Caspian Journal of Applied Sciences Research 4(8), pp. 19-26, 2015 Journal Homepage: www.cjasr.com ISSN: 2251-9114
Three Dimensional Numerical Energy Simulation of Double-Skin Façade Buildings Amir Mollaei Kouhi1,*, Mohammad Ghadimi2 1
Dept. of Energy Engineering, Islamic Azad University, Southern Tehran Branch, Iran.
2
Dept. of Energy Engineering, Islamic Azad University, Roudehen Branch, Iran.
Using double-skin layers is one way of preventing energy loss in buildings. The skins with mechanical ventilation between them (airflow window or AFW skins) have recently found widespread applications throughout the world. In the present article, the finite volume method was used to numerically analyze the efficiency of these skin facades based on an existing empirical model. The following flow conditions were considered: two and three dimensional, steady flow, turbulent flow, laminar flow (in some cases), incompressible, viscous, and single-phase flow. In the present analysis, the air flow through the window (AFW) case was solved. Geed agreement was observed between the obtained results and the existing empirical results. The K-ε/RNG turbulence model was used in the present analysis. This same model was then numerically solved for different temperature and temporal data in the shaded and unshaded cases. As a result, heat transfer rates were successfully optimized. © 2015 Caspian Journal of Applied Sciences Research. All rights reserved.
Keywords: Double-skin AFW facades; turbulence model; temperature profile.
1. Introduction Energy consumption optimization is today a very important concern. For this reason, double or multiple layers (facades) are used for obtaining high energy consumption efficiencies. These layers prevent heat transfer from the outside to the inside of the building and vice versa. In double skin facades with internal air flow, the optimum air flow velocity, temperature, as well as width and length thereof must be studied. First, a review of the existing literature is introduced. The first studies on double-skin facades were conducted by Brandle and Boehm, Riparti, and Park et al. All these studies were aimed at improving energy efficiency. The first innovation with regard to double-skin facades occurred in 1957 in Sweden. In 1967, ECONO Company built the first double-skin façade building
in Helsinki, Finland. Baker et al. conducted empirical tests on double-skin buildings in Japan. They studied the effect of the inlet fluid temperature, inlet airflow velocity, wind, etc. on the heat transfer behavior of these buildings. Saelens studied heat transfer optimization of double-skin buildings in terms of energy efficiency. Ghadamian and Ghadimi conducted an energy analysis on double-skin façade building models. In the present article, a numerical analysis is conducted similar to the numerical solution presented by Busselen based on the empirical Matelaer Vliet empirical tests. The K-ε/RNG geometry was selected for shearing analysis so that the results can be readily compared with those obtained from the existing numerical and empirical analyses to establish the reliability of the implemented numerical method. In some previous studies, this was presented as 2D
*
Corresponding address: Dept. of Energy Engineering, Islamic Azad University, Southern Tehran Branch, Iran. E-mail address:
[email protected] (Amir Mollaei Kouhi)
© 2015 Caspian Journal of Applied Sciences Research; www.cjasr.com. All rights reserved.
Amir Mollaei Kouhi; Mohammad Ghadimi / Three Dimensional Numerical Energy Simulation of Double-Skin Façade Buildings 4(8), pp. 19-26, 2015
∂(ρuj T) ∂ μ μt ∂T =− [( + ) ] ∂xj ∂xj Pr Prt ∂xj
analyses and the obtained results were in good agreement with both the empirical and the numerical results. Here, 2D and 3D analyses were performed and the results compared with those of the previous studies. It was revealed that the numerical results were reliable upon slight simplification of the equations. In the remaining parts of this article, the specific temperature distribution during a night and day (Feb. 16, 2000) is analyzed and the effect of mechanical ventilation (shading) as well as skin height and width on the skin behavior are studied.
∂(ρuj k) ∂ ∂k =− [Γk ] + Gk − Yk ∂xj ∂xj ∂xj ∂(ρuj ε) ∂ ∂ε =− [Γε ] + Gε + D − Yε ∂xj ∂xj ∂xj Where ui is the velocity component, T is temperature, P is pressure, µ is fluid viscosity, Pr is the turbulent Prandtl number, µt is the turbulence viscosity, k is the turbulence kinetic energy, ε is the specific turbulence damping, G is the production term, Y is the damping term, and D is the diffusion term, modeled by the abovementioned turbulence model. The following fluid properties were considered:
1.1. Governing flow equations The following flow conditions were considered in the present article: 3D, steady, incompressible, viscous, turbulent, laminar, and single-phase. To simulate the turbulence terms, the K-ε/RNG model was implemented. This model is the result of a powerful statistical technique (termed “rematching group theory”) and is similar to the standard model, but comprises a series of modifications including the following: - This model contains an additional term in the ε equation which considerably improves solution accuracy for fast compressed flows. The rotation on turbulence effect used in this model leads to increased solution accuracy for rotational flows. - The theory provides an analytic formula for the turbulent Prandtl numbers, whereas the standard model uses constant values determined by the user. - The theory provides an analytical differential formula for viscosity for low Reynolds numbers, whereas the standard model is used for high Reynolds numbers only. Nevertheless, effective use of this model depends on the flow behavior in the vicinity of the wall.
-
Air molecular mass Mw=28.966kg/kg.mol
=
constant
-
Universal gas constant = constant= R= 8.3145 kJ/kg.mol.K
-
Specific heat capacity Cp=1006.4 J/kg.K
-
Heat conduction coefficient= constant= K=0.0242 W/m.K
=
=
constant=
Also, µ is obtained from the Sutherland’s viscosity law and the working pressure is equal to the atmospheric pressure.
2. Materials and Methods 2.1. Computational methods In the present article, first the 2D and 3D models for the skin and the block are built in the Gambit environment, and the initial meshing of the boundary layer (initial thickness=0.1, growth ratio=1.2) formed. Then, the inner plate is divided into 0.3 intervals. The model thus built is subsequently solved in the Fluent environment upon entering the initial conditions and determining the solution model. Fluent applies the finite volume method at the cell center. This software was selected for solving the governing equations. The second approximation method was used for discretization of the equations and the Simple algorithm was implemented for pressure and velocity coupling.
The above show that the reliability and validity of the proposed model is greater than those of the standard model for a broad range of flows. Therefore, to obtain the flow field, we must simultaneously solve the continuity, momentum, and energy equations as well as the k and ε heat transfer equations: ∂(ρUj ) =0 ∂xj ∂(ρuj ui ) ∂P ∂uj ∂ = + [(μ + μt ) ] ∂xj ∂xj ∂xj ∂xj
The first order upwind difference scheme was used for the time separation term and the power 20
Amir Mollaei Kouhi; Mohammad Ghadimi / Three Dimensional Numerical Energy Simulation of Double-Skin Façade Buildings 4(8), pp. 19-26, 2015
law scheme for the displacement terms. The momentum and energy equations were coupled with the continuity equations and written as appp=∑anbpnb+sp . The details of this method are given in. The discount coefficient was obtained based on the empirical results for producing the maximum convergence velocity. The convergence limit for the dependent variables θ, p, v, and u varies relatively between consecutive iterations. Here, this limit was assumed to be |(∅^(k+1)-∅^k)/∅^k |≤ε where ε=10-4 for φ=u.v, and ε=10-5 for φ=θ.
and lower parts are insulated and no heat transfer occurs through them (Figs. 3 and 4). Gambit was used for meshing, creating control volumes, setting the boundary conditions on various surfaces as well as at the inlet and outlet of the main flow, and selecting the fluid for the present problem. The structured meshing was implemented for both the models. For increased numerical analysis accuracy in the vicinity of walls and at the flow inlet, a finer mesh was considered for these regions.
2.2. Selected geometry and network The selected geometry for the heat transfer flow comprised two air inlet channels and one air outlet channel placed at the top of the skin through which hot air can enter during winter and cold air during summer. These currents subsequently exit from the top of the building upon passing through the skin (Figs. 1 and 2).
Fig. 3: The 2D meshing of the numerical model
Fig. 1: The 2D view of the analyzed geometry for the case with no shading
Fig. 4: The 3D meshing of the numerical model Fig. 2: The 2D view of the analyzed geometry for the case with shading
The skin width, height, and length were assumed to be 300 mm, 2700 mm, and 1000 mm respectively. As shown in Fig. 4, there are two inlets for air (diameter=7mm) and one outlet (diameter=10mm). The origin of the coordinate system was assumed to be at the corner of the
The present geometry consists of two hot and cold walls at constant temperature, placed on the left and right of the model respectively. The upper 21
Amir Mollaei Kouhi; Mohammad Ghadimi / Three Dimensional Numerical Energy Simulation of Double-Skin Façade Buildings 4(8), pp. 19-26, 2015
selected geometry. This meshing contains several separate planes for numerical analysis: one plate corresponds to the outlet compartment and comprises 110, 20 and 30 mesh elements along the X,Y, and Z axes respectively. The second plane contains the skin which passes the main flow and the greatest heat transfer between the fluid and the walls occurs at this section. There are 110 * 30 * 222 mesh elements along the X, y, and Z axes respectively. Overall, there are 798,600 nodes in the completed meshing. The mesh elements are sp arranged that at all sections, the y+
