Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Contents lists available at ScienceDirect

Renewable and Sustainable Energy Reviews

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

Application of artificial neural network for predicting performance of solid MARK

desiccant cooling systems – A review
D.B. Jania, , Manish Mishrab, P.K. Sahoob
⁎

a
Department Gujarat Technological University, GTU, Ahmedabad, India
b
Department of Mechanical & Industrial Engineering, Indian Institute of Technology Roorkee, 247667, India

A R T I C L E I N F O A BS T RAC T

Keywords: In present study, an attempt has been made to review the applications of artificial neural network (ANN) for
ANN predicting the performance of solid desiccant cooling systems. Different types of neural networks are applied to
Desiccant cooling model the solid desiccant cooling systems. With use of experimental data, an ANN model was developed which
Dehumidifier is based on different algorithms. Available experimental data were divided into two categories for training and
COP
testing of the ANN model. Later on, trained ANN model was tested for predicting the performance of system
Regeneration
based on various input and output parameters such as air stream flow rates, temperatures and humidity ratios,
pressure drop, dehumidifier effectiveness, cooling capacity, regeneration temperature, power input, coefficient
of performance etc. So, present review proposes the use of ANN based model to simulate the relationship
between inlet and outlet parameters of the system. The ANN predictions for these parameters usually agreed
with the experimental values with higher correlation co-efficient. The previous studies show that ANNs can be
used with a higher precision in guessing the performance of solid desiccant cooling systems. This review is
useful for making opportunities to further research of ANNs and its feasibility which is becoming common in the
coming days.

1. Introduction
The performance of solid desiccant cooling systems is evaluated in
terms of energy analysis (first law) and exergy analysis (second law) of
thermodynamics using conventional approaches i.e. analytical, numer-
ical and experimental methods. The conventional analytical and
numerical approaches involve complex equations and assumptions,
whereas experimental studies are rigorous and costlier. During last two
decades, the use of artificial intelligence systems in the field of space
cooling is increasing gradually to solve the complicated problems.
Artificial intelligence systems include areas such as artificial neural
network (ANN), genetic algorithms, fuzzy logic and various hybrid
systems, which combine two or more techniques. The main advantage
of ANN as compared to the other artificial intelligent systems is its
accuracy, speed, simplicity and ability to model a multivariable
problem for solving complex relationships between the variables and
establishing non linear relationships by means of training network.
ANN model can forecast the desired output of the system using limited
training data. Moreover, ANN overcomes the limitations of traditional
approaches by extracting the required information using training data,
which has not required any specific analytical equations or exhaustive
experimental study. So, manufacturers may employ the ANN technique
for evaluating the performance of solid desiccant cooling that can save
both engineering effort and funds. Earlier many review studies were
reported with applications of ANN in the field of drying processes [1],
forecasting [2], atmospheric sciences [3], solar photovoltaic systems
[4] and modeling of energy systems [5–7]. Ding [8] summarized
various simulation techniques for modeling and performance estima-
tion of conventional VCR systems.
Following the cited literature, it is seen that there is no specific
review has been reported on applications of ANN for solid desiccant
cooling systems. The present review set out more broadly about up to
date study covering the applications of ANN in the performance
forecasting of solid desiccant cooling systems by prediction of para-
meters in terms of cooling capacity, dehumidifier effectiveness, regen-
eration temperature, power input and COP of the system. Moreover,
the limitations with ANN models are also highlighted. The present
review can be impacted in term of application of ANN for performance
estimation of desiccant cooling system requiring less formal statistical
training, ability to implicitly detect complex nonlinear relationships
between dependent and independent variables, ability to detect all
possible interactions between predictor variables and the availability of

⁎
Corresponding author.
E-mail addresses: dbjani@rediffmail.com (D.B. Jani), mmishfme@iitr.ac.in (M. Mishra), sahoofme@iitr.ac.in (P.K. Sahoo).

http://dx.doi.org/10.1016/j.rser.2017.05.169
Received 29 November 2016; Received in revised form 10 March 2017; Accepted 20 May 2017
1364-0321/ © 2017 Elsevier Ltd. All rights reserved.
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Nomenclature VCS vapor compression system

W humidity ratio (g/kg)
a actual output Wc compressor work (kW)
ANN artificial neural network w synaptic weights
b bias wj,i the weight between input layer vectors and hidden layer
COP coefficient of performance vectors
COPH thermal coefficient of performance X input signal
COPT total coefficient of performance Y output signal
COPw total work coefficient of performance
Cp specific heat (J/kg K) Greek letters
DEC direct evaporative cooling
Et total energy consumption (kW) ε effectiveness
EC evaporative cooling ϕ relative humidity
h enthalpy (kJ/kg) υk,j the weight between hidden layer vectors and output layer
MRR moisture removal rate (kg/h) vectors
MSE mean square error Δm moisture removal rate (kg/h)
ṁ mass flow rate of air stream (kg/s)
p predicted output (network output) Subscript
Qcc cooling capacity (kW)
Qr regeneration heat (kW) a dry air
R correlation coefficient amb ambient air
RH Relative humidity (%) DW desiccant wheel
RHSF room sensible heat factor d desiccant
RMSE Root mean square error DEC desiccant enhanced evaporative cooler
SHR Sensible heat ratio HRW heat recovery wheel
T temperature (K) in inlet
TANSIG Tan-sigmoid transfer function i,j,k the number of nodes
TRAINLM Levenberg–Marquardt backpropagation p process air
VC vapor compression r regeneration air
VCOP coefficient of vapor compression wb wet bulb
VCR vapor compression refrigeration 1,2, etc. reference state points

multiple training algorithms. The objective of the present review is to
aware about an introduction of an application of artificial neural
network in the field of space cooling energy modeling as a powerful
data-driven, self-adaptive, flexible computational tool having the
capability of capturing nonlinear and complex underlying character-
istics of any physical process with a high degree of accuracy. Moreover,
it is found that artificial neural network is very suitable for inverse
energy modeling when the numerical relations between input and
output variables are unknown, and can't be established. Thus, this
review is useful for making opportunities to further research of use of
artificial neural network in performance prediction of solid desiccant
cooling systems and its feasibility which is becoming common in the
coming days.
2. Overview of solid desiccant cooling
2.1. Brief history of solid desiccant cooling
Different configurations of solid desiccant cooling system have been
proposed earlier by many researchers so far to attain better system
performance in terms of maintaining a thermal comfort. The earliest
form of solid desiccant cooling cycle was proposed by coupling
desiccant dehumidifier with heat source and evaporative cooler [9].
Similar cycle was proposed later by Dunkle [10] using desiccant wheel

Fig. 1. Principle of solid desiccant cooling.

353
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

made up of molecular sieve with additional heat exchanger to achieve high temperature of hot surrounding.
the better performance than the earlier one. Later on, Munter [11] Dunkle cycle combines the advantages of both the ventilation cycle,
further enhanced the performance of the solid desiccant cooling cycle having supply of large amount of ventilation fresh air with relatively
by introducing parallel passages in rotary dehumidifier and provided low temperature and the recirculation cycle, having enhanced cooling
backup of traditional vapor compression system to handle the sensible capacity. As shown in Fig. 4, an additional sensible heat wheel is
and latent load separately. Since then, number of efforts have been incorporated in system which can provide colder process air with
made for enhancing the performance of rotary desiccant dehumidifiers relatively low temperature for the sensible heat exchanger [23].
used in the desiccant cooling systems. Important among those were the Fig. 5 shows comparison between different solid desiccant cooling
analogy theory suggested by Banks [12], the pseudo-steady state model cycles on psychrometric chart. It was depicted from the psychrometric
by Barlow [13], combined potential technique by Jurinak [14], finite chart that the reactivation temperature required for desorption of
difference method for cross-cooled dehumidifiers [15] and finite rotary dehumidifier in case of recirculation cycle is highest due to
difference method by Maclaine-Cross [16] which are now widely used excess humidity ratio at the dehumidifier inlet in the regeneration air
by other investigators in getting better performance of desiccant side. The temperature of supply room air is observed lowest in case of
cooling cycles [17]. Burns et al. [18] evaluated the performance of recirculation cycle because it reuses 100% building air as a process air
hybrid desiccant cooling cycle used for supermarket and shows better at the dehumidifier inlet.
performance than the traditional VCR system in terms of energy In case of Pennington cycle, maximum amount of fresh ventilated
consumption in humid climate. Van den Bulck et al. [19] have carried air supply is possible because it uses 100% outdoor air as a process air
out experiments to investigate the effects of process and regeneration at the dehumidifier inlet. But, in case of ventilation cycle thermal
air mass flow rates, temperature, humidity ratio etc. on the perfor- coefficient of performance and specific cooling capacity would be
mance of solid desiccant cooling system. Higher regeneration tempera- reduced in comparison with other standard cycles because humidity
ture was recommended by them to reduce the required mass flow rates ratio and temperature of outdoor air are usually higher than that of the
of air streams. Pesaran [20] proposed the use of renewable energy room recalculated air especially in hot and humid surroundings.
(solar heat) for the regeneration of dehumidifier used in solid desiccant
cooling system to increase the overall system efficiency. Comparison
3. Overview of artificial neural network
among various solid desiccant cooling cycles for air conditioning in hot
and humid climates was also carried out by Jain et al. [21].
An artificial neural network consists of large number of processing
elements called neurons. They are joined by connecting links called
2.2. Principle of solid desiccant cooling
weights. A simplified ANN model has an input layer, an output layer,
and at least one hidden layer. The selection of layer is determined by
A solid desiccant is a hygroscopic material made up of silica gel, Li-
the network architecture and the method of input data required. A
Cl and Molecular sieves. Solid desiccant cooling system works on
generalized neural network model (Fig. 6) consists of three primary
principle of adsorption of water vapor from air. In solid desiccant
elements; synapses or connecting link, summing node with a squashing
cooling system, the excess water vapor present in the process air is
function and an externally applied bias to increase or decrease the net
removed by passing through rotary dehumidifier. The temperature of
input of the activation function. The network performance is deter-
this warm and dried process air is then lowered further to the desired
mined by value of weights and biases in every single neuron. The
room comfort conditions by use of sensible air-to-air heat recovery
network needs to be trained with the help of measured data sets to give
wheel and cooling coils. Dehumidifier needs to be regenerated by
the desired output using input data sets [24].
supplying hot regenerated air via heater to remove adsorbed water
Multi layer feed forward networks (MLFFN) consists of an input
vapor (Fig. 1). So, desiccant cooling system comprises mainly four
layer, followed by one or more hidden layers and an output layer.
components, namely desiccant wheel, sensible heat recovery wheel,
Multiple layers of neurons with nonlinear transfer functions make the
cooling unit and the reactivation heater.
network to learn linear and nonlinear relationships between input and
output vectors. Back propagation (BP) learning algorithm is used
2.3. Types of solid desiccant cooling cycles
widely to train the MLFFN. The network is trained with selected
number of neurons in the hidden layer, momentum factor, learning
Solid desiccant cooling system which was introduced first by
rate and transfer function. Radial biased function network (RBFN)
Pennington is shown in Fig. 2. It was also known as ventilation cycle
having three layers as input layer, hidden layer followed by an output
[22]. The moist ambient air (state 1) dehumidifies (state 2) by passing
layer. The interconnections between input layer and hidden layer form
through a rotary dehumidifier owing to adsorption and converted into
hypothetical connections and between the hidden layer and output
warm and dry air. It is then cooled sensibly by passing through sensible
layer form weighted connections. The Gaussian transfer function is
heat recovery wheel (2-3). Then after, it is cooled further up to comfort
used to operate the weighted inputs to produce neuron output. The
supply conditions (state 4) by passing through direct evaporative
activation function used in hidden layer is radial base function. The
cooler. Return room air (state 5) is cooled and humidified by passing
linear transfer function is used in output layer. RBFN has faster
through another direct evaporative cooler (state 6). This air is then
sensibly heated with process air by passing through sensible heat
recovery wheel (state 7). This warm air is then heated up further to the
required regeneration temperature of desiccant material used in the
desiccant wheel by passing through regeneration heat source. After
reactivating the desiccant wheel, this air is then exhausted to surround-
ing (state 9). A small portion of return air steam (state 7) bypasses the
regeneration heat source in order to reduce the regeneration heat
consumption.
To enhance cooling capacity, recirculation cycle shown in Fig. 3
which is a modified form of Pennington (ventilation) cycle, is config-
ured to reuse the room return air as a dehumidifier process air inlet in
severe hot and humid climatic conditions [22]. This is because due to
the fact that this lowers the energy consumption to lower the extreme Fig. 2. Pennington (ventilation) cycle.

354
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Fig. 3. Recirculation cycle.

Fig. 6. Structure of an artificial neural network [24].

square nodes. The second layer is called product layer and every node
in this layer is a fixed node marked by a circle node and labelled by M.
The third layer is a normalized layer and every node in this layer is a
fixed node, marked by a circle node and labelled by N. The fourth layer
is the de-fuzzy layer having adaptive nodes and marked by square
nodes. The fifth layer computes the overall system output as the
summation of all incoming signals [102–104].
Outputs from the ANN model are compared with the actual output.
Fig. 4. Dunkle cycle.
There may be small variation between the network’s output and the
target output. The weights are adjusted such a way that the error
function minimizes the variation between actual outputs and model
outputs. This process is continued until the error function comes under
the desired tolerance limit [25]. This repetitive process of training and
correction of the weights, is known as back propagation algorithm.
While training the ANN model, the weights and bias which minimize
the error between the measured output and the ANN network output
are obtained [26] as follows
⎡ N ⎤
Y = F (S) = F ⎢ Σ XK wK + b⎥
⎣K=1 ⎦ (1)

Neural network tool box of MATLAB© was used for the formation
of network architecture. ANN structure can be made by inserting the
weights into each layer of layered structure by selecting the weight
matrices. The values of these weights are determined using the feed
forward backpropagation (BP) method. Error calculation is iterated
using different inputs and outputs during learning until the root mean
square error (MSE) of the network reaches to an acceptable level. The
performance estimation through the neural network is done by mean
square error (MSE) [27] between the estimated and the actual values as
per the following relation
N
Σ X predicted − X experimental 2
i=1
MSE =
N (2)

Lesser the MSE is, better fit the results will be. The root mean
square error [28] can be given by
Fig. 5. Comparison between different solid desiccant cooling cycles.

1 N
convergence, smaller extrapolation errors and has higher reliability RMSE = Σ (ai−pi )2
N i =1 (3)
compared to MLFFN. Generalized regression neural network (GRNN)
made up of four-layer feed forward neural network which is based on where, ai and pi are number of nodes for actual output and estimated
the non-linear regression theory consisting of input layer, pattern layer, outputs, respectively. The absolute fraction of variance (R2) [29], a
summation layer and output layer. GRNN is a memory-based feed statistical indicator that can be applied to multiple regression analysis,
forward networks as it consists of estimation of probability density is obtained as
functions. Adaptive neuro fuzzy interface systems (ANFIS) consisting ⎛ N ⎞
of nodes and directional links, which combines the learning capabilities ⎜ Σ (ai − pi )2 ⎟
i =1
of a neural network and reasoning capabilities of fuzzy logic. ANFIS R2 = 1−⎜ N ⎟
⎜⎜ Σ ( p )2 ⎟⎟
architecture made up of fiver layers. The first layer in the network is ⎝ i =1 i ⎠ (4)
called fuzzy layer. The adjustable nodes in this layer are represented by
2 2
R ranges between 0 and 1. A very good fit yields R value of 1 or

355
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

closer, whereas a poor fit results to a value near 0. streams at the dehumidifier inlet and outlet respectively.
Fig. 7 illustrates the flow chart that describes the development, Dehumidifier effectiveness (εdw) can be expressed by [32]
training and simulation of an artificial neural network. The experi-
w1 − w2
mental measurements are input parameters to the model. Neural εdw =
w1 − w2,ideal (6)
network understands the underlying correlations in the entered input
data and stores them as inter-neuron connection strengths or corrected where w2,ideal is the ideal humidity ratio of the process air stream at exit
weights. Based on number of neurons, number of iterations and the of the rotary desiccant dehumidifier, it is to be taken zero by assuming
desired accuracy, the training set and the target set are developed. The that the air is completely dehumidified at this point.
network needs to be trained using training data set consisting of a Effectiveness of sensible heat wheel (εhw) can be determined by [33]
group of input data and corresponding output data. In an artificial
neural network model, different input parameters are used such as T2 − T3
εhw =
temperature, humidity ratio or relative humidity, dehumidifier pres- T2 − T6 (7)
sure drop, flow rates of process as well as regeneration air streams at
different state points in the cycle, while moisture removal rate (MRR), where the subscripts correspond to the state point in the solid desiccant
dehumidifier effectiveness, cooling capacity, power consumption, coef- cooling system.
ficient of performance (COP) of the system are used as output Effectiveness of evaporative coolers (εec) can be evaluated by [34]
parameters. Training involves the revision of synaptic weights. The T3 − T4 T − T6
network reads and processes each set of input data and produces an εec1 = and εec2 = 5
T3 − Twb,3 T5 − Twb,5 (8)
output, which is compared with the actual (measured) output. Based on
the difference between the network output and the target output, the
model parameters are adjusted so that the network would exhibit the
targeted results. The network performance was mainly determined by
the weights and bias values in every single neuron.
The three-layer back propagation (BP) ANN network for rotary
dehumidifier is shown in Fig. 8 as a sample study. wj,i represents the
weights between input layer vectors and hidden layer vectors, and υk,j
represents the weights between hidden layer vectors and output layer
vectors. The training process requires a proper set of valid data i.e.
input and target output. The input layer in the present sample case has
seven nodes namely inlet temperature, humidity ratio and flow rate of
process air and ambient air respectively, and regeneration air tem-
perature. The output layer has six layers namely outlet temperature
and humidity ratio of process air and regeneration air respectively,
including the effectiveness and moisture removal rate of rotary
desiccant dehumidifier. Back propagation is a multilayer feed forward
network with hidden layers between the input and output. The hidden
layer has seven nodes. During training, the weights and biases of the
network are iteratively adjusted to minimize the network performance
function. The training parameters used for the simulation of sample
ANN model are summarized in Table 1. TRAINLM, MSE, TANSIG are
the training, performance and transfer functions respectively in the
simulation. Typical performance function that is used for training feed
forward back propagation neural network is the network mean square
errors (MSE). During training, the weights and biases of the network
are iteratively adjusted to minimiseminimize the network performance
function. LM is the fastest training algorithm for network of moderate
size and it has the memory reduction feather to be used when the
training set is large. While mu is the iteration step size and epoch is a
run through all training input-output sets.
The trained artificial neural network model of the sample case is
shown in Fig. 9. This shows 7-7-6-6 network structure with 7 hidden
layers yielded the good model for the accurate estimation of outputs,
with minimum MSE during training because it has higher stability and
faster convergence rate. This trained network can be used for simulat-
ing the system outputs for the inputs that have been introduced before.

4. Performance indices

The performance of solid desiccant wheel used in solid desiccant

cooling system was determined by calculating the moisture removal
rate (MRR) and dehumidifier effectiveness. The moisture removal rate
or moisture transfer rate [31] from air to the desiccant surface given by
Δ m = mp (W1 − W2) (5)
where, mp is the mass flow rate of process air stream at the
dehumidifier inlet, W1 and W2 are the humidity ratios of process air Fig. 7. Flow chart of ANN model [30].

356
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Q cc mp (h5 − h 4)
COP = =
Qr mr (h8 − h7) (10)
where mp and mr denotes the mass flow rate of process air as well as
regeneration air streams, respectively.
Similarly, the coefficient of performance (COP) for the solid
desiccant – vapor compression hybrid cooling system can be given as
follows [37]
Q cc
COPT =
Et (11)
where Qcc is the cooling capacity which is defined as following [38]
Q cc = mp (h1 − h5) (12)
where Et is the total energy consumption of the system expressed by
[39]
E t = Q r + Wc + Woth (13)
Woth shows energy consumption of other equipments are fans, desiccant
wheel and heat wheel motor. While Qr is the heat supply to regenera-
tion air stream by a reactivation heat source is given as [40]
Q r = mr (h8 − h7) (14)
COPw and COPH can also be defined by [41]
Q cc Q
COPw = and COPH = cc
W Qr (15)
Fig. 8. Architecture of proposed ANN network (BP) for rotary dehumidifier [30].
From last equation we obtain:
Table 1 1 1 1
Training parameters used for the ANN model [30]. = +
COPT COPW COPH (16)
Sr. no. Training parameter Type/value
Similarly, the coefficient of performance (COP) for the solar assisted
1 Training function TRAINLM solid desiccant – vapor compression hybrid air-conditioning can be
2 Performance function MSE given as follows [42–57]
3 Transfer function TANSIG
4 epochs 1000 Q DEC
COPDEC =
5 min_gradmin _grad 0.000001 Q HC (17)
6 mu 0.001
7 mu_dec 0.1 where solar cooling coefficient of performance (COPDEC) is the ratio
8 mu_inc 10 between useful cooling output of the desiccant cooling cycle (QDEC) and
9 max_timemax _time inf the regeneration heat delivered by the solar heating coil (QHC). So,
10 goal 0.000000000000001
11 max_failmax _fail 50 Q DEC = mp (h3 −h2) and Q HC = mr (h8 −h7) (18)
while coefficient of performance (COP) of vapor compression
where Twb is the wet bulb temperature of moist air. systems (COPVCR) is expressed as follows [58–70]
The rates of moisture added to air by the evaporative coolers in the QL
process and return lines are obtained by following [35] COPVCR =
Wi (19)
mw1 = mp (w4 − w3) and mw2 = mr (w6 − w5) (9) where QL is cooling effect in kW and Wi is electric power consumption
for VCR compressor in kW.
where m w1 and m w2 are the rates of moisture added to air in the
evaporative coolers in supply and return lines respectively. 5. Important literature survey on applications of ANN
The coefficient of performance (COP) for the solid desiccant
evaporative cooling system can be expressed by [36] The earliest effort was made to apply artificial neural network to

Fig. 9. The trained ANN model for rotary dehumidifier [30].

357
D.B. Jani et al.
optimize energy consumption in HVAC system by Curtiss et al. [71].
The network was able to estimate energy use better than the traditional
regression techniques, and the energy management system was able to
maintain thermal comfort and saves more energy as compared to a
fixed set-point or basic temperature reset algorithm. Later on, Kah
et al. [72] proposed a modular network using a multilayer perceptron
for controlling the temperature of the air-conditioning system used in a
public transport. The input parameters used for training the network
were the temperature, number of passengers and time of day. A model
was also used for decision making to set the system to an optimum
state of operation that maintains thermal comfort inside conditioned
space. The authors claimed that the system offered fast learning with
good accuracy. Datta et al. [73] presented a similar application where a
number of networks were compared, with the objective of identifying
the important inputs to the network which will facilitate on-line
prediction and thereby implement refrigeration and HVAC system
diagnostics, process control, optimization and energy management in
retail stores. The network showed a better performance as compared to
conventional multiple regression techniques. Kajl et al. [74] proposed
an artificial neural network model which can fill the gap between
simplified and detailed prediction methods for building energy con-
sumption. The neural network training and testing data set were based
on simulation results of numerous office buildings carried out with the
DOE-2 software program. It is found that the proposed model retains
all the merits of the simple steady state methods and additionally, it
gives certain advantages of the detailed dynamic methods, as for
example, the interaction between the envelope and the HVAC systems
of the building. Kalogirou [5] carried out an important study to review
various applications of artificial neural network in energy systems used
in building cooling. Errors reported in models discussed in review were
well a within acceptable limit, which suggests that artificial neural
networks can readily be used for modeling in other fields of energy
production and use.
Artificial neural network modeling of rotary desiccant dehumidifier
was carried out first by Cejudo [75]. The neural network model consists
of a four-input–four-output network that calculates the outlet condi-
tions from inlet ones. Real data are used to validate the ANN model and
to train the neural network. A small discrepancy between calculated
and measured values was found mainly due to the fact that the
dehumidification is assumed to be adiabatic in theoretical model as
well as the heat losses in the ducts and the dehumidifier were not
considered in the ANN model, but in the experimental facility these
losses can occur. The second attempt was made by Parmar and
Hindoliya [76] to model desiccant dehumidifier with help of ANN. A
neural network model was proposed using a neural network toolbox of
MATLAB with feed forward back propagation method. The model was
validated against experimental data sets. A number of training algo-
rithms with feed forward back propagation method were used for the
modeling of rotary dehumidifier to identify a training algorithm with
least mean square error (MSE). The performance of all training
algorithms was analyzed and training algorithm trainlm (Levenberg–
Marquardt back propagation) is found most suitable for the estimation
of outputs which have least mean square error. The proposed model
was predicted the specific humidity and temperature at the dehumidi-
fier exit well within the range of experimentally measured values.
Characteristics of a dehumidifier made up of silica-gel were predicted
using artificial neural networks (ANNs) over a wide range of operating
conditions by Mahmoud et al. [77]. Two different Multi-layer
Perceptron feed forward Neural Networks (MLPNN) were developed
to predict dry bulb temperature and relative humidity of process air
stream at the dehumidifier exit during dehumidification and reactiva-
tion operating modes. Training and testing process of the proposed
ANN model was based on experimentally measured data. Effects of
desiccant mass on air mass flow rate, shelves span and air dry bulb
temperature were predicted. Results showed that the ANN model
predicts data that are very close to the available experimental data
358
Renewable and Sustainable Energy Reviews 80 (2017) 352–366
with a maximum possible error of ± 5% and a correlation coefficient up
to 0.989 and 0.999 during dehumidification and regeneration operat-
ing modes, respectively. The reported results revealed that the desic-
cant bed can be modelled using ANNs with a high degree of accuracy.
Uçkan et al. [78] proposed mathematical equations derived from
Artificial Neural Networks (ANNs) for the prediction of dry bulb
temperature and the specific humidity at the dehumidifier exit. The
neural network model consists of five inputs and two output neurons
that define the outlet conditions (DBT and humidity ratio) of the rotary
dehumidifier. The results predicted by the ANNs model are compared
with the actual measured data by using input variables. It is found that
the mean absolute percentage errors (MAPE) for DBT and humidity
ratio are found to be 0.80% and 1.56%, respectively and the correlation
coefficient (R) values are obtained to be approximately 0.986 for both
output variables while the root mean square errors (RMSE) are found
as 0.54% and 0.18% for DBT and humidity ratio, respectively. Zhu and
Chen [79] developed ANN model of marine rotary desiccant air-
conditioning by using multivariate linear regression analysis method.
The results show that the square of multiple coefficient correlation (R2)
of the prediction models reached 0.999 after the seventh corrected and
the relative error much less than 1%. Neural network is used to analyze
the effect of significant operating parameters on moisture adsorption
rate in desiccant wheel by Rafique [80] because in analytical model it is
difficult to predict relationship between inlet and outlet parameters.
Gawad et al. [81] proposed artificial neural network (ANN) and
adaptive neuro-fuzzy inference system (ANFIS) artificial intelligent
techniques to estimate the critical characteristics of the rotary desiccant
dehumidifier. The parameters used in the modeling are like as flow
velocity, shape of cross-section of flow path and matrix material.
Moreover, both these artificial intelligence techniques use experimen-
tally measured results for training and validation of model.
Zendehboudi [82] proposed hybrid model based on least squares
support vector machine (LSSVM) and genetic algorithm (GA) to
estimate DBT and humidity ratio of process air as well as regeneration
air at the dehumidifier exit, dehumidifier effectiveness, moisture
removal capacity (MRC), and sensible energy ratio (SER) for desiccant
materials as silica gel and molecular sieve by considering different
process/regeneration section (P/R) area ratios. The capability of the
model was evaluated through three different statistical error tests. The
obtained results show that, integration of LSSVM and GA is very
accurate technique for estimating the performance of desiccant wheel,
having a mean average error (MAE) is less than 0.23, determination
coefficient (R2) is greater than 0.994, and mean squared error (MSE) is
less than 0.072. Jani et al. [30] developed an artificial neural network
(ANN) model to predict the performance of a desiccant wheel for
different process air inlet conditions. DBT, humidity ratio and flow rate
of the process as well as regeneration air streams of dehumidifier and
regeneration temperatures are used as inputs to the model. The outputs
of the model are outlet DBT and humidity ratio of process as well as
regeneration air streams of dehumidifier, the moisture removal rate
(MRR) and the effectiveness of the dehumidifier. Performance estima-
tions through ANN are compared with the experimentally measured
data and a close agreement is observed between them.
Summary of dehumidifier parameters which are studied by pre-
vious researchers have been tabulated in Table 2.
Ertunc and Hosoz [83] developed an ANN model to predict the
performance of evaporative condenser cooling system. The ANN model
was used for estimating various performance parameters of the system,
namely the condenser heat rejection rate, refrigerant mass flow rate,
compressor power, electric power input to the compressor motor and
the system coefficient of performance (COP). The ANN estimations are
in good match with the experimental values with higher correlation
coefficients and very low root mean square errors. The obtained results
show that such systems can alternatively be modelled using ANNs
within a high degree of accuracy. Yigit and Ertunc [84] predicted dry
bulb temperature and humidity ratio at exit of a sensible heat
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Table 2 artificial neural network (ANN) based on back propagation algorithm

Summary of dehumidifier parameters studied by previous work. to predict performance of a direct evaporative cooling system.
Experimentally measured data is used for training a three-layer feed-
Author Network Parameters MSE (%) R
architecture forward ANN model. . The estimations usually close match with the
experimental values with correlation coefficients (R) in the range of
Cejudo [75] Liner sigmo-dial Tpro_air 0.016 – 0.969–0.993, mean relative errors in the range of 0.66–4.04%, and
Treg_air 0.01 –
very low root mean square errors. This study reveals that, as an
Wpro_air 0.107 –
Wreg_air 0.05 – alternative to classical modeling techniques, the ANN approach can be
Parmar and Multilayer feed Tpro_air 0.07 – used successfully for estimating the performance of direct evaporative
Hindoliya [76] forward Treg_air 0.04 – air coolers. Kiran and Rajput [87] developed ANN models using the
Wpro_air 0.06 – training algorithm Levenberg–Marquardt (LM) for estimating perfor-
Mahmoud et al. Multilayer Tpro_air 0.05 0.999
mance of indirect evaporative cooler (IEC). The models are trained by
[77] perceptron Treg_air 0.994
Wpro_air 0.975 using simulation data and are then compared and validated using
mdes 0.980 experimental data from the literature. It was found that the ANN model
ṁf_pro_air 0.998 gives accurate results with good statistical values. A neural network
Uckan et al. [78] Feed forward BP Tpro_air 0.05 0.995
model based on the training of a black box model with experimental
Treg_air – 0.986
Wpro_air 0.03 – data was proposed by Koronaki et al. [88] for estimating the state
Wreg_air – – conditions of air at the process and regeneration stream. Later on, they
ṁf_pro_air – – have also performed a Sensitivity Analysis on these ANN estimated
ṁf_reg_air – – results. The parameters were taken like as dry bulb temperature and
Zhu and Chen [79] Multivariate linear Tpro_air – 0.879
humidity ratio of process and regeneration air streams, the sensible
regression Treg_air
ϕ_pro_air
heat factor of the room, and the mass air flow ratio of the regeneration
COP and process air streams. Estimated output values of the developed ANN
Rafique [80] Feed forward BP Tpro_air 0.022 – model for desiccant cooling systems are in line with results from other
Treg_air
correlation models based on the interpolation of experimental data
Wpro_air
Wreg_air
obtained from industrial air conditioning installations. Table 3 shows
ṁf_(P/R)_air comparison between the mean relative and absolute errors of the feed-
Gawad et al. [81] Multilayer feed Channel shape – forward and the radial-basis neural network models while Fig. 10
forward Circular 0.05 depicts influence of input variables on outlet conditions of process and
Hexagonal 0.06
reactivation air streams for standardised regression coefficient values.
Sinusoidal 0.04
Jani et al. [30] Feed forward BP Tpro_air – 0.998 Sohani et al. [89] carried out an investigative study on the
Treg_air – parameters having potential impact on the performance of the desic-
Tamb_air – cant integrated evaporative cooling system by use of an artificial neural
Wpro_air –
network. The developed ANN models were trained and tested using
Wamb_air –
ṁf_reg_air –
experimental data. Base case values and parameter variation ranges of
MRR – effective parameters of the desiccant dehumidifier are given in Table 4.
εdw – An ANN model for a solid desiccant – vapor compression hybrid
Zendehboudi [82] – Tpro_air 0.072 0.996 air-conditioning system was proposed by Jani et al. [29] to estimate the
Wpro_air
cooling capacity, power input and COP of the system. The experimental
Treg_air
εdw measurements were taken at steady state conditions while varying the
MRC input parameters such as air stream flow rates and regeneration
temperature. Most of the experimental test data (80%) are used for
training the ANN model while remaining (20%) are used for the testing
exchanger coupled to a VCR unit using artificial neural networks. DBT of ANN model. The outputs estimated from the ANN model have a high
and humidity ratio of ambient air entering the coil, its velocity, frost coefficient of correlation (R > 0.988) in forecasting the system
weight, the temperature at the coil surface etc. are taken as input performance. It is found that the ANN model can be applied success-
parameters to the model. Moreover, DBT and humidity ratio of the air fully and can provide high accuracy and reliability for estimating the
exiting the coil can be taken as output parameters. A feed-forward performance of the hybrid desiccant cooling systems. The regression
neural network based on backpropagation algorithm was developed to plot between ANN estimations and the experimental measurements for
model the thermal performance of the coil. The artificial neural (a) Training, (b) Validation, (c) Test, and (d) Combined set is shown in
network (ANN) was trained using the experimental data to predict Fig. 11.
the air conditions at the exit of the coil. The predicted values are found Hosoz and Ertunc [90] developed artificial neural networks (ANNs)
to be in close agreement with the actual values from the experiments.
This demonstrates that the ANN presented can help the manufacturer
to forecast the performance of cooling coils in air-conditioning systems Table 3
under various operating conditions. The performance of an evaporative Comparison between the mean relative and absolute errors for the feed-forward and the
radial-basis neural network models [88].
cooling condenser integrated to conventional VCR system was pre-
dicted by Ertunc and Hosoz [85], using artificial neural network Feed-forward neural network Radial-basis neural network
(ANN). DBT, WBT, flow rates of air and water streams, pressure dropt model model
etc. are used as inlet parameters. While the model was used for
Absolute Relative mean Absolute Relative mean
predicting the condenser heat rejection rate, refrigerant temperature
mean error error (%) mean error error (%)
leaving the condenser along with dry and wet bulb temperatures of the
leaving air stream. The obtained results show that the artificial neural T_pro_out 0.0026 °C 4.5533 0.004 °C 7.5014
network models yielded a good statistical prediction performance in RH_pro_out 0.1242 0.5239 0.1941 0.8208
terms of correlation coefficient, mean relative error, root mean square T_reg_out 0.0027 °C 5.4877 0.0069 °C 17.804
RH_reg_out 0.1159 0.3043 0.363 0.9540
error and absolute fraction of variance. Hosoz et al. [86] developed

359
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Fig. 10. Influence of different input variables on exit conditions of process and reactivation air streams for standardised regression coefficient values [88].

Table 4 Empirical correlations usually provide vapor pressure with limited

Base case values and parameter variation ranges of effective parameters of the desiccant
dehumidifier [89].
Parameter Base case Variation range
value
Process air temperature at the inlet (°C) 32 25–38
Process air humidity ratio at the inlet 13.5 9–21
(gmmoisture/kgdry air)
Process air velocity at the inlet (m/s) 1.4 0.4–3.5
Desiccant mass ratio (DAR) 0.9 0.1–1.3
Length of the product channel (m) 0.4 0.1–0.5
Height of the product channel (m) 3.3 –
Width (m) 0.5 –
Height(m) 0.5 – ANN model was proposed having 2 cells in its input layer and 3 cells in
Air pressure (kPa) 100 – its output layer. The least problem-yielding ANN structure was
based on the back propagation algorithm to estimate the performance
of automotive air conditioning (AAC) systems. The experimental
measurements were taken while varying the compressor speed, cooling
capacity and condensing temperature. The experimental data were
used for training, testing and validation of an ANN model for the
system. The model was used for estimating various performance
parameters of the system, such as the compressor power, heat rejection
rate in the condenser, refrigerant mass flow rate, compressor discharge
temperature and the coefficient of performance of the system. The ANN
estimations for these parameters usually agreed well with the experi-
mental values with correlation coefficients in the range of 0.968–0.999,
mean relative errors in the range of 1.52–2.51% and very low root
mean square errors.
Hosoz and Direk [91] investigated the performance automotive air
conditioning system by operating it as an air-to-air heat pump using
ambient air as a heat source. Evaluation of the data gathered in steady
state test runs has shown the effects of the operating conditions on the
cooling capacity, coefficient of performance, compressor discharge
temperature and the rate of exergy destroyed by each component of
the system for different operating modes. It was found that the heat
pump operation provides adequate heating only in mild weather
conditions, and the heating capacity drops sharply with decreasing
outdoor temperature. Yilmaz and Atik [92] predicted the performance
of mechanical cooling cycle (MCC) by using an artificial neural network
(ANN) model. The network has three layers as input, output, and
hidden layer, has one input and four outputs. Training was continued
until the square error comes (≤ 0.005), for which back propagation
algorithm was used. The obtained results show that the low relative
error value, which indicates the accuracy of predictions by the ANN
model. Gandhidasan and Mohandes [93] presented a novel approach
based on artificial neural networks (ANNs) to predict the vapor
pressure of different desiccant solutions. The vapor pressure of
desiccants depends mainly on temperature and concentration.
accuracy. Moreover, the expressions currently in practice are tedious
and valid for narrow ranges and must be adjusted constantly. Neural
networks were trained to predict vapor pressure of desiccant solutions
with a reasonable accuracy without mathematical formulae. Trained
neural network models provided wide ranges of vapor pressure for
desiccant solutions without the need to cross reference several tables or
charts. Results showed potential of using ANNs for the prediction of
vapor pressure of desiccant solution for cooling applications. Modeling
of a mobile air conditioner system (MAC) for different mass flow rates
of refrigerant and compressor revolution speeds was carried out by Atik
et al. [94] by using artificial neural network (ANN). The three-layered
analyzed by examining the number of cells in hidden layer varies from
6 to 19. The best result was obtained in the ANN with 10 hidden cells.
A coefficient of correlation (R), 0.945 was obtained in estimating
cooling capacity, 0.985 in the power consumed in compressor, and
0.994 in coefficient of performance of the system. The obtained
correlation coefficients showed that ANNs can be used with a high
precision and accuracy in predicting the performance of mobile air
conditioner (MAC) systems. Table 5 shows the network performance
with various neuron numbers in the hidden layer.
Li et al. [95] proposed an artificial neural network based dynamic
model for variable speed direct expansion (DX) air conditioning
system. The average relative error (ARE) and maximum relative error
(MRE) were found as 0.33%, 0.27%, 0.27% and 0.89%, 0.99%, 1.15%,
respectively for three inputs, indicating the high accuracy of the ANN
based dynamic model. An ANN based controller was then developed as
shown in Fig. 12 for controlling the indoor air temperature and
humidity simultaneously by varying the compressor speed and supply
fan speed in a space served by the experimental system. The test results
showed that the ANN based controller was able to track the changes in
set points and to resist the disturbances.
Hernandez et al. [96] developed a direct and inverse artificial neural
network to estimate the coefficient of performance (COP) of a solar
refrigeration system. The network structure 6-6-1 (6 inputs, 6 hidden
and 1 output neurons) revealed an excellent agreement (R > 0.986)
between experimental and predicted values. The parameters such as
the solution concentration, the cooling water temperature, the genera-
tion temperature, the ambient temperature, the generation pressure
and the solar radiation were used as input to the model. The sensitivity
analysis shows that all studied input variables effect on the COP
prediction but the generation pressure is found the most influential
parameter (Fig. 13). The system performance was also examined by
inverting ANN model to evaluate the unknown input parameter from a
required COP. Because of the high precision and short computing time
makes this methodology very useful to simulate and to optimize the
solar refrigerator system.

360
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Fig. 11. The regression plot between ANN estimations and the experimental data base for (a) Training, (b) Validation, (c) Test, and (d) Combined set [29].

Table 5
The network performance with various neuron numbers in the hidden layer [94].

Number of hidden neurons MRE (%)

Q W COP Mean value

6 5.248 6.221 6.250 5.906

7 5.185 6.148 6.192 5.842
8 5.164 6.139 6.239 5.847
9 5.229 6.231 6.399 5.953
10 5.254 6.137 3.108 4.833
11 4.749 6.150 6.497 5.799
12 5.330 6.057 3.323 4.903
13 4.598 5.929 6.647 5.725
14 4.605 5.577 6.501 5.561
15 4.603 5.921 6.797 5.774
16 4.533 5.527 6.754 5.605
17 4.819 5.370 6.822 5.670
18 4.823 5.336 6.882 5.680
19 4.581 5.225 6.898 5.568

Fig. 14 shows comparison between ANN predicted COP and

experimental data base. It depicts that the COP prediction correlated
with high accuracy (R2 > 0.9875).
Kamar et al. [97] proposed an ANN model to estimate the cooling
capacity, compressor power input and the coefficient of performance
(COP) of the automotive air-conditioning (AAC) system. Compressor
speed, air temperature and velocity at evaporator inlet and air
Fig. 12. Schematic layout of the ANN based controller arrangement [95].
temperature at condenser inlet are used as input parameters to the
model. The network using Lavenberge- Marquardt (LM) architecture
was optimized for 4-3-3 (neurons in input-hidden-output layers)
configuration. The trained ANN model for the AAC system shows good

361
D.B. Jani et al.
Fig. 13. Influence of different input variables in estimation of the system COP [96].
Fig. 14. Comparison between ANN predicted COP and experimental data base [96].
performance with an error index in the range of 0.65–1.65%, mean
square error (MSE) between 1.09×10−5 and 9.05×10−5 and the root
mean square error (RMSE) in the range of 0.33–0.95%. Further, the
correlation which relates the predicted outputs of the ANN model to
the experimental results has a high coefficient (R) in guessing the AAC
system performance. Table 6 shows the effect of various input para-
meters on ANN model for five different cases. For each analysis error
index (EI) of model is evaluated and results are shown. The largest EI
depicts that the ANN model is less sensitive and has lower perfor-
mance. The EI is lowest when all parameters are included in the
analysis. EI is highest when compressor speed excluded from the
analysis.
Mora et al. [98] developed ANN model for estimation of thermo-
dynamic properties of refrigerants. It is found that ANNs are capable of
learning the complex relationships between input and output data, So,
it can be a good replacement of the commonly used equations of state
for thermodynamic properties estimation. Multilayer perceptron ANNs
with back-propagation algorithm were employed to obtain accurate
thermodynamic properties by use of estimation models. The developed
ANN model is found to predict thermodynamic properties of refriger-
ants preciously opening a promissory way to process optimization and
construction of intelligent devices, impacting in both cost and energy
savings. Fig. 15 shows variation in mean square error (MSE) over
different number of ANN model. It is found from the obtained results
that the lowest MSE is observed for model 2.
362
Renewable and Sustainable Energy Reviews 80 (2017) 352–366
Tian et al. [99] proposed artificial neural network (ANN) model to
predict performance of electric vehicle air-conditioning system
(EVACS) based on experimental data. Parameters such as refrigerant
mass flow rate, condenser heat rejection, refrigeration capacity and
compressor power consumption are used as input to the ANN model.
The ANN model was tested with two transfer functions (logsig and
tansig) and different hidden neurons (3–13) using Levernberg-
Marquardt algorithm. Comparison of ANN training performances
using Log-Sig transfer function and Tan-Sig transfer function are
showed in Fig. 16(a) and (b), respectively. Fig. 16 shows that ANN
model with Log-Sig transfer function has a faster convergence rate.
Number of epochs obtained for ANN model with Log-Sig and Tan-Sig
are 106 and 171, respectively. The optimized ANN was determined as a
configuration of 4-13-4 with Log-Sig transfer function, which demon-
strated the best capability with mean relative errors (MRE), root mean
square errors (RMSE) and correlation coefficients (R) in the range of
0.92–2.71%, 0.0044–0.0141 and 0.9975–0.9998, respectively.
Wang et al. [100] developed an ANN model to estimate the
performance of a hybrid ejector air-conditioning system (HEACS) by
use of three different types of neural networks, namely multi-layer
perceptron (MLP), radial basis function (RBF) and support vector
machine (SVM). The obtained result shows a good agreement between
estimated and experimentally measured value which indicates the
excellent ability of MLP. Sohani et al. [101] proposed an artificial
neural network (ANN) model to estimate the supply air temperature
and pad pressure drop in evaporative cooling system. The prediction
abilities of proposed model was analyzed and compared with analytical
models, and a comprehensive error analysis was also conducted. It was
found that the ANN model perform better than the other analytical
approaches for estimating the supply air temperature and the pad
pressure drop of evaporative cooling system. A summary of ANN
applications for different space cooling systems are listed in Table 7.
Zheng et al. [110] summarizes recent researches and developments
on novel solid desiccant materials that can be adopted in solid
desiccant based space cooling systems. Solid desiccant cooling systems
have gained increasing interest as an alternative air conditioning
technology. Performance of desiccant plays a crucial role in overall
performance of the whole system, especially in terms of dehumidifica-
tion and cooling capacity. Yang et al. [111] presents a hybrid artificial
intelligent control technique, including neural networks and genetic
algorithms, for the optimal control of the set points of the condensing
temperature to improve the coefficient of performance (COP) of air-
cooled chillers under various operating conditions. The neural network
is used to model the air-cooled chillers, and genetic algorithm is
adopted in searching optimal set points of condensing temperature
based on the predicted fitness values. Results show that this hybrid
control technique allows optimal set point of the condensing tempera-
ture to be successfully determined, and the chiller performance can be
improved considerably. Deb et al. [112] developed a methodology to
forecast diurnal cooling load energy consumption for institutional
buildings using data driven techniques. The results show that the
ANN is able to train and forecast the next day energy use based on five
previous days’ data with good accuracy. Workable operation strategies
for the desiccant enhanced evaporative cooling (DEVap) system were
suggested by Lee at al. [113] and applied to an energy simulation for
Table 6
Performance of ANN model with relative influence of various input parameters [97].
Neural network N_comp Teva_air_in Tcond_air_in Veva_air_in Error index
model (EI)
NN1 × √ √ √ 0.018
NN2 √ × √ √ 0.016
NN3 √ √ × √ 0.014
NN4 √ √ √ × 0.015
NN5 √ √ √ √ 0.013
D.B. Jani et al.
Fig. 15. Variation in MSE for different number of ANN model during validation [98].
estimating the energy saving potential of the system over the conven-
tional variable air volume (VAV) system. Guo et al. [114] reviewed a
recent advance which lowers the required heat source temperatures for
solar assisted solid desiccant space cooling to the range of 50–60 °C.
Rafael et al. [115] developed linear regression models and artificial
neural networks (ANN) to predict the heating and cooling energy
demands, energy consumptions and CO2 emissions of a building. The
multilayer perceptron adjusted over the variables without being
transformed, provides greater accuracy in the determination of the
demand, consumption and CO2 emissions both for heating and cooling,
offering ECM values closer to 0, with an R2 coefficient above 99%.
Erdemir and Ayata [116] have developed an artificial neural network
model to estimate decreasing in temperature of green roof of an office
building. The obtained results show 0.3982% RMSE and 99.05% R2.
6. Comparison between ANN and other soft computing and
statistical tools (SCST)
In the present review, thorough evaluation on parameters that
Fig. 16. Comparison of ANN model training results of (a) Log-Sig and (b) Tan-Sig transfer functions [99].
363
Renewable and Sustainable Energy Reviews 80 (2017) 352–366
having the potential influence on the performance prediction of the
solid desiccant cooling system (SDCS) system was carried out by use of
artificial neural network (ANN). The results produced by the ANN can
also be compared with that of the other soft computing and statistical
tools (SCST) to check the precision and accuracy for the prediction
[89]. They are group method of data handling (GMDH), genetic
programming (GP), multiple linear regression (MLR), and stepwise
regression method (SRM). These SCST models were trained and tested
using numerical and experimental data to estimate the performance of
desiccant enhanced evaporative cooling system (DEVAP). The dehu-
midifier stage was assumed to be incorporated separately into two
different configurations of counter flow indirect evaporative coolers
(IEC). For each stage, the best SCST models have been evaluated by
comparing estimated out put with experimental data via error criteria,
including the mean square error (MSE), and coefficient of determina-
tion (R2). It was found that the GMDH and SRM methods propose the
foremost models for determining the performance of the second stage.
Furthermore, SRM approach was found to be the best model in
guessing the performance of the dehumidifier.
Based on relative influence of various input parameters such as
inlet air temperature, relative humidity, velocity, dimensionless length
and working to inlet air ratio, the different model for determining
behavior of the performance parameters [117,118], outlet absolute
humidity and temperature, was compared on basis of training, test and
validation data. Tables (8a) and (8b) compared accuracy of different
SCST models on basis of mean squared error (MSE) while estimating
temperature and absolute humidity at the dehumidifier (DEH) exit.
The predicting correlations were also mentioned for different cases.
7. Limitations of artificial neural network modeling
The limitation of artificial neural network modeling is over training
of the network. Over training comes in to existence when the capacity
of artificial neural network for training is too high or too many training
iterations were allowed per network [119]. In most of the artificial
neural network applications, a very high training precision of 10−5 (or
large number of training cycles) is preset to terminate the training
processes. Nevertheless, in actual thermal systems, few training
samples are usually erroneous due to experimental uncertainties.
Hence, too high precision will over fit the training samples and degrade
the estimation performance of artificial neural network model. To
overcome the problem of over training with ANN, the numbers of
training data need to be normalized. While preparing the training data,
firstly, the maximum and minimum values of the input parameters are
selected from the experimentally measured data samples. Secondly,
some artificial training samples can be drawn from the existing
empirical formulae which cover the entire rage as much as possible.
D.B. Jani et al. Renewable and Sustainable Energy Reviews 80 (2017) 352–366

Table 7
Summary of ANN applications for different space cooling systems.

Authors [References] ANN network architectures Year Application

Bacthler et al. [107] GRBF 2001 Heat pump

Swider et al. [108] GRNN 2001 Chillers
Hosoz et al. [83] MLFFN 2006 Evaporative cooler
Saidur et al. [105] MLFFN 2006 Domestic refrigerator
Hosoz and Ertunc [106] MLFFN 2006 Cascade refrigeration
Yilmaz and Atik [92] MLFFN 2007 Mechanical cooling system
Ertunc and Hosoz [85] ANFIS 2008 Evaporative condensers
Parmar and Hindoliya [76] MLFFN 2011 Solid desiccant dehumidifier
Kiran and Rajput [87] ANFIS 2011 Indirect evaporative cooling
Koronaki et al. [88] MLFFN 2012 Evaporative solid desiccant cooling system
Zhu and Chen [79] GRNN 2015 Marine rotary desiccant air-conditioning
Jani et al. [29] MLFFN 2016 Solid desiccant - vapor compression hybrid air-conditioning system
Gawad et al. [81] ANFIS 2016 Heat recovery wheel in desiccant cooling system
Sohani et al. [89] GRNN 2016 Desiccant enhanced evaporative cooling system
Wei et al. [109] MLFFN 2016 Heat source tower heat pump system

Table 8a testing study or developing a complicated mathematical model with

Comparison between different SCST models for estimating air temperature at the enormous engineering efforts. The mathematical approach requires a
desiccant dehumidifier exit [89]. large number of geometrical parameters defining the system, which
Technique Training data Test data Validation data
may not be readily available and their predictions may not be
sufficiently accurate in many cases. As an alternative, use of ANN
R2 MSE R2 MSE R2 MSE requires less effort, time and cost to model the system. This new
modeling technique is used in many engineering applications, where
ANN 0.9981 0.0086 0.9976 0.0111 0.9962 0.01753
GMDH 0.9884 0.05372 0.9530 0.2183 – –
classical approaches are too complex to be used. So, ANN allows
GP 0.9725 0.1275 0.9154 0.3929 – – modeling of physical phenomena in complex systems without requiring
MLR 0.9814 0.0866 0.9285 0.3322 – – explicit mathematical representations or without requiring exhaustive
SRM 0.9880 0.0555 0.9759 0.1121 – – and costlier experiments. The cited literature in this review confirmed
that artificial neural network can be successfully used to predict
performance of solid desiccant cooling systems with acceptable accu-
Table 8b
racy. Through this review it has also been concluded that the use of
Comparison between different SCST models for estimating air humidity ratio at the
desiccant dehumidifier exit [89]. solid desiccant based space cooling technique is energy saving and
environment friendly approach for building air-conditioning.
Technique Training data Test data Validation data Numerous researchers have conducted feasibility study by using
2 2 artificial neural network approach in terms of its energy efficiency
R MSE R MSE R2 MSE
and cost effectiveness. Ongoing research and development works
ANN 0.9994 0.0029 0.9989 0.0050 suggests that use of artificial neural network for estimation of its
GMDH 0.9887 0.0524 0.9523 0.2215 – – performance have significant potential for improvement in its relia-
GP 0.9676 0.1506 0.9186 0.3784 – –
bility. Information presented in this paper may be beneficial to the
MLR 0.9852 0.0687 0.9000 0.4648 – –
SRM 0.9991 0.0043 0.9643 0.1657 – –
researchers working in the area of solid desiccant cooling and also
those using the ANN methodology in their studies.

The range of the training data must be representative of entire References

operating range of the equipment as much as possible. ANN is quite
suitable to the problems having enormous training data. The available
data can be split as training data, validation data and testing data.
Improper data splitting can lead to a poor estimation capability. Hence,
the selection of data for training and testing is more important in order
to attain high precision. Instrumental uncertainties and hypothetical
assumptions may effect the reliability of the results in case of handling
large training data. Hence, the number of training data required for the
network needs to be optimized. Novel optimization techniques may be
used to minimize the training data requirement.
The guessing capability of least square support vector machines
(LSSVM) is slightly better than traditional ANN models [59]. Hence,
the suitability of least square support vector machines for humidity
control in dehumidifier needs further research and investigation.
8. Conclusion
Because of high accuracy and short computing time, an approach of
ANN prediction can be very useful to simulate the performance of solid
desiccant cooling systems at different operating conditions by conduct-
ing limited number of experimental tests instead of comprehensive
[1] Tayagarajan T, Poonavaikko M, Shanmugam J, Panda RC, Rao PG. Artificial
neural networks: principles and application to model based control of drying
systems – a review. Dry Tech 1998;16:931–66.
[2] Zhang G, Zhang BE, Patuwo M, Hu Y. Forecasting with artificial neural networks.
State Art Int J Forecast 1998;14:35–62.
[3] Gardner MW, Dorling SR. Artificial neural networks (the multilayer perceptron) –
a review of applications in the atmospheric sciences. Atmos Environ
1998;32:2627–36.
[4] Mellit A, Kalogirou SA, Hontoria L, Shaari S. Artificial intelligence techniques for
sizing photovoltaic systems: a review. Renew Sustain Energy Rev 2009;13:406–19.
[5] Kalogirou S. Applications of artificial neural networks in energy systems: a review.
Energy Convers Manag 1999;40:1073–87.
[6] Kalogirou SA. Applications of artificial neural networks for energy systems. Appl
Energy 2000;67:17–35.
[7] Kalogirou SA. Artificial neural networks in renewable energy systems applications:
a review. Renew Sustain Energy Rev 2001;5:373–401.
[8] Ding GL. Recent developments in simulation techniques for vapour compression
refrigeration systems. Int J Refrig 2007;30:1119–33.
[9] Pennington NA. Humidity changer for air conditioning. USA Patent No. 2, 700,
537; 1955.
[10] Dunkle RV. A method of solar air conditioning. Mech Chem Eng Trans Inst Eng
1965;73:73–8.
[11] Munters CG. Inorganic, fibrous, gas-conditioning packing for heat and moisture
transfer, U.S. Patent 3, 377, 225; 1968.
[12] Banks PJ. Coupled equilibrium heat and single adsorbate transfer in fluid flow
through porous media - I, characteristic potentials and specific capacity ratios.

364
D.B. Jani et al.
Chem Eng Sci 1972;27:1143–55.
[13] Barlow R. Analysis of the adsorption process and desiccant cooling systems: a
pseudo-steady-state model for coupled heat and mass transfer. Technical report
no. SERI/TR-631-1330, Solar Energy Research Institute, Golden, CO; 1982.
[14] Jurinak JJ. Open cycle solid desiccant cooling-component models and system
simulation [PhD thesis]. Madison: University of Wisconsin; 1982.
[15] Worek WM, Lavan Z. Performance of a cross-cooled desiccant dehumidifier
prototype. J Sol Energy Eng 1982;104:187–96.
[16] Maclaine-Cross IL. Proposal for a desiccant air conditioning system. ASHRAE
Trans 1988;94:1997, [09].
[17] Davanagere BS, Sherif SA, Goswami DY. A feasibility study of solar desiccant air
conditioning system- Part I: psychrometrics and analysis of the conditioned zone.
Int J Energy Res 1999;23:7–21.
[18] Burns PR, Mitchell RB, Bechman WA. Hybrid desiccant cooling systems in
supermarket applications. ASHRAE Trans 1985;91:457–68.
[19] Van den Bulck E, Mitchell JW, Klein SA. The use of dehumidifiers in desiccant
cooling and dehumidification systems. Trans ASME 1986;108:684–91.
[20] Pesaran AA, Mills AF. Modelling of solid-side mass transfer in desiccant particle
beds. Sol Eng 1984:177–85.
[21] Jain S, Dhar PL, Kaushik SC. Evaluation of solid based evaporative cooling cycles
for typical hot and humid climates. Int J Refrig 1995;18:287–99.
[22] Daou K, Wang RZ, Xia ZZ. Desiccant cooling air conditioning: a review. Renew
Sustain Energy Rev 2006;10:55–77.
[23] Jani DB, Mishra M, Sahoo PK. Solid desiccant air conditioning – a state of the art
review. Renew Sustain Energy Rev 2016;60:1451–69.
[24] Rojas R. Neural networks: a systematic introduction. Germany: Springer; 1996.
[25] Yang KT. Artificial neural networks (ANN): a new paradigm for thermal science
and engineering. ASME J Heat Transf 2008;130:1–18.
[26] Mohammad AT, Mat SB, Sulaiman MY, Sopian K, Abduljalil AA. Artificial neural
network analysis of liquid desiccant dehumidifier performance in a solar hybrid
air-conditioning system. Appl Therm Eng 2013;59:389–97.
[27] Mohammad AT, Mat SB, Sulaiman MY, Sopian K, Abduljalil AA. Implementation
and validation of an artificial neural network for predicting the performance of a
liquid desiccant dehumidifier. Energy Convers Manag 2013;67:240–50.
[28] Gandhidasan P, Mohandes MA. Artificial neural network analysis of liquid
desiccant dehumidification system. Energy 2011;36:1180–6.
[29] Jani DB, Mishra M, Sahoo PK. Performance prediction of solid desiccant – vapor
compression hybrid air-conditioning system using artificial neural network.
Energy 2016;103:618–29.
[30] Jani DB, Mishra M, Sahoo PK. Performance prediction of rotary solid desiccant
dehumidifier in hybrid air-conditioning system using artificial neural network.
Appl Therm Eng 2016;98:1091–103.
[31] Hurdogan E, Buyukalaca O, Yılmaz T, Hepbasli A, Uckan R. Investigation of solar
energy utilization in a novel desiccant based air conditioning system. Energy Build
2012;55:757–64.
[32] Saeed A, Al-Alili A. A review on desiccant coated heat exchangers. Sci Technol
Built Environ 2016. http://dx.doi.org/10.1080/23744731.2016.1226076.
[33] Jani DB, Mishra M, Sahoo PK. Performance analysis of hybrid solid desiccant –
vapor compression air conditioning system in hot and humid weather of India.
Build Serv Eng Res Technol 2016;37:523–38.
[34] Parmar H, Hindoliya DA. Performance of solid desiccant based evaporative
cooling system in warm and humid climatic zone of India. Int J Eng Sci Technol
2010;2:5504–8.
[35] Bourdoukan P, Wurtz E, Joubert P. Comparison between the conventional and
recirculation modes in desiccant cooling cycles and deriving critical efficiencies of
components. Energy 2010;35:1057–67.
[36] Panaras G, Mathioulakis E, Belessiotis V, Kyriakis N. Theoretical experimental
investigation of the performance of a desiccant air-conditioning system. Renew
Energy 2010;35:1368–75.
[37] Li Y, Sumathy K, Dai YJ, Zhong JH, Wang RZ. Experimental study on a hybrid
desiccant dehumidification and air conditioning system. J Sol Energy Eng
2006;128:77–82.
[38] Jani DB, Manish Mishra, Sahoo . PK. Performance studies of hybrid solid
desiccant-vapor compression air-conditioning system for hot and humid climates.
Energy Build 2015;102:284–92.
[39] Jani DB, Mishra M, Sahoo PK. Experimental investigation on solid desiccant –
vapor compression hybrid air-conditioning system in hot and humid weather.
Appl Therm Eng 2016;104:556–64.
[40] Uckan I, Yilmaz T, Hurdogan E, Buyukalaca O. Experimental investigation of a
novel configuration of desiccant based evaporative air conditioning system. Energy
Convers Manag 2013;65:606–15.
[41] Bourdoukan P, Wurtz E, Joubert P. Experimental investigation of a solar desiccant
cooling installation. Sol Energy 2009;83:2059–73.
[42] Khalid A, Mahmood M, Asif M, Muneer T. Solar assisted, pre-cooled hybrid
desiccant cooling system for Pakistan. Renew Energy 2009;34:151–7.
[43] Fong KF, Chow TT, Lin Z, Chan LS. Simulation optimization of solar-assisted
desiccant cooling system for subtropical Hong Kong. Appl Therm Eng
2010;30:220–8.
[44] Zhang T, Liu X, Liu J. Performance investigation and exergy analysis of air-
handling processes using liquid desiccant and a desiccant wheel. Sci Technol Built
Environ 2016. http://dx.doi.org/10.1080/23744731.2016.1215689.
[45] Pandelidis D, Anisimova S, Worek WM, Drag P. Numerical analysis of a desiccant
system with cross-flow Maisotsenko cycle heat and mass exchanger. Energy Build
2016;123:136–50.
[46] Baniyounes AM, Liu G, Rasul MG, Khan MMK. Analysis of solar desiccant cooling
system for an institutional building in subtropical Queensland, Australia. Renew
365
Renewable and Sustainable Energy Reviews 80 (2017) 352–366
Sustain Energy Rev 2012;16:6423–31.
[47] Enteria N, Yoshino H, Satake A, Mochida A, Takaki R, Yoshie R, et al.
Development and construction of the novel solar thermal desiccant cooling system
incorporating hot water production. Appl Energy 2010;87:478–86.
[48] White SD, Kohlenbach P, Bongs C. Indoor temperature variations resulting from
solar desiccant cooling in a building without thermal backup. Int J Refrig
2009;32:695, [04].
[49] Gagliano A, Patania F, Nocera F, Galesi A. Performance assessment of a solar
assisted desiccant cooling system. Therm Sci 2014;18:563–76.
[50] Yadav YK, Kaushik SC. Psychrometric technoeconmic assessment and parametric
studies of vapor – compression and solid/liquid desiccant hybrid solar space
conditioning systems. Heat Rev Syst 1991;11:563–72.
[51] Bourdoukan P, Wurtz E, Joubert P, Spérandio M. Potential of solar heat pipe
vacuum collectors in the desiccant cooling process: modelling and experimental
results. Sol Energy 2008;82:1209–19.
[52] Li M, David I, Ursula E, Dennis L, Volker F. Cooling potential of ventilated PV
facade and solar air heaters combined with a desiccant cooling machine. Renew
Energy 2006;31:1265–78.
[53] Finocchiaro P, Beccali M, Nocke B. Advanced solar assisted desiccant and
evaporative cooling system equipped with wet heat exchangers. Sol Energy
2012;86:608–18.
[54] Khalid AJ, Nabeel SD. Application of solar assisted heating and desiccant cooling
systems for a domestic building. Energy Convers Manag 2001;42:995, [22].
[55] Kabeel AE. Solar powered air conditioning system using rotary honeycomb
desiccant wheel. Renew Energy 2007;32:1842–57.
[56] Khalid AD, Madhi SM. An experimental investigation into a solar assisted
desiccant-evaporative air-conditioning system. Sol Energy 1987;39:97, [07].
[57] Yadav A, Bajpai VK. Numerical and experimental investigation of operating
parameters of solar-powered desiccant wheel in India. Heat Trans 2013;42:1–30.
[58] Wang H, Cai W, Wang Y. Modeling of a hybrid ejector air conditioning system
using artificial neural networks. Energy Convers Manag 2016;127:11–24.
[59] Zendehboudi A. Implementation of GA-LSSVM modelling approach for estimating
the performance of solid desiccant wheels. Energy Convers Manag
2016;127:245–55.
[60] Comino F, Ruiz , de Adana M. Experimental and numerical analysis of desiccant
wheels activated at low temperatures. Energy Build 2016;133:529–40.
[61] Seung JO, Kim CN, Thu K, Ja MK, Islam MR, Chun W, Chua KJE. Studying the
performance of a dehumidifier with adsorbent coated heat exchangers for tropical
climate operations. Sci Technol Built Environ 2016. http://dx.doi.org/10.1080/
23744731.2016.1218234.
[62] Jani DB, Mishra M, Sahoo PK. Exergy analysis of solid desiccant–vapour
compression hybrid air conditioning system. Int J Exergy 2016;20:523–38.
[63] La D, Dai YJ, Li Y, Wang RZ, Ge TS. Technical development of rotary desiccant
dehumidification and air conditioning: a review. Renew Sustain Energy Rev
2010;14:130–47.
[64] Angrisani G, Minichiello F, Roselli C, Sasso M. Experimental investigation to
optimise a desiccant HVAC system coupled to a small size cogenerator. Appl
Therm Eng 2011;31:506–12.
[65] Ge TS, Ziegler F, Wang RZ, Wang H. Performance comparison between desiccant
cooling system and vapor compression system. Appl Therm Eng 2010;30:724–31.
[66] Fong KF, Chow TT, Lee CK, Lin Z, Chan LS. Solar hybrid cooling system for high-
tech offices in subtropical climate: radiant cooling by absorption refrigeration and
desiccant dehumidification. Renew Energy 2011;52:2883–94.
[67] Charoensupaya D, Worek WM. Parametric study of an open-cycle adiabatic, solid,
desiccant cooling system. Energy 1988;13:739–47.
[68] Mittal V, Khan BK. Experimental investigation on desiccant air-conditioning
system in India. Front Energy Power Eng 2010;4:161–5.
[69] Subramanyam N, Maiya MP, Murthy SS. Application of desiccant wheel to control
humidity in air-conditioning systems. Appl Therm Eng 2004;24:2777–88.
[70] Jia CX, Dai YJ, Wu JY, Wang RZ. Analysis on hybrid desiccant air-conditioning
system. Appl Therm Eng 2006;26:2393, [00].
[71] Curtiss PS, Brandemuehl MJ, Kreider JF. Energy management in central HVAC
plants using neural networks. ASHRAE 1995:199–216.
[72] Kah AH, San QY, Guan SC, Kiat WC, Koh YC. Smart air conditioning system using
multilayer perceptron neural network with a modular approach. In: Proceedings of
the IEEE international conference ICNN '95, Perth, Western Australia; 1995, p.
2314–9.
[73] Datta D, Tassou SA, Marriott D Application of neural networks for the prediction
of the energy consumption in a supermarket. In: Proceedings of CLIMA 2000
conference, Brussels, Belgium; 1997, p. 98.
[74] Kajl S, Roberge MA, Lamarche L, Malinowski P Evaluation of building energy
consumption based on fuzzy logic and neural networks applications. In:
Proceedings of CLIMA 2000 conference, Brussels, Belgium; 1997, p. 264.
[75] Cejudo JM, Moreno R, Carrillo A. Physical and neural network models of a silica-
gel desiccant wheel. Energy Build 2002;34:837–44.
[76] Parmar H, Hindoliya DA. Artificial neural network based modelling of desiccant
wheel. Energy Build 2011;43:3505–13.
[77] Mahmoud O, Abd El-Rahim D, Fatouh M. Artificial neural network model of a
silica-gel desiccant bed during dehumidification and reactivation operating modes.
In: Proceedings of the eleventh international conference of fluid dynamics.
Alexandria, Egypt; 2013, p. 1-9.
[78] Uçkan I, Yılmaz T, Hürdoğan E, Büyükalaca O. Development artificial neural
network model for the prediction of the performance of a silica-gel desiccant
wheel. Int J Green Energy 2014. http://dx.doi.org/10.1080/
15435075.2014.895733.
[79] Zhu J, Chen W. A novel multivariate linear prediction model for the marine rotary
D.B. Jani et al.
desiccant air-conditioning by adding a dynamic correction factor. Appl Therm Eng
2015;78:101–9.
[80] Rafique MM. A statistical analysis of desiccant dehumidifier for air conditioning
application. Int J Hybrid Inform Technol 2015;8:245–56.
[81] Gawad AFA, Radhwi MN, Wafiah AM, Softah GJ. Computational and artificial
intelligence study of the parameters affecting the performance of heat recovery
wheels. Am J Energy Eng 2015;3:79–94.
[82] Zendehboudi A. Implementation of GA-LSSVM modelling approach for estimating
the performance of solid desiccant wheels. Energy Convers Manag
2016;127:245–55.
[83] Ertunc HM, Hosoz M. Artificial neural network analysis of a refrigeration system
with an evaporative condenser. Appl Therm Eng 2006;26:627–35.
[84] Yigit KS, Ertunc HM. Prediction of the air temperature and humidity at the outlet
of a cooling coil using neural networks. Int Commun Heat Mass Transf
2006;33:898–907.
[85] Ertunc HM, Hosoz M. Comparative analysis of an evaporative condenser using
artificial neural network and adaptive neuro-fuzzy inference system. Int J Refrig
2008;31:1426–36.
[86] Hosoz M, Ertunc HM, Ozguc AF. Modelling of a direct evaporative air cooler using
artificial neural network. Int J Energy Res 2008;32:83–9.
[87] Kiran TR, Rajput SPS. An effectiveness model for an indirect evaporative cooling
(IEC) system: comparison of artificial neural networks (ANN), adaptive neuro-
fuzzy inference system (ANFIS) and fuzzy inference system (FIS) approach. Appl
Soft Comput 2011;11:3525–33.
[88] Koronaki IP, Rogdakis E, Kakatsiou T. Thermodynamic analysis of an open cycle
solid desiccant cooling system using artificial neural network. Energy Convers
Manag 2012;60:152–60.
[89] Sohani A, Sayyaadi H, Balyani HH, Hoseinpoori S. A novel approach using
predictive models for performance analysis of desiccant enhanced evaporative
cooling systems; 2016. 〈http://doi.org/10.1016/j.applthermaleng.2016.06.121〉.
[90] Hosoz M, Ertunc HM. Artificial neural network analysis of an automobile air
conditioning system. Energy Convers Manag 2006;47:1574–87.
[91] Hosoz M, Direk M. Performance evaluation of an integrated automotive air
conditioning and heat pump system. Energy Convers Manag 2006;47:545–59.
[92] Yilmaz S, Atik K. Modeling of a mechanical cooling system with variable cooling
capacity by using artificial neural network. Appl Therm Eng 2007;27:2308–13.
[93] Gandhidasan P, Mohandes MA. Predictions of vapor pressures of aqueous
desiccants for cooling applications by using artificial neural networks. Appl Therm
Eng 2008;28:126–35.
[94] Atik K, Aktas A, Deniz E. Performance parameters estimation of MAC by using
artificial neural network. Expert Syst Appl 2010;37:5436–42.
[95] Li N, Xia L, Shiming D, Xu X, Chan MY. Dynamic modeling and control of a direct
expansion air conditioning system. Appl Energy 2011;91:290–300.
[96] Hernandez JA, Rivera W, Colorado D, Moreno-Quintanar G. Optimal COP
prediction of a solar intermittent refrigeration system for ice production by means
of direct and inverse artificial neural networks. Sol Energy 2012;86:1108–17.
[97] Kamar HM, Ahmad R, Kamsah NB, Mustafa AFM. Artificial neural networks for
automotive air-conditioning systems performance prediction. Appl Therm Energy
2013;50:63–70.
[98] Mora JE, Carlos PT, Gonzalez FN, Ocampo JD. Thermodynamic properties of
refrigerants using artificial neural networks. Int J Refrig 2014;46:9–16.
[99] Tian Z, Qian CH, Gu B, Yang L, Liu F. Electric vehicle air conditioning system
performance prediction based on artificial neural network. Appl Therm Eng
2015;89:101–14.
366
Renewable and Sustainable Energy Reviews 80 (2017) 352–366
[100] Wang H, Cai W, Wang Y. Modeling of a hybrid ejector air conditioning system
using artificial neural networks. Energy Convers Manag 2016;127:11–24.
[101] Sohani A, Zabihigivi M, Moradi MH, Sayyaadi H, Balyani HH. A comprehensive
performance investigation of cellulose evaporative cooling pad systems using
predictive approaches. Appl Therm Eng 2017;110:1589–608.
[102] Mohanraj M, Jayaraj S, Muraleedharan C. Applications of artificial neural
networks for refrigeration, air-conditioning and heat pump systems—a review.
Renew Sustain Energy Rev 2012;16:1340–58.
[103] Mohanraj M, Jayaraj S, Muraleedharan C. Applications of artificial neural
networks for thermal analysis of heat exchangers – a review. Int J Therm Sci
2015;90:150–72.
[104] Sahlot M, Riffat SB. Desiccant cooling systems: a review. Int J Low-Carbon
Technol 2016;11:489–505.
[105] Saidur R, Masjuki HH, Jamiludhin MY. A new method to investigate the energy
performance of a household refrigerator-freezer. Int Energy J 2006;7:9–15.
[106] Hosoz MH, Ertunc M. Modeling of a cascade refrigeration system using artificial
neural networks. Int J Energy Res 2006;30:1200–15.
[107] Bacthler H, Browne MW, Bansal PK, Kecman V. Neural networks—a new
approach to model vapour compression heat pumps. Int J Energy Res
2001;25:591–9.
[108] Swider DJ, Browne MW, Bansal PK, Kecman V. Modeling of vapour compression
liquid chillers with neural networks. Appl Therm Eng 2001;21:311–29.
[109] Wei X, Li N, Peng J, Cheng J, Su L, Hu J. Analysis of the effect of the CaCl2 mass
fraction on the efficiency of a heat pump integrated heat-source tower using an
artificial neural network model. Sustainability 2016;8:1–14.
[110] Zheng X, Ge TS, Wang RZ. Recent progress on desiccant materials for solid
desiccant cooling systems. Energy 2014;1:1–15.
[111] Yang J, Chan KT, Dai T, Yu FW, Chen L. Hybrid artificial neural network – genetic
algorithm technique for condensing temperature control of air – cooled chillers.
Procedia Eng 2015;121:706–13.
[112] Deb C, Eang LS, Yang J, Santamouris M. Forecasting diurnal cooling energy load
for institutional buildings using artificial neural network. Energy Build
2015;1:1–14.
[113] Lee SJ, Kim HJ, Dong HW, Jeong JW Energy saving assessment of a desiccant –
enhanced evaporative cooling system in variable air volume applications. Appl
Therm Eng. 〈http://dx.doi.org/10.1016/j.applthermaleng.2017.02.007〉.
[114] Guo J, Lin S, Bilbao JI, White SD, Sproul AB. A review of photovoltaic thermal
(PV/T) heat utilisation with low temperature desiccant cooling and dehumidifi-
cation. Renew Sustain Energy Rev 2017;67:1–14.
[115] Rafael PM, Alexis PF, Carlos RB, Jesus APA. Comparison of linear regression and
artificial neural networks models to predict heating and cooling energy demand,
energy consumption and CO2 emissions. Energy 2017;118:24–36.
[116] Erdemir D, Ayata T. Prediction of temperature decreasing on a green roof by using
artificial neural network. Appl Therm Eng 2017. http://dx.doi.org/10.1016/
j.applthermaleng.2016.10.145.
[117] Alaidroos A, Krarti M. Numerical modeling of ventilated wall cavities with spray
evaporative cooling system. Energy Build 2016;130:350–65.
[118] Speerforck A, Schmitz G. Experimental investigation of a ground-coupled
desiccant assisted air conditioning system. Appl Energy 2016;181:575–85.
[119] Yin C, Rosendahl L, Luo Z. Methods to improve prediction performance of ANN
models. Simul Model Pract Theory 2003;11:211–22.