ABSTRACT
An intelligent integrated approach of Jaya optimization algorithm and neuro-fuzzy network is introduced to model the natural convection in an open round cavity against the decision parameters. The decision parameters are the Rayleigh number (Ra) in range of 105 to 2 × 105 and ratio of the nonconductor barrier distance from the bottom of the cavity to the cavity diameter (H/D) in range of 0 to ∞. After that, the obtained hybrid model is applied to forecast the average Nusselt number in the cavity. In the next step, the experimentally obtained data by using a Mach–Zehnder interferometer is utilized to train the hybrid model. The accuracy of the hybrid model is evaluated through the calculation of average testing and checking errors. According to the obtained results, there is an optimum ratio (H/D = 0.5), in which the heat transfer is maximum. Moreover, it was found that the best hybrid model forecasts the natural convection in the cavity with the mean absolute percentage error of less than 0.75% and 1.40% for the testing and checking data, respectively.
Nomenclature
ANFIS | = | Adaptive neuro-fuzzy inference system |
B | = | Body force (N/m3) |
D | = | Diameter of the cylinders (m) |
FS | = | Fuzzy System |
g | = | Gravitational acceleration (m/s2) |
h | = | Heat transfer coefficient (W/m2.°K) |
H | = | Distance of adiabatic plate from the bottom of cylinder (m) |
k | = | Thermal conductivity of air (W/m.°K) |
l | = | Cylinder length (m) |
MF | = | Membership function |
MAPE | = | Mean absolute percentage error, % |
Nr | = | Specific refractivity of air (m3/kg) |
Nu | = | Nusselt number |
P | = | Pressure (Pa) |
r | = | Radius of the cylinders (m) |
R2 | = | Correlation coefficients |
R0 | = | Gas constant (J/Kg.K) |
Ra | = | Rayleigh number based on the cylinder diameter (gβ(Tw-T∞)D3/να) |
RMSE | = | Root-Mean Square Error |
STD | = | Standard Deviation |
SSE | = | Sum-Squared Errors |
t | = | Slot width (m) |
T | = | Temperature (°K) |
Greek symbols
α | = | Thermal diffusivity of air (m2/s) |
β | = | Coefficient of volumetric thermal expansion of air (1/°K) |
μ | = | Dynamic viscosity (Pa.s) |
ρ | = | Density (kg/m3) |
ε | = | Fringe shift |
θ | = | Angle from stagnation point (degree) |
λ | = | Laser wave length (m) |
ν | = | Kinematic viscosity (m2/s) |
Subscripts
∞ | = | Refers to ambient condition |
f | = | Refers to film condition |
L | = | Refers to local value of parameter |
ref | = | Refers to reference condition |
W | = | Refers to the cylinder surface |
Disclosure statement
No potential conflict of interest was reported by the authors.
Additional information
Notes on contributors
E. Akbari
E. Akbari now is working at the University of Applied Science and Technology of Kermanshah Branch. Main research interests of Ehsan are mechanical Engineering, manufacturing Engineering and new aspects of artificial intelligence. His current project is ‘Hybrid modelling of heat transfer & numerical investigation of heat transfer due to non-Newtonian nanofluid’
A. Karami
A. Karami obtained all his major university titles (M. Sc, B. Sc and Ph. D) at the Razi University, Iran. He is interested in the hybrid modelling of heat transfer processes using artificial intelligence approach. He has written more than 40 journal papers and 10 conference papers.
S. Nazari
S. Nazari is a Ph.D student at the Razi University of Kermanshah. He is working on solar energy as his Ph.D. thesis. His research interests are energy and exergy analysis of solar energy systems, fluid flow and heat transfer, enhanced heat transfer by nanofluid and non-newtonian nanofluid. He has published several conference papers.
M. Ashjaee
M. Ashjaee currently works at the School of Mechanical Engineering, University of Tehran. Mehdi does research in Mechanical Engineering and Optical Engineering. His current project is ‘combustion, battery pack cooling with PCMs and metal foam, miniature heat sink with ferrofluids under alternate magnetic field’.