ÇATI GEOMETRİSİN BİNA ISIL DAVRANIŞI ÜZERİNDEKİ ETKİSİNİN İNCELENMESİ

Bu calismanin amaci yaz aylari icin enerji verimliligi acisindan, bina cati geometrisinin ic ortam havasina etkisini arastirmaktir. Bu amacla Harran konik catili yapilari ayni taban alani, hacim ve termo-fiziksel ozelliklere sahip duz catili bina ile karsilastirilmistir. k-e turbulans modeli kullanilarak uc boyutlu CFD simulasyonu gerceklestirilmistir. Cati geometrisinin dogal havalandirmaya etkisi arastirilmistir. Bu kapsamda capraz havalandirma debisi ve tasinim isi transfer katsayilari degerlendirilmistir. Harran evi capraz havalandirma acisindan duz catili yapiya gore % 8 daha iyi performans gostermektedir.  Sayisal analizler sonucundan tipik bir yaz gunu icin Harran evinin kubbesinin duz catidan % 30 daha az isi atagina maruz kaldigi ve bunun daha dusuk ic ortam havasi sicakligina neden oldugu ortaya cikmistir .


INTRODUCTION
Harran houses are located in Harran district of Şanlıurfa Province in Turkiye, which especially differ in the shape of roof from the modern buildings. Interesting roof forms of Harran House shown in Fig. 1 have construction dates back to 7. BC (Özdeniz, et al., 1998). These vernacular buildings are important structures to understand the history and architectural features of the region as well as the change in the adaptation to climatic conditions. For the no air conditioning situation, Harran conical domed houses retain the internal air cooler than the flat roofed buildings in the summer days. The effect of the conical roof shape on building thermal behavior needs to be examined for its adaptation to modern low energy buildings. Başaran (Basaran, 2011)investigated the thermal performance of Harran Houses and emphasized both of the high thermal capacity of the square base walls and the opening at the top of the dome that facilitates the natural ventilation have contributed providing relatively good indoor thermal conditions. Cardinale, et al., 2013, have recently focused on the analysis of Mediterranean vernacular buildings in the view of energy and indoor comfort without use of air conditioning systems. Results show that the thermal mass of the external walls and roof dampers the large external temperature fluctuations. For the summer season the indoor temperature ranges between 25-28 ºC, even with the high outdoor air temperature above 35 ºC. Geva, et al., 2014, investigated the thermal comfort based on morphological analyses including cross ventilation in a Synagogue. The results show that the synagogue exhibits high discomfort thermal levels especially in summer, because the design of this synagogue did not consider climate conditions. Faghih and Bahadori 2011, investigated the thermal performance of the domed roofs considering the parameters such like air flow around them, solar radiation, radiation heat transfer with sky and the ground, and openings. Their numerical simulation shows that the thermal performance of the domed roof buildings is better compared to the flat roof one under the same conditions. Laborda, et al., 2015, also investigated the effect of a roof monitor skylight to maximize thermal conditions in a house. Al-Jawadi ve Al-Sudany, 2010, compared the thermal behavior of single domed building and identical flat-roof building by a mathematical technique supported by experimental measurements. The final results arrived at dome-system works to lower temperature by about (2-6) ºC on average and aids in reaching thermal balance summertime. Pearlmutter, 1993, also compared the indoor temperatures of the flat roof and vaulted roof experimentally and observed that vaulted roof geometry has greater thermal stability. Nguyen and Reiter, 2014, numerically optimized the design of a low-cost housing to examine the role of thermal comfort criteria. The results show that the optimal design of a natural ventilated house and air-conditioned one has differences, so it is needed to propose an adequate design in the early stage of a project.
The objective of present study is to investigate the effect of the conical roof geometry on indoor air temperature by comparing with flat roofs to ascertain which of these two roof geometries is more energy efficient during summer months

Heat Transfer Through the Roofs and the Walls
In this study, for the convenience of the calculation of the radiation on the conical geometry, the conical roof was modeled as an octagonal pyramid. Each surface of this octagonal pyramid is assumed to be oriented to north, south, west, east, northeast, northwest, southeast, southwest. For the comparison, it is also assumed that the two building model has same material properties, same thickness, same volume and covering the same base area. And the other assumptions as below, a) Heat transfer through the roof and the walls is assumed one-dimensional; in the absence of any heat generation in the wall, the corner effects are negligible and the wall base is insulated. b) Ambient temperature and the solar radiation are assumed constant for an hourly period. c) Thermal conductivity of the wall material is constant. d) The convection heat transfer coefficient between the outer surface of the building and the ambient air is constant (Section 2.2) e) The convection heat transfer between the indoor air and the inner surface of the building has a value of (Wallentén, 2001). The general form of the heat transfer equation under given conditions is described below, by taking x-direction normal to the wall surface as shown in Fig. 2. (1) Heat transfer from the outer surfaces of the wall and the roof involves with solar radiation, convection, heat transfer with sky and conduction inside the wall and the roof The boundary conditions at the left and right boundaries can be expressed as, The initial condition is, Initial temperatures of the walls, the roof and the indoor air temperature are assumed to be equal to the outside air temperature at 6.00 am. ( Fig. 3). . Thus, there are five unknown nodes of that 1 inner, 1 outer and 3 internal nodes shown with 1, 2, and 3. The equations given below are respectively obtained by applying the energy balance at the left (node 0) and right (node 4) boundaries The valid equations for internal planes (nodes of 1, 2, and 3) where is the absorptivity, is the surface area, is the solar heat flux incident on the surface of the wall and the roof is the emissivity, is the Stefan-Boltzmann constant. We denote , and respectively as distance between the nodes, temperature of the node m at time t, thermal conductivities of the wall, specific heat, and, densities of the wall.
Hourly solar radiation on the horizontal surface with its direct, diffuse and the ground reflected components are calculated by using the methods given in Duffie and Beckman, 1991, based on hourly measured global radiation for horizontal surface in Şanlıurfa. Results are presented in Fig. 3. Sky temperature shown in the figure is on the other hand calculated according to the equation of ( ) , where describes measured hourly ambient temperature. Incoming solar radiation values per unit area of roof are then evaluated for both conical and flat roofs. Each unit area of the flat roof surface receives much higher solar radiation as demonstrated in Fig. 4.  The energy balance for the room air, assuming no opening on the building and negligible the radiation exchange between interior surfaces, can be written as, and are respectively temperatures of the jth inner surface and interior air at time t, On the orher hand, is air density, is specific heat of air, and, is volume of the room.

Calculation of Convection Heat Transfer Coefficient on the Roof and the Walls of the House
Three dimensional CFD simulation are performed to determine the convective heat transfer coefficient on the roof and the walls of the conical domed and the flat roof houses. Autodesk CFD Simulation 2013 is used for the analysis. The models are employed for an actual sized room of conical domed Harran house, provided that the flat roof building has same volume and same base with its domed counterpart.
The convective heat transfer coefficient is calculated with a fundamental equation given below, where ( ) is a constant value specified at that ambient conditions in software and the building surface temperature is accepted as constant,  Tetrahedral mesh type with approximately 1,5 million cells is used in this analysis, under the best allowable memory limit and speed of the computer available in our high performance computer lab. In addition, simulations were made in two different mesh structures and the deviations were found to be below 5%. Hence, the result is that the simulation is independent of mesh structure. The inlet plane wind speed is chosen 2 m/s to simulate average summer wind speed for Şanlıurfa (Şanlıurfa İl Çevre Durum Raporu) and the outlet plane is accepted 0 Pa

Investigation of Effect of Roof Geometry on Ventilation
This section will compare the conical and flat roofed buildings in natural ventilation flow rates. For this purpose, CFD simulation performed with using turbulence model which is effective and widely used in various engineering applications (Nguyen and Reiter, 2014). The domain decomposition technique (Meroney, 2009) is used to predict opening flow rates. Same opening locations added to both building models to find pressure coefficients which are shown in Fig. 6. The boundary conditions are set to the same as CFD analysis in Sec 2.2 except the inlet velocity which is accepted as 5 m/s according to wind profile in this case.
meteorological wind speed was assumed as 10 m/s and h is the height (Allocca, 2003).

Figure 6: Computational domain and wind flow pattern inside Harran house
The opening flow rates for the cross ventilation is calculated from the relations below, where is the discharge coefficient, is opening area, is approaching wind speed at building height, and are pressure coefficients.

Velocity inlet
pressure outlet symmetry symmetry The air flow velocities through the openings on the roof of the Harran house are measured to calculate the energy transferred by the air (Fig.7). The cross section areas of the openings are 15cm by 15 cm. The measurements were performed by Testo 435 anemometer during an hour with 10 minutes periods. The energy transported with air movement from the roof openings can be calculated by adding mass flow rate of air, ̇ in Equation (8). This kind of wind velocity roughly causes more than energy exchange between inside and outside, and this result is nearly same with Ref (Başaran T., 2011)

RESULTS AND DISCUSSION
As a result of the CFD simulation for the scenario which wind blows with 2 m/s speed from west, the distribution of the convection heat transfer coefficients for the building surfaces are determined and shown in Fig. 8. These point-wise values throughout surfaces are averaged and results are summarized in Table 1 for both houses with conical and flat roofs. It is clear from the table that the conical roof house has higher convective heat transfer coefficient on windward side but a lower value on the roof.  Table 1. The convection heat transfer coefficients for both the flat and the conical roof building surfaces, obtained by CFD simulation.
The mean values of volumetric flow rates for two roof models are computed based on CFD analysis outputs. Obtained results are presented in Table 2. The conical roofed house has relatively higher flow rate than the flat one. Percentagewise difference is nearly 8% in favor of the conical roof.
As a result of the analysis, y+ <5 result plotted for the y + value k-ε turbulence model was obtained ( Figure 9). As expected, in the standard k-ε turbulence model, the result is 50% higher than the LRNM k-ε turbulence model on the windward side for a flat roofed building. This result is similar to that of Blocken et al., 2009. The results of LRNM k-ε turbulence model were given in this study   Heat transfer analysis of the outer and the inner surface of the walls and the roofs allow calculation of surface temperatures. The average values of hourly outer surface temperatures for both roofs are presented in Fig. 10 for different surface azimuth angles facing in eight directions during the summer day. The outer surface temperatures for the flat roof are always greater than those of the conical roof. The flat roof outer surface temperature is the highest between 11.00 and 16.00. Variation of inner surface temperatures for both roofs are also calculated and shown in Fig. 11. The inner surface temperatures for the flat roof are again higher comparing with the conical roof one. Temperature measurements by a high resolution thermal camera were performed on the inner surface of the conical roof at surface azimuth angle γ=180º. The image is shown in Fig. 12. The calculated value for corresponding time and surface of the conical roofed house is nearly and there is well-match with the experimental value obtained from the thermal camera measurements.  The heat flux through indoor air from roofs and the walls are calculated for comparison and the results are shown in Fig. 13. The heat flux from the flat roof to indoor air is higher and it reaches the highest value between 17.00-18.00, corresponding to nearly 50% more heat attack to the interior air. On the other hand, Fig. 12 indicates that there is no significant difference between the heat fluxes of both houses through indoor air from the walls.
By using inner surface temperatures of the two building models, the energy balance is applied to the room air, with the assumption of no openings on the buildings. This analysis gives the indoor air temperature fluctuations for a typical summer day (Fig 13). This graph shows that heat flow to conical roof is lower than the flat roof during the day and this is the main cause for the conical roofed house to have better indoor air temperature in summer.

CONCLUSION
In this study, the conical and flat roof geometries were investigated in terms of energy efficient roof geometry for summer season. Both of the building models have equivalent thermo-physical properties, same base area and same indoor air volume. The following conclusions summarize the analysis in this study; (1) The conical roofed Harran house has lower indoor air temperature than the flat roofed house during a summer day. (2) The conical roofed house has an advantage of the less daily total heat transfer to indoor air from walls because of the less surface area. At the end of the typical summer day, the roof of the Harran house is also exposed to 30% less heat attack (370,7 ⁄ ) than roof of the flat building (553,8 ⁄ ).
(3) At any time of the day, the outer surface temperature of Harran house dome is different with changing surface azimuth angles but the outer surface temperature of flat roof is uniform. This difference for conical roof is 10º C respects to surface azimuth angle, mainly depends on solar radiation incidence In fact this temperature difference helps heat remove from the roof. (4) In the case of wind incidence angle of 90 and cross ventilation, the Harran house has better ( ) performance than flat roofed building. (5) In real conditions, Harran houses have one opening on the top and a few on the base of the roof, so these openings causes the vertical ventilation and thus help the heat rejection by the heated air accumulated under the roof, and its value is found 1000 W for Harran house dome. (6) The inner surface of the Harran dome has finned form and this situation increases the rate of heat transfer from the inner surface to medium, thus it causes the roof to cool and circulate of air under roof so further research is needed to evaluate this fact.