The partial differential equations are converted to algebraic equa­tions by means of finite-volume techniques using rectangular meshes on a staggered arrangement. Diffusive terms at the boundaries of the control volumes are evaluated by means of second-order central dif-

formulation EPF [5]. The resulting algebraic equation system was solved using a pressure — based SIMPLE-like algorithm [8], and the iterative convergence procedure was truncated when relative increments of the computed Nusselt number (Eq. 1) were below the con­vergence prescribed value (10~6). The meshes are intensified near the solid parts using a

tanh-like function with a concentration factor of 1 [4], in order to solve the boundary layer correctly. This aspect is indicated in Fig. 2 with solid triangles.

1.5 Verification.- The numerical solutions presented here have been calculated adopting a global h-refinement criterion. That is, all the numerical parameters (numerical scheme, numerical boundary conditions, etc.) are fixed, and the mesh is refined to yield a set of numerical solutions.

This set of numerical solutions has been post- processed by means of a tool based on the Richard­son extrapolation theory and on the concept of the Grid Convergence Index (GCI) [1][6]. The problems are solved on different meshes related by a mesh ratio r = 2. The tool processes a set of three con­secutive solutions in the h-refinement. The most relevant parameters arisen from the verification pro­cess are the GCI, the observed order of accuracy of the numerical solution (p), and the percentage of nodes of the post-processing grid where has been applied the post-processing procedure, which are called Richardson nodes Rn [1].

For all the post-processing results presented in this work (see for example Table 4), the observed order of accuracy (p), approaches the theorical values of the differential scheme used (between 1 and 3); the GCI values decrease as n increases; and the percentage of Richardson nodes obtained was high.

These results indicate that the estimator GCI is reliable, and that the solution is free of

SHAPE * MERGEFORMAT

According to results in Table 7 the percent­age differences dif% decrease with the de­crease in A’, and the maximum percentage difference was 8.3% at. A good

behaviour of Eq. 8 to calculate Kp parame­ter was found in A’ < 1.0 range. For A’ > 1.0, the percentage differences |<й/|% were higher due to the influence of the overall as­pect ratio A = 20. For the purpose of the design in transparent insulation technology, is an appropriate range.

Comparison with the Scozia results [7]

In Table 8 our numerical results of the Nus — selt number are compared with the Scozia nu­merical results. It was observed that the per­centage differences dif% increases with the decrease in A’, while the maximum percent­age difference was 04% at.