Applications of software in solar drying systems: A review

Prashant Singh Chauhan a, Anil Kumar b,n, Perapong Tekasakul b

a Energy Centre, Maulana Azad National Institute of Technology, Bhopal 462051, Indiab Energy Technology Research Center, Department of Mechanical Engineering, Faculty of Engineering, Prince of Songkla University, Hat Yai,Songkhla 90112, Thailand

a r t i c l e i n f o

Article history:Received 30 January 2014Received in revised form29 June 2015Accepted 8 July 2015Available online 30 July 2015

Keywords:Solar dryerSimulation modelingCFDCOMSOL multi physicsFORTRANMATLABSPSS

a b s t r a c t

This review paper is focused on the application of software in solar drying systems. The application ofsoftware is very important to develop and analyze the mathematical models and predicting theperformance of different kinds of solar drying systems. It is also useful for predicting the croptemperature, moisture content and drying rate, drying kinetics, and color of the crop. Computationalfluid dynamics can be used for the analysis and investigation of air flow and temperature distributionpattern through appropriate simulation with the help ANSYS and FLUENT. MATLAB and FORTRAN arevery useful tools to develop mathematical models for prediction the crop temperature, air temperature,the moisture evaporated. It is also very useful for training and testing of various models. For statisticaldata analysis, statistical software SPSS, Sigma Plot V and Statistica. All recent employed software andtheir utility in solar drying systems are emphasized in this communication. This comprehensive reviewof the various software applications in different solar drying systems is useful for academician, scientistand researchers.

& 2015 Elsevier Ltd. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13272. Simulation methodologies of different solar dryers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1327

2.1. Direct solar dryer. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13272.1.1. CFD simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13272.1.2. FORTRAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13282.1.3. MATLAB simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13282.1.4. SPSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13292.1.5. Statistica . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13292.1.6. TRNSYS simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1330

2.2. Indirect solar dryer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13302.2.1. CFD simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13302.2.2. Comsol multiphysics simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13302.2.3. MATLAB simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1331

2.3. Mixed mode solar dryer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13312.3.1. FORTRAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13312.3.2. MATLAB simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13312.3.3. Sigma plot V. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1333

2.4. Hybrid solar dryer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13332.4.1. CFD simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13332.4.2. FORTRAN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13342.4.3. MATLAB simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13342.4.4. SPSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1334

3. Case-study of MATLAB software based ANN model used in solar drying. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1335

Contents lists available at ScienceDirect

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

Renewable and Sustainable Energy Reviews

http://dx.doi.org/10.1016/j.rser.2015.07.0251364-0321/& 2015 Elsevier Ltd. All rights reserved.

n Corresponding author. Tel: þ66 95 043 9186.E-mail address: anilkumar76@gmail.com (A. Kumar).

Renewable and Sustainable Energy Reviews 51 (2015) 1326–1337

3.1. ANN model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13353.2. Development of ANN model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13363.3. Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1336

4. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1336References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1336

1. Introduction

Application of solar energy has been old since the existence ofhuman being on the earth. At present, the way of life of the peopleis dependent on the production and utilization of energy; as aresult, the demand and supplying of energy is increasing in humansocieties. Presently, 77% of the world's total energy is supplied byfossil fuels, which release polluting and greenhouse gases, bydegrading ozone layer excessively threatens environment andcontributes to more global warming. Therefore, in order to main-tain an environment, considering alternate energy sources hasbecome an essential mission [1]. Every day earth receives thou-sands of times more energy from the sun than it consumed fromall other resources. Solar energy has plenty of potential to fulfillour energy demand. Solar drying is one among the applications ofutilization of solar energy. Solar drying is one of the oldestmethods of preservation of crops and it is utilized everywhere[2]. Solar drying is a dual process of heat transfer to the productfrom the heat source and mass transfer in the form of moisture,from the product to its surface and from the surface to thesurrounding air [37].

Solar dryers are available in the variety of design and size basedon drying capacity. To test a dryer, it is essential to evaluate itscomplete and relative performance with the other dryers. The testresults give the related information to the researchers, manufac-turers and end users [3]. The application of software is veryimportant to develop and analyze the mathematical models andpredicting the performance of different kind of solar dryingsystems. The design of solar drying can be optimized with the helpof software and it saves time which consumed during experiments.It is also useful for predicting the crop temperature, moisturecontent and drying rate, drying kinetics, texture and color of thecrop. Computational fluid dynamics (CFD) can be used for theanalysis and investigation of air flow, air flow rate inside the solardryer, temperature distribution pattern and humidity, throughappropriate simulation of energy and momentum equations and

heat and mass transfer in both gaseous and solid phases [4].MATLAB is very useful tool for developing mathematical modelsto predict the crop temperature, air temperature, the moistureevaporated and for predicting the thermal performance of the solardryer. It is also very supportive of training and testing of variousmodels [5]. Statistical software SPSS is an important tool forstatistical data analysis of any solar dryer. It computes the coeffi-cient of determination (R2), reduced chi-square, and the percentageof root mean square error (RMSE) which can be used for selectingthe best-fit equation to describe the drying process. Another soft-ware Statistica can also be used for statistical analysis. Sigma Plot Vsoftware is used for data fitting [6]. TRNSYS software is applied formodeling and describes the drying behavior [7].

The aim of this review article is to provide the information of theexisting software applied in solar drying, simulation procedures andoptimization techniques to the researchers. At present there is not asingle available source which provides such type of information tothe researchers and scientists working in solar drying. The selectionof a dryer for a particular crop drying is a main challenge. Thiscommunication introduces a comprehensive review based on theapplication of different kinds of analysis and performance evalua-tion software for different solar drying systems.

2. Simulation methodologies of different solar dryers

Solar dryers are classified broadly into four categories such asdirect, indirect, mixed mode solar dryer and hybrid solar dryer.Simulation methodologies of these dryers with the help ofdifferent softwares have been discussed as:

2.1. Direct solar dryer

In the direct solar drying systems, crop is exposed to sunlightdirectly such that it can be dehydrating. With this type of dryingsystem a black painted heat absorbing surface is provided that cancollect the sunlight and converts it into heat; the crop to be driedis placed directly on this surface. These dryers may have glass lidcovers and vents to in order to increase efficiency [8]. The cabinetsolar dryer is a kind of direct solar dryer which is shown in Fig. 1.

2.1.1. CFD simulationMathioulakis et al. constructed an industrial batch-type, tray

dryer for the drying of fruits. CFD FLUENT software was used tosimulate the air movement inside the drying chamber. Threeboundary conditions were assumed for simulation. In the firstcondition, fixed-mass-inflow boundary condition was assumedat the inlet (25 m s�1 velocity and 4.0% turbulence intensity). Inthe second condition, no-resistance boundary condition wasimposed where as mass is allowed to leave the solution domainat the outlet. In third condition, a wall shear stress condition wasassumed on surfaces bounding domain. The variation in thedryness in several trays was observed. The non-uniformitywas also traced to in certain areas of the chamber. In this articleCFD FLUENT software was used to predict the air velocities inthe drying chamber and optimizing the drying condition andFig. 1. Direct solar dryer [8].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–1337 1327

performance of the unit. The data obtained from the CFD anddrying tests showed good correlation between the air velocity andthe drying rate. Hence CFD FLUENT may be used as a dryingoptimization tool [9].

Bartzanas et al. used computational fluid dynamics (CFD),FLUENT v.5.3.18 software to understand the effect of vent arrange-ment for air ventilation of a tunnel greenhouse dryer. Numericalinvestigation was done by using a commercial CFD code for theeffect of ventilation configuration of a tunnel greenhouse throughcrop airflow and different temperature patterns. The mathematicalmodel was validated against experimental data. The airflowpatterns were recorded by using a three-dimensional sonic anem-ometer and the greenhouse ventilation rate was derived by using atracer gas technique. CFD model was used to study the conse-quences of four different ventilator configurations of the naturalventilation system. It was found that the ventilation configurationaffects the ventilation rate of the greenhouse and air temperaturedistributions. The observed values for the different configurationsand computed ventilation rates were varied from 10 to 58 airchanges per hour for an outside wind speed of 3 ms�1 and for awind direction perpendicular to the openings. The simulationsdraw attention to the mean air temperature at the middle of thetunnels which varied from 28.2 to 29.88 1C for an outside airtemperature of 28 1C. In the crop cover the average air velocityvaried according to the placement of the vents from 0.2 to 0.7 m/s.In this paper CFD FLUENT software was used to select the exactlocation of ventilation holes for proper air movement inside thedryer [10].

Chen et al. used CFD FLUENT software for optimal design ofsolar energy-assisted photocatalytic closed type dryer by means ofnumerical experiments [4]. The flow field of drying chamber wasgoverned by the Reynolds-averaged equations of continuity andmomentum employed for incompressible and steady flow. Thegoverning equations with boundary conditions were solvednumerically. A uniform inlet velocity profile was used to set thevelocity at the inlet flow boundary. The turbulence intensity of theinlet streamwas assumed to be 10% and the turbulent length scalewas considered to be equal to the upstream duct diameter. Thecomputational domain was discretized with structural hexahedralmeshes included around 600,000 cells for ensuring a goodresolution of the mathematical results.

Krawczyk and Badyda applied fluent computational fluiddynamics software to develop a mathematical model for sewagedrying process in a forced convection solar greenhouse dryer. Theunsteady condition was considered for thermal and flow processesinside the solar dryer because of the thermodynamic character-istics of the sludge and drying conditions (solar radiation, tem-perature and humidity of ventilated air, change over time) [11].

Amjad et al. used ANSYS-FLUENT CFD based flow simulationsoftware for predicting the air distribution in drying chamber ofbatch type dryer for potato slice having a thickness of 4 mm. Adiagonally airflow inlet channel along with the length of dryingchamber have been proposed for this dryer. The simulated resultsof airflow and experimental value were found a good agreement interms of coefficient of correlation, i.e. 87.09% for airflow distribu-tion [12].

2.1.2. FORTRANMahapatra et al. used FORTRAN to develop a computer simula-

tion program to study various process and performance of atunnel-type solar dryer with an integrated collector. The heatand mass balance equations for the air flow through the dryingspace and multi-layer mass transport model of the static layer ofthe material under drying were considered in heat flow networkmodel for this drying system [13].

Mortaza et al. used Compaq Visual FORTRAN programminglanguage for writing thin-layer drying equations and equilibriumdrying model of deep-bed solar greenhouse drying. Experimentaldata of chamomile obtained at different specified loads such as 15,30, 45, and 60 kg/m2. Equilibrium drying model computed tem-perature profiles, moisture content of the material, humidity ratioprofiles and the temperature of drying air. A good agreement wasfound between simulated and experimental results of moisturecontent for deep-bed solar greenhouse drying of chamomile. Theroot normalized mean square error was lower than 9.3% [14].

2.1.3. MATLAB simulationChyi et al. developed MATLAB based software TLDRY for thin-

layer drying to examine the applicability of the equations in ASAES 448. TLDRY software was found to be best and convenient toolfor the Newton and Page model [15]. Seginera and Bux were usedMATLAB to develop a drying rate model for waste water sludge forgreenhouse dryer [16].

Kumar and Tiwari developed a program in MATLAB software toestimate the temperature of the greenhouse and jaggery, moistureevaporated of the jaggery and to predict the thermal performanceof the greenhouse on the basis of ambient parameters under naturalconvection greenhouse conditions for the drying of jaggery. Pro-posed thermal model was validated with the experimental obser-vations. The predicted values and experimental observations werefound in good agreement with a coefficient of correlation 0.90 and0.98 for jaggery temperature and greenhouse air temperaturerespectively and 0.96 to 1.00 for the jaggery mass during drying.It was concluded that the thermal model will be beneficial todesign, greenhouse dryer for a given mass of jaggery [5].

Tiwari et al. applied the MATLAB 7.0 software to solve themathematical model which developed from energy balance equa-tion for fish drying in greenhouse dryer. The necessary climaticparameters were the ambient air temperature, solar intensity, andrelative humidity inside the greenhouse dryer for developing themodel. The output of the program was given the hourly averagefish surface temperature and greenhouse room air temperature.The coefficient of correlation and root mean square of percentdeviation had been used to determine the closeness of thepredicted and experimental values of fish surface temperatureand greenhouse room air temperature. The predicted values werein good agreement with experimental values. The range ofcoefficient of correlation and root mean square percent deviationwere in between 0.94–0.99 and 2.4–10% respectively [17].

Prakash and Kumar used MATLAB software to develop an ANFISmodel to envisage the moisture evaporated greenhouse air tem-perature and the jaggery temperature in a natural convectiongreenhouse drying system. This ANFIS model was validatedthrough the experimental observations for the complete dryingof jaggery. The resulting network structural design is known asANFIS. In Fig. 2, x and y was taken as the input vector. The firingstrengths w1 and w2 were obtained from the product of themembership grades in the prescribed part. E1 and E2 were takenas the ratio of firing strength to the total of all firing strengths. The‘f’ was output and it was the average weighted of each rule. Forthis ANFIS model, the coefficient of correlation was rangedbetween 0.999 and 1 and percentage error of root mean squareranged between 0.7% and 1.0% for jaggery and greenhouse airtemperature. The comparison between ANFIS model and thethermal model shows, the ANFIS model is better than the thermalmodel. They were also used to forecast the thermal performance ofthe greenhouse on the basis of ambient parameters. Hence, theyfound that the use of the ANFIS model may be the best way fordeveloping the natural convection greenhouse dryer for a givenmass of jaggery with a thin layer [18].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–13371328

Prakash and Kumar applied MATLAB version R 2010a softwarefor the training and testing of the artificial neural network model(ANN) for prediction of jaggery mass during drying inside thenatural convection greenhouse dryer. They trained the ANN modelfrom 1 to 40 neurons in the hidden layers and after that, increasingthe size of the network in gradually and it is found that the ANNmodel, execute best for 20 neurons because of this condition,RMSE value was 0.714812 and coefficient of determination (R2)was 0.999948. During the experimentation, it is observed thatthere was a drastic change in the value of RMSE and R2 valuesbetween 10 and 20 neurons. The intention was to find a mostsuitable transfer function in order that ANN model predicts theexperimental data in maximum possible accuracy [19].

2.1.4. SPSSHossain and Bala used statistical analysis software SPSS 9.0 in

the solar tunnel dryer drying of hot chili. The color values andpungency indices of solar, improved sun dried and conventionalsun dried chilies were statistically analyzed using a randomizedblock design. The data of color values and pungency indices of redand green chilies obtained experimentally were examined byanalysis of variance through SPSS 9.0 software. The mean differ-ences of color values and pungency indices were graded byDuncan's Multiple Range Test (DMRT) [20].

Tunde-Akintunde used SPSS (Statistical Package for socialscientists) 11.5.1 software for fitting the models, i.e. Newtonmodel, Henderson and Pabis model, Page's model and Logarithmicmodel for the experimental data in their linearized form usingregression technique. The non-linear regression analysis was alsoperformed by SPSS 11.5.1 software package. The coefficient ofdetermination (R2) was one of the most important criteria forselecting the best fit equation. Besides to the coefficient ofdetermination (R2), the goodness of fit was determined by differ-ent statistical parameters, for example, reduced mean square ofthe deviation, and root mean square error (RMSE). For the best fit,the coefficient of determination value should be higher even asand the root mean square error's values should be lower. After theanalysis it was concluded that for solar drying, the range of R2,RMSE and χ2�10�3 varied 0.9286–0.9460, 5.091–7.494, 0.0774–0.0638 respectively for Henderson and Pabis model. Whereas thecorresponding values for Newton model, Logarithmic model andPage model were 0.9876–0.917, 8.729–13.01, 0.0836–0.1010;0.9463–0.9313, 9.025–12.75, 0.0901–0.1082; 0.9885–0.9900,1.332–1.109, 0.0322–0.0294 respectively [21].

2.1.5. StatisticaMidilli and Kucuk did the mathematical modeling of thin layer

drying of shelled and unshelled pistachio samples in a mix modesolar dryer under forced and natural convection mode. Statistica

Fig. 2. First-order Sugeno fuzzy inference system [18].

Fig. 3. Flow chart of the TRNSYS module wood drying [24].

Fig. 4. Indirect solar dryer [8].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–1337 1329

Computer Program was used for non-linear regression analysis ofeight different mathematical models and found that the logarith-mic model, possibly will sufficiently describe thin layer forcedsolar drying of shelled and unshelled pistachio. The coefficients ofdetermination and reduced chi square of shelled and unshelledpistachio were 0.9983, 2.697�105 and 0.9990, 1.639�105 respec-tively for thin layer forced solar drying [22].

2.1.6. TRNSYS simulationReuss et al. applied TRNSYS software for modeling of wood

drying and the behavior of a natural convection chimney in mixmode dryer. The transient conditions of the drying air were takenas inputs to the module. Material properties, like moisture trans-port coefficients and sorption isotherms, modeling could beadapted to different wood varieties. The output included thetemperatures of the wood, moisture values, the air as well asdrying and heat transfer rates. Fig. 3 shows a flow chart of theTRNSYS module “Wood Drying”. The computation of the dryingprocess was divided into 60 segments. A time step of 15 s wasselected. The segment width followed from the division into 60segments and the geometry of the staple. The goal of optimizingthe dryer was to constantly reach a high output of high qualitydried wood. Modeling makes simple and fast simulation as areplacement for of difficult and time-consuming experiments [23].

Awadalla et al. applied TRNSYS program in solar wood dryer atWood Research Institute of Munich, Germany. TRNSYS is a widelyused as scientific simulation tool in solar energy applications. Thewood drying process was investigated theoretically under transi-ent conditions. The finite element method was also used to solvethe set of governing equations. For validation of the present model,the wood model was executed by the TRNSYS program withexperimental data of wood drying. There was deviation in theore-tical and experimental results of wood average moisture contentranges from �12% to þ14% and the deviation of wood averagetemperature ranges from �0.5% to þ3%. The results of steadystate and transient conditions were in good agreement withexperimental and theoretical works and simulation model verifiedthat it can be an effective tool for the design of a solar timber dryerand the prediction of moisture content behavior [24].

Mortaza et al. used TRNSYS software for effective and innova-tive modeling and simulation of deep-bed solar greenhouse dryerfor biomaterials. The moisture content and temperature profiles ofthe material and the temperature and humidity ratio profiles of

drying air were computed using equilibrium model. Author foundgood agreement between experimental moisture content andsimulation results for deep-bed solar greenhouse drying of cha-momile with root mean square error which was lower than 9.3%.The findings of this exercise confirmed that TRNSYS is a powerfultool for optimizing of solar dryer operation and design [25].

2.2. Indirect solar dryer

In indirect solar dryers, the black painted heat absorbingsurface heats the ambient air, instead of direct exposure of cropto solar radiation. This heated air is subsequently passed throughthe crop, taking moisture and exit through a chimney [8]. Adiagram of an indirect solar dryer is shown in Fig. 4.

2.2.1. CFD simulationRomero et al. designed and fabricated an indirect solar dryer of

50 kg drying capacity for vanilla drying at Universidad del Caribein Cancun, Mexico. They used FLUENT ANSYS software for simula-tion and validation. The temperature distribution analysis for solardryer was done through CFD where input taken as inlet and outlettemperature of the cabinet and inlet temperature of solar collector.Three dimensional, transient and laminar flow was considered forCFD simulation. Temperature, pressure and speed of working fluidare necessary parameters to solve the various governing conserva-tion equations. The following steps were applied to solve thephysical phenomena using ANSYS-FLUENT code:

Step-1 Design the geometry and discretization of the controlvolume using ANSYS design modeler program.Step-2 Give the specification of construction material proper-ties and boundary conditions for each system element.Step-3 Solution of the equations in each element of the meshfor specified each time interval.Step-4 Graphical representation of the results.

A superior degree of harmony was found between CFD simula-tion and measured parameters for solar collector, whereas in caseof cabinet a little variation was found between measured andestimated temperatures. This variation was found because ofconsideration of constant convection heat transfer coefficient inthe ambient. It was concluded that it is necessary to define avariable convection heat transfer coefficient as a function of thetime along the day for predicting thermal parameter. Finally,approximately 62% (initial and final weight of vanilla were1267.5 and 491.2 g, respectively) weight reduction of the vanillawas found in one month during the cabinet drying instead of threemonths in case of conventional drying [26].

2.2.2. Comsol multiphysics simulationVintila et al. applied Comsol Multiphysics CFD software for

numerical simulation in an indirect solar dryer. In this experimentthe numerical simulation was done with Comsol Multiphysics CFDcommercial code by a reduced 2D domain model via neglectingend effects from the side walls. The physics settings of COMSOLconsist of two elements i.e. sub-domain settings and boundaryconditions. In the sub-domain settings, types of materials, modesof heat transfer (conduction and/or convection) and initial condi-tions can be considered whereas in the boundary conditionssettings, geometry of boundaries can be considered. The simula-tion was executed at noon clear sky sunny day and solar irradia-tion was assumed 300 W/m2. Walls of the drying cabinet and backside of the collector were considered as adiabatic (thermalinsulated). After the analysis of the coupled thermal-fluid model,different results obtained, such as velocity field, temperature

Fig. 5. Mixed-mode solar dryer [8].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–13371330

distribution and pressure distribution in the solar collector as wellas in drying chamber with different damper opening conditionssuch as fully closed, half open and fully open and for differentoperational conditions. The predicted results were compared withmeasurements and found to be best agreement [27].

2.2.3. MATLAB simulationJain and Jain used MATLAB 5.3 to solve the transient analytical

model for inclined multi-pass solar air heater integrated withthermal storage and attached to the deep-bed dryer. The experi-ments were conducted in the month of October in Delhi (India)climatic condition. Various parameters such as length and breadthof collector, mass flow rate and change in the tilt angle had beenconsidered for analysis. The prepared computer program wasapplied to solve the energy balance equations of solar air heater.It was also applied to observe the temperatures of the absorberplate, storage material air (in different streams) and moisturecontent of the crop. The MATLAB 5.3 based proposed model wasvery useful for evaluation of thermal performance of flat platesolar air heater for the grain drying applications [28].

Dissa et al. used MATLAB version 7.0.1 software for simulationof thin layer inside indirect solar drying of mango slices. MATLABversion 7.0.1 has been used for estimation of drying rate at regularinterval of time by derivation of the moisture content dry basiswith respect to time using a derivation program. The experimentwas done for three consecutive days during harvest period ofmangoes. It was observed that in the first day there was a very lessconstant drying rate period and in the second day it becomenegligible. The drying rate reaches maximum of 0.18 g kg�1 s�1 onthe first day, 0.13 g kg�1 s�1 on the second day and0.04 g kg�1 s�1 on the third day. The fittings were carried outusing a nonlinear regression tool of software MATLAB (version7.0.1) based on the nonlinear optimization method of Levenberg–Marquardt with standard error 0.007249 and correlation coeffi-cient 0.9979 [29].

2.3. Mixed mode solar dryer

Mixed-mode solar dryer, is a combination of direct and indirectsolar dryer. This works under the combined action of the solarradiation incident on the material to be dried and the airpreheated in solar collector provides the heat needed for the

drying operation [8]. A diagram of a mixed-mode solar dryer isshown in Fig. 5.

2.3.1. FORTRANSimate used FORTRAN to develop a mathematical model for

mixed mode natural convection solar drying of maize grainconstructed at Newcastle University, U.K. The experiments werecarried out under a solar simulator. A good agreement betweentheoretical and experimental results was examined [30].

Bennamoun and Belhamri developed FORTRAN programs tosolve partial differential equations for energy balance in betweenair flat collector and the drying chamber for solar dryer. TheGauss–Seidel iteration method was applied to resolve these sets ofequations which were written in matrix form. With and withoutthe heater were considered for study. Results showed that solarbatch dryer with 3 m2 effective collector surface area can dry250 kg of onion flakes per day at 50 1C temperature [31].

Hawlader et al. developed a simulation program using theFORTRAN language to analyze the performance of the solar-assisted heat-pump dryer and water heating systems under themeteorological conditions of Singapore. The coefficient of perfor-mance (with and without water heater) and solar fraction wereconsidered for the performance evaluation of the system. By thesimulation and experimental results the values of coefficient ofperformance were obtained 7.0, and 5.0, respectively, whereas thesolar fraction values of 0.65 and 0.61 respectively [32].

2.3.2. MATLAB simulationJain used the MATLAB 6.1 software to solve the energy balance

equations for onion drying in the reversed absorber with packedbed thermal storage natural convection solar crop dryer. Computerprogram was prepared to solve the energy balance equations fordifferent components of solar drying systems and to predict thetemperatures of the air in different streams packed bed, dryingchamber, temperature of different absorber plates, storage mate-rial and crop in tray-I and tray-II. Author concluded that theproposed mathematical model is very useful for performanceevaluation of the reversed absorber type collector and thermalstorage with natural convection solar crop dryer [33].

Tiwari et al. designed and developed a PV/T mixed mode dryer.MATLAB software was used to develop an analytical expression ofcharacteristic equation for PV/T mixed mode dryer and the resultwas validated with experimental observations [34].

Tripathy and Kumar applied artificial neural network (ANN) forprediction of temperature of potato (slice and cylindrical shape)during solar drying in a natural convection mixed-mode solardryer. The input parameters such as solar radiation and ambienttemperature were considered for ANN modeling. Researchersconcluded that logsig transfer function with a trainrp backpropagation algorithm and the neural network with 4 neurons isthe most suitable ANN configuration for prediction capability oftransient food temperature and food geometries. The standarderror and correlation coefficient for ANN model were 0.208 and0.952 respectively where as for statistical model it was 0.381and0.846 respectively for cylinder. For the slice, standard error andcorrelation coefficient for ANN model were 0.130 and 0.980respectively and for statistical model were 0.210 and 0.949respectively. The developed tool is useful for food process engi-neers and designers of solar drying system for estimation andoptimal control of the solar drying process without doing com-prehensive experimentation [35].

Cakmak and Yildiz developed Feed Forward Neural Networks(FNNs) model for the nonlinear behavior of the drying of seedygrapes in a novel mixed mode dryer. The drying rate wasestimated with the help of an exponential equation by using

Fig. 6. Hybrid solar dryer [40].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–1337 1331

Fig. 7. Velocity distributions for natural convection [41].

Fig. 8. Temperature distributions for natural convection [41].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–13371332

nonlinear regression analysis and further the drying rate of seedygrapes was also estimated by using FNN. Nonlinear and linearregression models were compared with the FNN model and foundmore accurate in estimating the drying rate. For FNN, the rootmean square error and correlation coefficient were found 0.0019and 0.9991 [36].

2.3.3. Sigma plot VTripathy applied SIGMAPLOT (version 11.0.0.77) for the regression

analysis of potato drying in natural convection mixed-mode solardryer. In order to explain the drying performance of potato samples,eight thin layers dryingmodels were tested and compared statistically.The drying datawere converted into dimensionless moisture ratio andeight thin layer drying models were fitted to the experimental data.For all the tested models, values of coefficient of determination werelying in the range of 0.9327–0.9956. Though, among all the dryingtested models, the Modified Page model attained the highest values ofcoefficient of determination and lowest values of root mean squareerror and reduced chi square. The values of coefficient of determina-tion, root mean square error and reduced chi square obtained for theModified Page model were 0.9956, 0.01914 and 0.0003929 respec-tively for potato cylinders and 0.9907, 0.02705 and 0.000765 respec-tively for slices. It was concluded that Modified Page model is themost suited to explain the drying kinetics of potato slices andcylinders [6].

2.4. Hybrid solar dryer

In solar hybrid dryers, other sources of energy can be usedalong with solar energy to continue the favorable drying condi-tions (Fig. 6). Hence the drying process is not dependable on solarradiation. Blower is used for proper air movement in hybrid solar

dryer which can be powered by solar photovoltaic system Hybridsolar dyer can control drying of any agricultural produce byefficient way and also helps to maintain product qualities [8].

2.4.1. CFD simulationAndrew et al. used computational fluid dynamic (CFD) software

to understand the heat and mass transfer inside the dryer undernatural and forced convection mode. The commercially availablecomputational fluid dynamics (CFD) software, STAR-CD was exe-cuted for modeling of temperature and velocity distributionsinside the dryer. On the basis of simulated temperature profiles,it was concluded that heat and mass transfer by natural convec-tion is more suitable for drying pepper berries.

In the modeling of natural convection mode drying, distribu-tion of heat energy in the drying chamber was examine with thehelp of CFD, which was most influensive parameter in pepperberries drying. Fig. 7 illustrates the velocity distributions fornatural convection phenomenon. The maximum air velocity wasfound, 0.2338 m/s near the chimney outlet. It was due pressuredifferential and software gives the more accurate simulationresults. Fig. 8 demonstrated the distribution of temperature insidethe solar dryer by means of contour-plot at the middle section. Thehighest temperature region i.e. 366.6 K was found in the center ofthe chimney and lowest at the inlet of collector (291.2 K) where astemperature inside chamber is varies from 340.9 to 352 K. it givesthe complete picture of temperature profile which is highlydesirable in the selection of crop to be dried inside solar dryer.

Figs. 9 and 10 show the air movement and temperaturevariation from the inlet of the solar collector to the dryingchamber under forced convection mode. The highest and mini-mum temperature was found 311.5 K and 298.3 K respectively. Inthe forced convection modeling the heated air was forced at 0.1 m/

Fig. 9. Velocity distributions for forced convection [41].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–1337 1333

s from the solar collector to the drying chamber and the heataccumulated in the drying chamber ranges between 303 K and311.5 K. The heat distribution in the drying chamber was morehomogeneous in the natural convection model [38].

2.4.2. FORTRANJanjai et al. used Compaq Visual FORTRAN version 6.5 for

simulation and modeling to solve the partial differential equationswhich were describing the heat and moisture transfer of peeledlongan in PV-ventilated solar tunnel dryer. The simulated resultswere found good agreement with the experimental data. Thepredictions model for drying were evaluated on the basis of rootmean square difference, even though, there were found smalldiscrepancies between the predicted and measured moisturecontents. Root mean square difference of the moisture andtemperature prediction was 6% and 3% respectively [39].

2.4.3. MATLAB simulationBarnwal and Tiwari used MATLAB 7.0 software to calculate the

rate of heat utilized for moisture evaporation, hourly convectivemass transfer coefficient and moisture evaporated. A hybridphotovoltaic–thermal (PV/T) greenhouse dryer was designed andconstructed with maximum carrying capacity of 100 kg. Twodifferent types (Grade I and Grade II) of Thompson seedless grapeswere used for drying. A DC fan was also used to create forcedconvection mode. It was found that the coefficient of convectiveheat transfer for Grade I and Grade II grapes were lying in between0.26–0.31 W/m2 K and 0.45–1.21 W/m2 K, respectively [40].

2.4.4. SPSSChavan et al. used statistical software “SPSS” for statistical analysis

in Solar-Biomass Hybrid Cabinet Dryer. Coefficient of determination(R2), reduced chi-square (χ2), and the percentage of root mean squareerror (RMSE) were applied for selecting the best-fit equation todescribe the drying process. Statistical analysis was performed bynonlinear regression using statistical software in all the cases. In thisstudy, 11 different thin-layer drying models were compared accordingto their root mean square error (RMSE), chi-square (χ2) and coefficientof determination (R2). The R2 values of solar-biomass hybrid cabinetdryer were 0.96660, 0.9729, 0.9793, 0.953 and 0.9729 for texture,taste, appearance, odor and overall acceptability respectively. As perthe result of thin-layer drying of mackerel, all models were showinggood correlation, but, Midilli and two-term drying models were thebest fitted and might be used to accurately forecast the moisturecontent of dried mackerel. The lowest RMSE, χ2 and highest R2 valueswere found as a result of Midilli and two-term drying models.Statistical software “SPSS” is user friendly software for statisticalanalysis [41].

Hossain et al. used statistical software SPSS 9.0 for statisticalanalysis of a hybrid solar drying system. A statistical analysis wasdone for color values, lycopene, ascorbic acid and total flavonoidsof sun-dried, and different pretreated samples with the help of theanalysis of variance (ANOVA) using the software SPSS 9.0. Dun-can's multiple range tests was used to compare the mean obtainedfrom each set of variations. The average air temperature at theoutlet of the collector was found 30 oC higher than the averageambient temperature where as the collector efficiency wasincreased by 10% using the solar reflector. The efficiency of thesolar dryer was varied from 17 to 29% depending on differentoperating conditions. It was concluded that the color, ascorbicacid, lycopene, and total flavonoids of tomato was considerably

Fig. 10. Temperature distribution for forced convection [41].

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–13371334

reduced during the drying process, however the losses of color andnutritional components were higher than the commercially avail-able samples in the European market [42].

Functions, applications and limitations of different software insolar drying are summarized in Table 1.

3. Case-study of MATLAB software based ANN model used insolar drying

An ANN modeling is applied for prediction of hourly mass ofjaggery drying during the natural convection greenhouse condition.

3.1. ANN model

ANN modeling is adopted to approach a healthy and positivedrying. This modeling takes very less time. The first step is toprepare the ANN model through the investigation data. After thepreparation of the ANN model, the most favorable ANN config-uration was determined and at the last, the ANN model wasvalidated with experimental data, which was not used duringtraining of ANN model. The objective of this modeling is todevelop the best ANN model for the prediction of moisture

Table 1Functions, applications and limitations of different software in solar drying.

S.no.

Softwarename

Functions Applications Limitations

1 CFDFLUENT/ANSYS

CFD FLUENT is simulation software whichprovides the information about the heat transferand fluid flow behavior inside the solar dryer[9,11,12].

Exact location of ventilation hole can be easilypredicted. The prediction of exact shape & sizecan be done easily and can save money by timeby elimination of repeated manufacturing &lengthy exercise.

Learning of software is time consuming andgeometry meshing is also time takingprocedure.

2 ComsolMultiphysics

It is simulation software and it is providing theinformation of the heat transfer profile and fluidflow pattern inside the solar dryer [27].

This software can be use to predict the airmovement through inlet to outlet and also exactlocation of ventilation hole. It can also be used topredict the exact shape and dimension of thedryer.

As compared to CFD FLUENT Learning ofsoftware is easy.

3 FORTRAN This is used for simulation and modeling to solvethe partial differential equations [13,14,32,39].

It can be used for performance analysis of solardrying systems. It can also save cost byminimizing material usage. It can optimizestructural performance with thorough analysisand eliminates expensive & lengthy trial-and-error exercise.

The Fortran program firstly develops in aprototype software like visual languages suchas Matlab and IDL (Interactive Data Language)and then port this code to FORTRAN.

4 MATLAB MATLAB is mathematical modeling software andit is used for non linear regression analysisaccurately with taking very less time [16,23,28, 29].

This software is very useful to developmathematical models to prediction the croptemperature, air temperature, the moistureevaporated. It is also very useful for training andtesting of various models.

MATLAB mathematical modeling requiresexcellent programming skills.It takes long timeto develop and test the models.

5 SPSS SPSS is analytical software for non-linearregression analysis [20,21,41, 42].

SPSS is used for fast and accurate analyticalanalysis such as coefficient of determination, rootmean square error, analysis of variance, reducedchi-square etc.

It is expensive software.

6 Sigma Plot V Analytical software [6] Simulation validation in terms of variousstatistical parameters for greenhouse dryingperformance such as moisture evaporation rate,greenhouse room temperature etc.

Repetitive analysis is typical in this software.

7 Statistica Statistica is also a analytical Software for non-linear regression analysis of mathematicalmodels [20,21]

Statistica is used for analytical analysis ofcoefficient of determination, root mean squareerror, analysis of variance etc.

Skills are necessary for the use of appropriatesoftware applications in statistical dataprocessing.

8 TRNSYS TRNSYS is a universal scientific simulation tool insolar energy. TRNSYS Software is used to developand describe the drying behavior of crops indifferent kind of dryers [24,25].

Its great advantage is the replacement of difficultdifferential equations by easy numericalcalculations.

It gives more accurate results with the shortertime steps and the closer segments lie in everynumerical method.

The calculation of moisture and heat transfer atthe crop surface to the drying air and ofmoisture & heat transport in the inner crop canbe easily described.

Table 2List of transfer functions and back propagation training algorithms used in ANNtraining [19].

S. no. Transfer functions Training algorithm

1 Logsig Lm2 Tansig Cfg3 Poslin Bfg4 Satlin Br

Table 3Results of measures of error in food temperature prediction results of the ANNmodel considering 10 neurons and trainlm algorithm for different transfer func-tions [19].

S. no. Measures of error Transfer function

Logsig Tansig Poslin Satlin

1 Correlation coefficient 0.99 1.00 1.00 1.002 Root mean square error 1.44 0.59 0.76 1.44

Table 4Results of error analysis of the jaggery mass during the process of drying predictionresults of ANN model considering 10 neurons and logsig transfer function fordifferent training algorithms [19].

S. no. Measures of analysis Training algorithm

Trainlm Traincfg Trainbfg Trainbr

1 Correlation coefficient 0.999791 0.99934 0.999463 0.9998632 Root mean square error 1.446353 5.477226 2.239475 1.144925

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–1337 1335

content during the jaggery drying in the natural convectiongreenhouse condition. The feed-forward back-propagation net-work structure was found the best possible method under thenatural convection greenhouse condition. Equal numbers of neu-rons were used for input and output layers. MATLAB versionR2010a software was used for the training as well as testing ofthe neural model [19].

3.2. Development of ANN model

The development of the ANN model for jaggery drying wasderived from trial and error approach in an iterative method. Theperformance of the ANN construction had found minimum statis-tical error between the experimental data and predicted data ofhourly jaggery mass. To justify the performance of developed ANNmodel, the root mean square error and correlation coefficient wereconsidered. An efficient ANN model has to show highest correla-tion coefficient and small root mean square error values. Transferfunctions, training algorithms and the total numbers of neurons inthe hidden layers were considered in the network architecture,configuration.

3.3. Result

The input parameters were global solar radiation, diffuseinsolation, ambient relative humidity and ambient temperature.The experiments were performed from 10.00 am to 5.00 pm andall the parameters of the model were trained for the four daysdrying data. Every day, eight sets of experimental data werecollected. Two sets of data were used to authenticate the model.In the Table 2 the influence of varying neurons in the ANN modelarchitecture for the jaggery mass drying has shown.

They trained the ANN model from 1 to 40 neurons in thehidden layers and after that, increase the size of the networkgradually.

The ANN model was executed the best for 20 neurons becauseof RMSE value was 0.714812 and coefficient of determination (R2)was 0.999948. It was also observed that there is a drastic change inthe value of RMSE and R2 values between 10 and 20 neurons. Theintention was to found a most suitable transfer function in order tothat ANN model can predict the experimental data with maximumaccuracy. To improve the ANN architecture, the sensitivity studywith 10 neurons were also performed in association with trainlm(Levenberg–Marquardt back propagation) algorithm in subsequenttransfer functions: hyperbolic tangent sigmoid transfer function(tansig), logarithmic sigmoid transfer function (logsig), saturatinglinear transfer function (satlin) and positive linear transfer func-tion (poslin). The results have been presented in Table 3.

The investigation of the effect of various training algorithms wasrequired to obtain the best ANN model, such as Fletcher–Powellconjugate gradient back-propagation (traincfg), Levenberg–Mar-quardt back-propagation (trainlm), Bayesian regularization (trainbr)and Broyden, Fletcher, Goldfarb, and Shanno quasi-Newton backpropagation (trainbfg). The result has shown in Table 4.

4. Conclusions

It can be concluded that researchers and scientist have beenused various software in solar drying applications especially inoptimization of design parameters of solar dyers. Software are verysupportive in simulation of different types of solar drying systemsbefore fabrication. These software based analysis is not only savetime but also save the capital investment in solar drying systems.

CFD (ANSYS, FLUENT), Comsol Multiphysics, MATLAB andTRNSYS Simulation software are commonly applied for thermal

performance analysis. Partial differential equations based onenergy balance for different component solar drying system,regression analysis and fitting the drying kinetics data Sigma PlotV, SPSS and Statistica are widely used.

Software simulation gives clear cut picture in terms air flowand temperature inside the drying chamber as well as in wholedrying unit. The software application are helpful in deciding cropto be dried in existing solar drying system on the other hand it canhelp to design a new solar drying system for particular product.Simulation models are proposed by the academician/researcher toprovide the design data for the selection of the optimal dryerwould be helpful in further developments in this area.

References

[1] Lotfalian A, Ghazavi M, Hoseinzadeh B. Reviewing drying of dill and spearmintby a solar dryer and comparing with traditional dryers. World Appl Sci J2010;8:364–8.

[2] Prakash O, Kumar A. Solar greenhouse drying: a review. Renew Sustain EnergyRev 2014;29:905–10.

[3] Augustus Leon M, Kumar S, Bhattacharya SC. A comprehensive procedure forperformance evaluation of solar food dryers. Renew Sustain Energy Rev2002;6:367–93.

[4] Chen HH, Huang TC, Tsai CH, Mujumdar AS. Development and performanceanalysis of a new solar energy-assisted photocatalytic dryer. Drying Technol:Int J 2008;26:503–7.

[5] Kumar A, Tiwari GN. Thermal modeling of a natural convection greenhousedrying system for jiggery: an experimental validation. Solar Energy 2006;80:1135–44.

[6] Tripathy PP. Investigation into solar drying of potato: effect of samplegeometry on drying kinetics and CO2 emissions mitigation. J Food Sci Technol2015;52(3):1383–93.

[7] Reuss M, Benkert St M, Aeberhard A, Martina P, Raush G, Rentzell BV, Sogari N.Modelling and experimental investigation of a pilot plant for solar wooddrying. Solar Energy 1997;59:259–70.

[8] Pardhi C, Bhagoria JL. Development and performance evaluation of mixed-mode solar dryer with forced convection. Int J Energy Environ Eng 2013;4:1–8.

[9] Mathioulakis E, Karathanos VT, Belessiotis VG. Simulation of Air movement ina dryer by computational fluid dynamics: application for the drying of fruits.J Food Eng 1998;36:183–200.

[10] Bartzanas T, Boulard T, Kittas C. Effect of vent arrangement on windwardventilation of a tunnel greenhouse. Bio-Syst Eng 2004;88:479–90.

[11] Krawczyk P, Badyda K. Two-dimensional CFD modeling of the heat and masstransfer process during sewage sludge drying in a solar dryer. Arch Thermo-dyn 2011;32:3–16.

[12] Amjad W, Munir A, Esper A, Hensel O. Spatial homogeneity of drying in abatch type food dryer with diagonal air flow design. J Food Eng 2015;144:148–55.

[13] Mahapatra A K, Imre DSc L, Barcza J, Bitai A, Farkas I. Simulation of a directlyirradiated solar dryer with integrated collector. Int J Ambient Energy1994;15:195–204.

[14] Aghbashlo M, Muller J, Mobli H, Madadlou A, Rafie S. Modeling and simulationof deep-bed solar greenhouse drying of chamomile flowers. Drying Technol:Int J 2015;33(6):684–95. http://dx.doi.org/10.1080/07373937.2014.981278.

[15] Wang DC, Fon DS, Fang W, Sokhansanj S. Development of a visual method totest the range of applicability of thin layer drying equations using MATLABtools. Drying Technol: Int J 2004;22:1921–48.

[16] Seginera I, Bux M. Modeling solar drying rate of wastewater sludge. DryingTechnol: Int J 2006;24:1353–63.

[17] Tiwari GN, Das T, Chen CR, Barnwal P. Energy and exergy analyses ofgreenhouse fish drying. Int J Exergy 2009;6:620–36.

[18] Prakash O, Kumar A. ANFIS modeling of a natural convection greenhousedrying system for jaggery: an experimental validation. Taylor & Francis. Int JSustain Energy 2014;33:316–35.

[19] Prakash O, Kumar A. Application of ANN for prediction of jaggery mass duringdrying inside natural convection greenhouse dryer. Int J Ambient Energy2014;35:186–92.

[20] Hossain MA, Bala BK. Drying of hot chili using solar tunnel dryer. Solar Energy2007;81:85–92.

[21] Tunde-Akintunde TY. Mathematical modelling of thin layer sun and solardrying of cassava chips. Int J Food Eng 2010;6:1–22.

[22] Midilli A, Kucuk H. Mathematical modeling of thin layer drying of pistachio byusing solar energy. Energy Convers Manag 2003;44:1111–22.

[23] Reuss M, Benkert ST, Aeberhard A, Martina P, Raush G, Rentzell BV, Sogari N.Modelling and experimental investigation of a pilot plant for solar wooddrying. Solar Energy 1997;59:259–70.

[24] Awadalla HSF, El-Dib AF, Mohamad MA, Reuss M. Hussein HMS. Mathematicalmodelling and experimental verification of wood drying process. EnergyConvers Manag 2004;45:197–207.

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–13371336

[26] Romero VM, Cerezo E, Garcia MI, Sanchez MH. Simulation and validation ofvanilla drying process in an indirect solar dryer prototype using CFD Fluentprogram. Energy Proc. 2014;57:1651–8.

[27] Vintila M, Ghiaus AG, Fatu V. Prediction of air flow and temperature profilesinside convective solar dryer. Bull UASVM Food Sci Technol 2014;71:188–94.

[28] Jain D, Jain RK. Performance evaluation of an inclined multi-pass solar airheater with inbuilt thermal storage on deep-bed drying application. J FoodEng 2004;65:497–509.

[29] Dissa AO, Bathiebo J, Kam S, Savadogo PW, Desmorieux H, Koulidiati J.Modelling and experimental validation of thin layer indirect solar drying ofmango slices. Renew Energy 2009;34:1000–8.

[30] Simate IN. Simulation of the mixed-mode natural-convection solar drying ofmaize. Drying Technol: Int J 2001;19:1137–55.

[31] Bennamoun L, Belhamri A. Design and simulation of a solar dryer foragriculture products. J Food Eng 2003;59:259–66.

[32] Hawlader MNA, Chou SK, Jahangeer KA, Rahman SMA, Eugene Lau KW. Solar-assisted heat-pump dryer and water heater. Appl Energy 2003;74:185–93.

[33] Jain D. Modeling the performance of the reversed absorber with packed bedthermal storage natural convection solar crop dryer. J Food Eng 2007;78:637–47.

[34] Tiwari GN, Nayak S, Dubey S, Solanki SC, Singh RD. Performance analysis of aconventional PV/T mixed mode dryer under no load condition. Int J EnergyRes 2009;33:919–30.

[35] Tripathy PP, Kumar S. Neural network approach for food temperature predic-tion during solar drying. Int J Therm Sci 2009;48:1452–9.

[36] Cakmak G, Yıldız C. The prediction of seedy grape drying rate using a neuralnetwork method. Comput Electron Agric 2011;75:132–8.

[37] Prakash O, Kumar A. Historical review and recent trends in solar dryingsystems. Int J Green Energy 2013;10:690–738.

[38] Andrew RH, Rigit, Patrick TK. Low heat and mass transfer in a solar dryer withbiomass backup burner. World Acad Sci, Eng Technol 2010;62:115–8.

[39] Janjai S, Lamlert N, Intawee P, Mahayothee B, Bala BK, Nagle M, Muller J.Experimental and simulated performance of a PV-ventilated solar greenhousedryer for drying of the peeled longan and banana. Solar Energy 2009;83:1550–65.

[40] Barnwal P, Tiwari GN. Grape drying by using hybrid photovoltaic–thermal(PV/T) greenhouse dryer: an experimental study. Solar Energy 2008;82:1131–44.

[41] Chavan BR, Yakupitiyage A, Kumar S. Mathematical modeling of dryingcharacteristics of indian mackerel (rastrilliger kangurta) in solar-biomasshybrid cabinet dryer. Drying Technol: Int J 2008;26:1552–62.

[42] Hossain MA, Amer BMA, Gottschalk K. Hybrid solar dryer for quality driedtomato. Drying Technol: Int J 2008;26:1591–601.

P. Singh Chauhan et al. / Renewable and Sustainable Energy Reviews 51 (2015) 1326–1337 1337