Abstract
This study deals with the convective drying characteristics of porous moist objects in different shaped enclosures which are important in the design of drying chambers and the development of energy efficient systems. In numerical analysis of these systems, the coupled heat and mass transport equations in the moist object domain and enclosure domain should be solved simultaneously to obtain accurate results while different soft computing techniques can be used to assist high-fidelity parametric computational analysis of convective drying. In this work, two-dimensional heat and mass transfer of different shaped porous moist objects inside square and L-shaped cavities are investigated by using Galerkin weighted residual finite element method. The porous objects have three different geometries: square, triangle, and circle with equal edge and diameter sizes. Effects of cavity geometries for varying Reynolds numbers (Re), inlet/exit port sizes, and distances between objects are investigated on a coupled transport mechanism. The results show that the increasing value of Re has a positive effect on moisture transfer, although the effects are different for objects with different geometries. The maximum difference in moisture content values is seen between triangle and circle objects with a value of 36.6%, although highest heat transfer values are achieved in the square geometry. The shape of cavity and distance between the objects are also effective. The L-shaped cavity leads to higher impact on evaporation, and heat and moisture transfer is higher when the spacing between the objects is narrower. The maximum difference in moisture content values is observed between triangle and circle objects, with the triangle object having 36.6% lower moisture content. There is 61.22% variation in the moisture content between the highest and lowest values of the distance between the objects. An artificial neural network is used in convective drying performance estimation for L-shaped cavity using a four input, two output system. Dynamic features and moisture reduction amount are efficiently obtained by using 25 neurons in the hidden layer.