comparison of the numerical and ex­perimental fluid flow data is shown in Fig. 8. They correspond to case Th = 30 0C, considering studies on cells c13 and c26. The upper — index * in the velocities indicates that they have been normalized using the reference value of, while

the length was normalized with L. Fig.

8 contains two sets of data. Velocity profiles at the central vertical section fo each observation window are shown at the bottom. At the top there are maps of local differences and between the numerical and experimental data in

the two observation windows. These maps of differences were obtained by interpolating the experimental and numerical data at the nodes of a regular mesh of 16 60 nodes, and by calculating the differences at each of the nodes. Third-order accurate interpolations were used to minimize additional errors in the data processing.

For cell c13, it was observed that Дм*[%] values were higher than Дм*[%] values. The maximum values of local differences Дм* were localized in small regions close to isother­mal walls, while the Дм* distribution was uniform. For cell c26 results of Дм*[%] were lower than Дм*[%], which corresponds to the local distribution of Дм* (Fig. 8b), where their max­imum values were localized in small regions close to isothermal walls. As for for Дм*, their maximum values were localized in extensive regions located in the central zones of each cell.

For Tft = 30 0C case, the maximum value of local differences was 10.1 %, for Tft = 70 0C case was 10.2 %, and for Tft = 120 0C case was 11.4%. These maximum values were localized in cell 26 and for the Дм* map. For two cells, and for all cases, the experimental u* — profiles are perfectly reproduced by the numerical profiles.

5. — Conclusions

Except in some special cases, it is demonstrated that the numerical model proposed to solve reduced computational domains with periodic boundary condi­tions, perfectly reproduces cases over whole computa­tional domains. A final validation test was carried out by comparing numerical data with experimental heat transfer and fluid flow measured in an ad-hoc experi­mental setup. The fluid flow was measured by a Digital Particle Image Velocimetry device.

References

[1] J. Cadafalch, C. D. Perez-Segarra, R. Consul and A. Oliva. Verification of finite volume computations on steady-state fluid flow and heat transfer. Journal of Flu­ids Engineering. 124(1):11-21, March 2002.

[2] K. M. Kelkar and D. Choudhury. Numerical Prediction of Periodically Fully Developed Natural Convection in a Vertical Channel with Surface Mounted Heat Gener­ating Blocks. International Journal of Heat and Mass Transfer. 36(5):1133—1145, 1993.

[3] B. A. Meyer, M. M. El-Wakil and J. W. Mitchell. Natural convection heat transfer in small and moderate aspect ratio enclosures; An application to flate — plate collec­tors. American Society of Mechanical Engineers, New York. 1978.

[4] C. D. Perez-Segarra, A. Oliva, M. Costa and F. Es — canes. Numerical experiments in turbulent natural and mixed convection in internal flows. International Jour­nal of Numerical Methods for Heat and Fluid Flow. 5:13-33, 1995.

[5] M. Quispe, J. Cadafalch, M. Costa and M. Soria. Com­parative Study of Flow and Heat Transfer Periodic Boundary Conditions. Proceedings of the ECCOMAS 2000. Barcelona, September, 2000.

[6] PJ. Roache. Perspective: A method for uniform report­ing of grid refinement studies. Journal of Fluids Engi­neering. 116:405-413, 1994.

[7] R. Scozia and R. L. Frederick. Natural Convection in Slender Cavities with Multiple Fins Attached to an Ac­tive Wall. Numerical Heat Transfer, Part. A. 20:127­158, 1991.

[8] J. P Van Doormal and G. D. Raithby. Enhancement of the SIMPLE method for predicting incompressible fluid flows. Numerical Heat Transfer. 7(6):147-163, 1984.