C and rotation
loss of the dryer was continuously reduced. This can be
speed of 15 rpm. This was consistent with the results reported
explained by the fact that the difference between the inlet and
in the literature for the process of drying various crops such
outlet temperatures of the dryer chamber initially increased,
as carrot slices (Aghbashlo et al., 2009), kodo millet grains
which caused more water to evaporate in the green pea and
(Yogendrasasidhar and Setty, 2018) and rough rice (Khanali
more exergy utilization and more exergy loss. Then, the dif-
et al., 2013). Increasing the drying temperature increased the
ference between the inlet and outlet temperature of the dryer
exergy loss in the dryer chamber, but this value was less than
178 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

Fig. 6 – Energy utilization ratio at different drying temperatures and at rotary rotation speeds.

the increase in the input exergy. Therefore, as the dryer tem- 3.7. Exergy improvement potential
perature increased, the exergy efficiency increased according
to Eq. (28). The average exergy improvement potential was calculated
Khanali and Rafiee (2014) also showed that with an increas- using Eq. (29) and the obtained values are shown in Table 5.
ing inlet air temperature from 40 to 70 ◦ C, the exergy efficiency It can be seen that the minimum average value of the exergy
of rough rice drying increased. In another study, Aviara et al. improvement potential was 0.0089 kJ/s (at the air temperature
(2014) studied the effect of temperature on the exergy effi- of 40 ◦ C and rotary rotation speed of 5 rpm) and the maximum
ciency of native cassava starch in a tray dryer and showed value was 0.0157 kJ/s (at the air temperature of 70 ◦ C and rotary
that increasing the temperature from 40 to 60 ◦ C increased rotation speed of 15 rpm). Also, it was observed that increas-
the exergy efficiency. As mentioned earlier, the exergy loss ing the temperature and rotary rotation speed increased the
increased with an increasing rotary rotation speed, and as a exergy improvement potential. This increase in the exergy
result, with increasing the rotary rotation speed, the exergy improvement potential may be attributed to the fact that the
efficiency increased. Similarly, with an increasing air temper- higher drying temperature can improve the enthalpy of the
ature and rotation speed, the exergy efficiency would increase. inlet air. The results were comparable to the observations
This increase in exergy efficiency can be attributed to the high reported by the researchers. Aghbashlo et al. (2012) reported
rate of heating and evaporation of moisture from the mate- that when drying at a drying temperature of 140–160 ◦ C during
rial at high temperatures and rotation speeds. In the drying of the drying of fish oil in a spray dryer, the exergy improve-
apple in a rotating-tray air dryer in the inlet air temperature ment potential increased from 0.08141 to 0.12571 kJ/s. Aviara
and rotary rotation speed, Ghasemkhani et al. (2016) showed et al. (2014) calculated the average rate of exergy improvement
that with an increasing temperature and rotary rotation speed, potential for the process with a tray dryer on the drying of
the exergy efficiency increased from 0.23 to 0.961. The value of native cassava starch was in the range of 0.2–1.6 kJ/s. They
the exergy efficiency in the drying of poplar wood chips was also reported that this value increased when the drying tem-
obtained in the range from 0.465 to 0.759 in a hot-air dryer perature increased from 40 to 65 ◦ C. Ghasemkhani et al. (2016)
with the return air cycle (Zohrabi et al., 2020b). dried the apple slices using the rotating-tray air dryer method,
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 179

Fig. 7 – Variation of exergy losses in the drying chamber with drying time under different temperature and rotary rotation
speed.

Fig. 8 – Effect of drying variables on the average exergy efficiency of drying chamber.

Table 5 – Variation of average exergy improvement potential rate at drying temperature and rotation speed.
Drying air temperature (◦ C) Average exergetic improvement potential rate (kJ/s)

Rotary rotation speed (rpm)

5 10 15

40 0.0089 ± 0.0002a 0.0099 ± 0.0003b 0.0112 ± 0.0002c

55 0.0108 ± 0.0003c 0.0119 ± 0.0002d 0.0132 ± 0.0003e
70 0.0131 ± 0.0003e 0.0148 ± 0.0003f 0.0157 ± 0.0003g
180 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

Fig. 9 – Comparison of predicted and desired output values for the kinetic, energy utilization, energy utilization ratio,
exergy loss and exergy efficiency, using selected ANFIS.

and observed that increasing the temperature from 50 to 65 ◦ C In this study, the effect of the main drying parameters on
increased the exergy improvement potential and reduced this the energy and exergy was examined and it was demonstrated
parameter from 65 to 80 ◦ C. They also reported that increas- that the change in these parameters affected the dryer perfor-
ing the rotary rotation speed (from 0 to 12 rpm) at different mance. For example, the green pea with higher temperature
temperatures reduced the exergy improvement potential. and rotary rotation speed increased the energy utilization and
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 181

Table 6 – Statistical parameters related to Kinetic, energy and exergy analysis using ANFIS.
R2 MSE MAE Learning method Number of MF Type of MF Parameters

Cycle Input Output Input

0.9996 0.0026 0.045 Hybrid 1000 3-3-3 Linear Gaussmf Moisture ratio (MR)
0.9999 0.0003 0.0085 Hybrid 1000 3-3-3 Linear Gaussmf Energy utilization (Eu)
0.9995 0.0049 0.072 Hybrid 1500 3-3-3 Linear Trimf Energy utilization ratio (EUR)
0.9989 0.0082 0.128 Hybrid 1500 3-5-3 Linear Trimf Exergy loss
0.9996 0.0018 0.021 Hybrid 1200 3-3-3 Linear Trimf Exergy efficiency

the optimal part of the used exergy called exergy loss. As men- neural network (ANN). They showed that the ANFIS model per-
tioned earlier, increasing the energy utilization and exergy loss formed better with R2 = 0.9997 than the ANN with R2 = 0.9995.
led to the increase in energy utilization ratio and the decrease Other researchers, including Ojediran et al. (2020); Ziaforoughi
in exergy efficiency. Therefore, the entry of energy and exergy et al. (2016) and Kaveh et al. (2018) used the ANFIS model in
into the dryer chamber can be reduced by increasing the the drying process, and the optimal performance of the ANFIS
energy utilization ratio and reducing the exergy efficiency. model for the modeling has been shown in these studies.

3.8. Adaptive neuro-fuzzy inference system (ANFIS) 4. Conclusions

In order to achieve the best ANFIS model for predicting the The aim of this study was to investigate the effect of the new
amounts of moisture ratio, energy utilization, energy utiliza- dryer (hot air- rotary drum) on the qualitative (color, shrinkage
tion ratio, exergy loss, and exergy efficiency of dried green and rehydration ratio) and thermodynamic (moisture ratio,
pea, some changes were made in various parameters such energy utilization, energy utilization ratio, exergy loss, and
as the number (3–5) and type of input (psigmf, dsigmf, pimf, exergy efficiency) properties of green peas at different air tem-
Gasuss2mf, Gaussmf, gbellmf, trimf, and trapmf) and output peratures and rotation speeds. The thermodynamic properties
membership functions (Linear), algorithm methods (Hybrid were predicted using the ANFIS model. It was found that the
and Back-propagation) and number of epochs (500–2000). rate of energy utilization and energy utilization ratio increased
Finally, the best ANFIS structure, including the number of with increasing the temperature and rotary rotation speed in
membership functions for each input, number of rules, num- the drying chamber, while the drying time decreased with
ber of epochs, and type of input and output membership increasing the rotary rotation speed and air temperature. The
functions are shown in Table 6. The correlation coefficients increase in temperature and rotary rotation speed led to the
(R2 ) of the desired output values calculated in the laboratory increase in exergy loss. The average exergy efficiency of the
and the predicted values in this study are given in Fig. 9. The green pea drying process varied from 0.5434 at the air temper-
high correlation coefficients between the laboratory results ature of 40 ◦ C and rotation speed of 5 rpm to 0.8382 at the air
and the model outputs indicated the acceptable accuracy and temperature of 70 ◦ C and rotation speed of 15 rpm. The mini-
applicability of this model in controlling the industrial pro- mum average value of exergy improvement potential at the air
cesses. temperature of 40 ◦ C and rotation speed of 5 rpm was obtained
According to Table 6, the values of R2 , MSE and MAE as 0.0089 kJ/s and its maximum value at the air temperature
indices related to the modeling of the moisture ratio for dry- of 70 ◦ C and rotary rotation speed of 15 rpm as 0.0157 kJ/s.
ing the green pea were 0.9996, 0.0026 and 0.045, respectively, The ANFIS model had shown a good performance in predict-
and for the energy utilization, they were 0.9999, 0.0003 and ing the parameters (moisture ratio, energy utilization, energy
0.0085, respectively. The best ANFIS models for predicting the utilization ratio, exergy loss, and exergy efficiency), with the R2
moisture ratio and energy utilization with the type of input values reaching above 0.99 and had an optimal mean squared
membership function, type of output membership function, error. The model could be used for the better control of indus-
type of learning algorithm, number of membership func- trial processes, reduction in cost and time, and improvement
tions for each input, and number of epochs were obtained as of the final product.
Gaussmf, Linear, Hybrid, 3-3-3 and 1000, respectively.
The best performance of the ANFIS model was obtained by Conflict of interest
predicting the energy utilization ratio, exergy loss and exergy
efficiency with the type of input membership function (Trimf), None declared.
type of output membership function (Linear), type of learning
algorithm (Hybrid) and number of epochs (1500). The R2 , MSE References
and MAE of the best network structure for the energy utiliza-
tion ratios were 0.9995, 0.0049 and 0.072, for the exergy loss Abbaspour-Gilandeh, Y., Jahanbakhshi, A., Kaveh, M., 2020.
were 0.9989, 0.0082 and 0.128, and for the exergy efficiency Prediction kinetic, energy and exergy of quince under hot air
were 0.9996, 0.0018 and 0.021, respectively. dryer using ANNs and ANFIS. Food Sci. Nutr. 8 (1), 594–611.
Abbaspour-Gilandeh et al. (2020) reported that the adap- Afzali, F., Darvishi, H., Behroozi-Khazaei, N., 2019. Optimizing
tive neuro-fuzzy inference system (ANFIS) model has a greater exergetic performance of a continuous conveyor infrared-hot
air dryer with air recycling system. Appl. Ther. Eng. 154,
ability than the neural network model to predict the energy
358–367.
and exergy parameters of quince in the hot-air dryers. In
Aghbashlo, M., Kianmehr, M.H., Arabhosseini, A., 2008. Energy
another study, Jahanbakhshi et al. (2020) modeled the mois- and exergy analyses of thin-layer drying of potato slices in a
ture ratio for drying the pistachio in a microwave dryer with an semi-industrial continuous band dryer. Dry. Technol. 26,
ultrasound pretreatment dryer using the ANFIS and artificial 1501–1508.
182 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183

Aghbashlo, M., Kianmehr, M.H., Arabhosseini, A., 2009. Dehghannya, J., Bozorghi, S., Heshmati, M.K., 2018a. Low
Performance analysis of drying of carrot slices in a temperature hot air drying of potato cubes subjected to
semi-industrial continuous band dryer. J. Food Eng. 91, 99–108. osmotic dehydration and intermittent microwave: drying
Aghbashlo, M., Mobli, H., Rafiee, S., Madadlou, A., 2012. Energy kinetics, energy consumption and product quality indexes.
and exergy analyses of the spray drying process of fish oil Heat Mass Transf. 54 (4), 929–954.
microencapsulation. Biosyst. Eng. 111, 229–241. Dehghannya, J., Farshad, P., Heshmati, M.K., 2018b. Three-stage
Aghbashlo, M., Mobli, H., Rafiee, S., Madadlou, A., 2013. A review hybrid osmotic–intermittent microwave–convective drying of
on exergy analysis of drying processes and systems. Renew. apple at low temperature and short time. Dry. Technol. 36 (16),
Sustain. Energy Rev. 22, 1–22. 1982–2005.
Aghbashlo, M., Hosseinpour, S., Mujumdar, A.S., 2015. Application Dinçer, I., Zamfirescu, C., 2015. Energy and exergy analyses of
of artificial neural networks (ANNs) in drying technology—a drying processes and systems. Chapter 4. In: Drying
comprehensive review. Dry. Technol. 33 (12), 1397–1462. Phenomena: Theory and Applications.,
Akpinar, E.K., 2004. Energy and exergy analysis of drying of red http://dx.doi.org/10.1002/9781118534892, ch4.
pepper slices in a convective try dryer. Int. Commun. Heat Erbay, Z., Icier, F., 2011. Energy and exergy analyses on drying of
Mass Transf. 31 (8), 1165–1176. olive leaves (Olea europaea L.) in tray drier. J. Food Process Eng.
Akpinar, E.K., 2005. Energy and exergy analyses of drying of 34, 2105–2123.
eggplant slice in a cyclone type dryer. J. Mech. Sci. Technol. 19 Ghasemkhani, H., Keyhani, A., Aghbashlo, M., Rafiee, S.,
(2), 692–703. Mujumdar, A.S., 2016. Improving exergetic performance
Akpinar, E.K., Midilli, A., Bicer, Y., 2005. Energy and exergy of parameters of a rotating-tray air dryer via a simple heat
potato drying process via cyclone type dryer. Energy Convers. exchanger. Appl. Ther. Eng. 94, 13–23.
Manage. 46, 2530–2552. Horuz, E., Maskan, M., 2015. Hot air and microwave drying of
Aktas, M., Khanlari, A., Amini, A., Sevik, S., 2017. Performance pomegranate (Punica granatum L.) arils. J. Food Sci. Technol. 52
analysis of heat pump and infrared–heat pump drying of (1), 285–293.
grated carrot using energy-exergy methodology. Energy Jahanbakhshi, A., Kaveh, M., Taghinezhad, E., Sharabiani, V.R.,
Convers. Manage. 132, 327–338. 2020. Assessment of kinetics, effective moisture diffusivity,
Armaghani, D.J., Mohamad, E.T., Momeni, E., Narayanasamy, M.S., specific energy consumption, shrinkage, and color in the
Amin, M.F.M., 2015. An adaptive neuro-fuzzy inference pistachio kernel drying process in microwave drying with
system for predicting unconfined compressive strength and ultrasonic pretreatment. J. Food Process. Preserv.,
Young’s modulus: a study on Main Range granite. Bull. Eng. e14449.
Geol. Environ. 74, 1301–1319. Johnson, P.W., Langrish, T.A.G., 2018. Exergy analysis of a spray
Aviara, N.A., Onuoha, L.N., Falola, O.E., Igbeka, J.C., 2014. Energy dryer: methods and interpretations. Dry. Technol. 36 (5),
and exergy analyses of native cassava starch drying in a tray 578–596.
dryer. Energy 73, 809–817. Kaensup, W., Chutima, S., Wongwises, S., 2002. Experimental
Aydogdu, A., Sumnu, G., Sahin, S., 2015. Effects of study on drying of chilli in a combined
microwave-infrared combination drying on quality of microwave-vacuum-rotary drum dryer. Dry. Technol. 20 (10),
eggplants. Food Bioprocess Technol. 8 (6), 2067–2079.
1198–1210. Karthikeyan, A.K., Murugavelh, S., 2018. Thin layer drying
Azadbakht, M., Aghili, H., Ziaratban, A., Vahedi Torshizi, M., 2017. kinetics and exergy analysis of turmeric (Curcuma longa) in a
Application of artificial neural network method to exergy and mixed mode forced convection solar tunnel dryer. Renew.
energy analyses of fluidized bed dryer for potato cubes. Energy 128, 305–312.
Energy 120, 947–958. Kaveh, K., Abbaspour-Gilandeh, Y., 2020. Impacts of hybrid
Azadbakht, M., Torshizi, M.V., Noshad, F., Rokhbin, A., 2018. (convective-infrared-rotary drum) drying on the quality
Application of artificial neural network method for prediction attributes of green pea. J. Food Process Eng. 43 (7), e13424,
of osmotic pretreatment based on the energy and exergy http://dx.doi.org/10.1111/jfpe.13424.
analyses in microwave drying of orange slices. Energy 165, Kaveh, M., Jahanbakhshi, A., Abbaspour-Gilandeh, Y.,
836–845. Taghinezhad, E., Moghimi, M.B.F., 2018. The effect of
Bundschuh, J., Chen, G. (Eds.), 2014. Sustainable Energy Solutions ultrasound pre-treatment on quality, drying, and
in Agriculture. CRC Press, Taylor & Francis Books, 480 pages. thermodynamic attributes of almond kernel under convective
Castro, M., Román, C., Echegaray, M., Mazza, G., Rodriguez, R., dryer using ANNs and ANFIS network. J. Food Process Eng. 41,
2018. Exergy analyses of onion drying by convection: influence e12868.
of dryer parameters on performance. Entropy 20, 310, Kaveh, M., Amiri Chayjan, R., Taghinezhad, E., Sharabiani, V.R.,
http://dx.doi.org/10.3390/e20050310. Motevali, A., 2020. Evaluation of specific energy consumption
Chen, G., 2015. Drying Operations: Agricultural and Forestry and GHG emissions for different drying methods (case study:
Products. In: Capehart, B. (Ed.), Encyclopedia of Energy Pistacia atlantica). J. Clean. Prod. 259, 120963.
Engineering and Technology, vol. 1, Second Ed. Taylor & Khanali, M., Rafiee, S., 2014. Investigation of hydrodynamics,
Francis Books, London, UK, ISBN 978-0-8493-3653- kinetics, energetic and exergetic aspects of fluidized bed
3659. drying of rough rice. Int. J. Food Eng. 10 (1), 39–50.
Corzo, O., Bracho, N., Vasquez, A., Pereira, A., 2008. Energy and Khanali, M., Aghbashlo, M., Rafiee, S., Jafari, A., 2013. Exergetic
exergy analyses of thin layer drying of coroba slices. J. Food performance assessment of plug flow fluidized bed drying
Eng. 86, 151–161. process of rough rice. Int. J. Exergy 13 (3), 387–408.
Dai, J.-W., Xiao, H.-W., Zhang, L.-H., Chu, M.-Y., Qin, W., Wu, Z.-J., Khoshnevisan, B., Rafiee, S., Iqbald, J., Shamshirbande, S., Omid,
Han, D.-D., Li, Y.-L., Liu, Y.-W., Yin, P.-F., 2019. Drying M., Anuarf, N., Abdul Wahabg, A., 2015. A comparative study
characteristics and modeling of apple slices during between artificial neural networks and adaptive neuro-fuzzy
microwave intermittent drying. J. Food Process Eng. 42 (6), inference systems for modeling energy consumption in
e13212. greenhouse tomato production—a case study in Isfahan
Darvishi, H., Zarein, M., Farhudi, Z., 2016. Energetic and exergetic province. J. Agric. Sci. Technol. 17, 49–62.
performance analysis and modeling of drying kinetics of kiwi Liu, Z.-L., Bai, J.-W., Wang, S.-X., Meng, J.-S., Wang, H., Yu, X.-L.,
slices. J. Food Sci. Technol. 53, 2317–2333. Gao, Z.-J., Xiao, H.-W., 2019. Prediction of energy and exergy of
Darvishi, H., Azadbakht, M., Noralahi, B., 2018. Experimental mushroom slices drying in hot air impingement dryer by
performance of mushroom fluidized-bed drying: effect of artificial neural network. Dry. Technol.,
osmotic pretreatment and air recirculation. Renew. Energy http://dx.doi.org/10.1080/07373937.2019.1607873.
120, 201–208. Manalu, L.P., Tambunan, A.H., Subandrio, Wibowo, T.Y., 2016.
Energy and exergy analysis of zedoary slices drying based on
 Food and Bioproducts Processing 1 2 4 ( 2 0 2 0 ) 168–183 183

experimental data. Int. J. Innov. Res. Sci. Eng. Technol. 5 (7), Sun, Q., Zhang, M., Mujumdar, A.S., 2019. Recent developments of
13607–13617. artificial intelligence in drying of fresh food: a review. Crit.
Marvasti, M.B., Zare, D., Stroshine, R., 2015. An integrated energy Rev. Food Sci. Nutr. 59 (14), 2258–2275.
and quality approach to optimization of green pea drying in a Taghinezhad, E., Kaveh, M., Khalife, E., Chen, G., 2020. Drying of
hot air infrared-assisted vibratory bed dryer. J. Food Eng. 166, organic blackberry in combined hot air-infrared dryer with
302–315. ultrasound pretreatment. Dry. Technol.,
Mokhtarian, M., Tavakolipour, H., Kalbasi ashtari, A., 2016. Energy http://dx.doi.org/10.1080/07373937.2020.1753066.
and exergy analysis in solar drying of pistachio with air Taheri-Garavand, A., Karimi, F., Karimi, M., Lotfi, V., Khoobbakht,
recycling system. Dry. Technol. 34, 1484–1500. G., 2018. Hybrid response surface methodology–artificial
Mondal, M.d. H.T, Hossain, Md. A, Sheikh, Md. A.M., neural network optimization of drying process of banana
Akhtaruzzaman, Md., Sarker, Md. S.H., 2020. Energetic and slices in a forced convective dryer. Food Sci. Technol. Int. 24
exergetic investigation of a mixed flow dryer: A case study of (4), 277–291.
maize grain drying. Dry. Technol., Vega-Gálvez, A., Ah-Hen, K., Chacana, M., Vergara, J.,
http://dx.doi.org/10.1080/07373937.2019.1709077. Martínez-Monzó, J., García-Segovia, P., Lemus-Mondaca, R., Di
Motevali, A., Minaei, S., 2012. Effects of microwave pretreatment Scala, K., 2012. Effect of temperature and air velocity on
on the energy and exergy utilization in thin-layer drying of drying kinetics, antioxidant capacity, total phenolic content,
sour pomegranate arils. Chem. Ind. Chem. Eng. Q. 18 (1), colour, texture and microstructure of apple (var. Granny
63–72. Smith) slices. Food Chem. 132 (1), 51–59.
Motevali, A., Minaei, S., Banakar, A., Ghobadian, B., Vijayan, S., Arjunan, T.V., Kumar, A., 2020. Exergo-environmental
Khoshtaghaza, M.H., 2014. Comparison of energy parameters analysis of an indirect forced convection solar dryer for drying
in various dryers. Energy Convers. Manage. 87, bitter gourd slices. Renew. Energy 146, 2210–2223.
711–725. Wang, J., Law, C.-L., Nema, P.K., Zhao, J.-H., Liu, Z.-L., Deng, L.-Z.,
Nazghelichi, T., Kianmehr, M.H., Aghbashlo, M., 2010. Gao, Z.-J., Xiao, H.-W., 2018. Pulsed vacuum drying enhances
Thermodynamic analysis of fluidized bed drying of carrot drying kinetics and quality of lemon slices. J. Food Eng. 224,
cubes. Energy 35, 4679–4684. 129–138.
Nazghelichi, T., Aghbashlo, M., Kianmehr, M.H., Omid, M., 2011. Xiao, H.-W., Law, C.-L., Sun, D.-W., Gao, Z.-J., 2014. Color change
Prediction of energy and exergy of carrot cubes in a fluidized kinetics of American Ginseng (Panax quinquefolium) slices
bed dryer by artificial neural networks. Dry. Technol. 29, during air impingement drying. Dry. Technol. 32, 418–427.
295–307. Yogendrasasidhar, D., Setty, Y.P., 2018. Drying kinetics, exergy and
Nguyen, T.K., Mondor, M., Ratti, C., 2018. Shrinkage of cellular energy analyses of Kodo millet grains and Fenugreek seeds
food during air drying. J. Food Eng. 230, 8–17. using wall heated fluidized bed dryer. Energy 151, 799–811.
Nikbakht, A.M., Motevali, A., Minaei, S., 2014. Energy and exergy Yu, X.-L., Zielinska, M., Ju, H.-Y., Mujumdar, A.S., Duan, X., Gao,
investigation of microwave assisted thin-layer drying of Z.-J., Xiao, H.-W., 2019. Multistage relative humidity control
pomegranate arils using artificial neural networks and strategy enhances energy and exergy efficiency of convective
response surface methodology. J. Saudi Soc. Agric. Sci. 13, drying of carrot cubes. Int. J. Heat Mass Transf. 149, 119231.
81–91. Zhang, Y., Zielinska, M., Vidyarthi, S.K., Zhao, J.-H., Pei, Y.-P., Li, G.,
Ojediran, J.O., Okonkwo, C.E., Adeyi, A.J., Adeyi, O., Olaniran, A.F., Zheng, Z.-A., Wu, M., Gao, Z.-J., Xiao, H.-W., 2020. Pulsed
George, N.E., Olayanju, A.T., 2020. Drying characteristics of pressure pickling enhances acetic acid transfer, thiosulfinates
yam slices (Dioscorea rotundata) in a convective hot air dryer: degradation, color and ultrastructure changes of “Laba” garlic.
application of ANFIS in the prediction of drying kinetics. Innov. Food Sci. Emerging Technol. 65, 102438.
Heliyon 6, e03555. Ziaforoughi, A., Yousefi, A.R., Razavi, S.M.A., 2016. A comparative
Onwude, D.I., Hashim, N., Chen, G., 2016. Recent advances of modeling study of quince infrared drying and evaluation of
novel thermal combined hot air drying of agricultural crops. quality parameters. Int. J. Food Eng. 12, 901–910.
Trends Food Sci. Technol. 57, 132–145. Zielinska, M., Zapotoczny, P., Alves-Filho, O., Eikevik, T.M.,
Parizi, S.N., Beheshti, B., Roustapour, O.R., 2016. Investigation of Blaszczak, W., 2013a. Microwave vacuum–assisted drying of
pistachio (Kalleh Ghoochi v.) drying kinetics in a new green peas using heat pump and fluidized bed: a comparative
intelligent rotary dryer under vacuum. Food Sci. Technol. 54 study between atmospheric freeze drying and hot air
(13), 135–143 (In Farsi). convective drying. Dry. Technol. 31, 633–642.
Peinado, D., de Vega, M., García-Hernando, N., Marugán-Cruz, C., Zielinska, M., Zapotoczny, P., Alves-Filho, O., Eikevik, T.M.,
2011. Energy and exergy analysis in an asphalt plant’s rotary Blaszczak, W., 2013b. A multi-stage combined heat pump and
dryer. Appl. Ther. Eng. 31, 1039–1049. microwave vacuum drying of green peas. J. Food Eng. 115,
Sheikhshoaei, H., Dowlati, M., Aghbashlo, M., Rosen, M.A., 2019. 347–356.
Exergy analysis of a pistachio roasting system. Dry. Technol., Zohrabi, S., Seiiedlou, S.S., Aghbashlo, M., Scaar, H., Mellmann, J.,
http://dx.doi.org/10.1080/07373937.2019.1649276. 2020a. Enhancing the exergetic performance of a pilotscale
Sivakumar, R., Saravanan, R., Elaya, A.P., Iniyan, S., 2016. convective dryer by exhaust air recirculation. Dry. Technol. 38
Fluidized bed drying of some agro products– a review. Renew. (4), 518–533.
Sustain. Energy Rev. 61, 280–301. Zohrabi, S., Aghbashlo, M., Seiiedlou, S.S., Scaar, H., Mellmann, J.,
Soysal, A., Oztekin, S., Eren, O., 2006. Microwave drying of parsley: 2020b. Energy saving in a convective dryer by using novel
modelling, kinetics, and energy aspects. Biosyst. Eng. 93 (4), real-time exergy-based control schemes adjusting exhaust air
403–413. recirculation. J. Clean. Prod. 257, 120394.