Vous êtes sur la page 1sur 8

Numerical Studies of Heat and Air flow in Ventilated Insulated Slanting Roofs

Arild Gustavsen, Professor, Department of Architectural Design, History and Technology, Norwegian University of Science and Technology; Arild.Gustavsen@ntnu.no and www.ntnu.no Sivert Uvslkk, Senior Scientist, SINTEF Building and Infrastructure; Sivert.Uvslokk@sintef.no and www.sintef.no KEYWORDS: Ventilated roof, Numerical simulation, CFD, Natural convection, Air flow. SUMMARY: A common roof design in the Scandinavian countries is the ventilated slanting roof. Ventilation of the roof has two purposes; to transport any moisture that has penetrated the roof construction out of the roof and to keep the outer roof surface cold enough to avoid snow melting. Snow melting on the roof outer surface can cause pools of water freezing near the eave construction, which again can cause mechanical degradation and leakage. There have been some discussions regarding how large the air gap in ventilated roofs should be in order to provide sufficient ventilation for ensuring low temperatures and for allowing transport of moisture out of the roof. Previous studies have mostly been based on simplified analyses where the driving pressure and pressure loss coefficients have been used to calculate the air flow through the roof. These pressure loss coefficients are however usually for simpler geometries than the air gap geometries found in roofs, and some questions can be raised regarding their usability for such analyses. In this paper we use a CFD tool to calculate the mass flow rates through some ventilated slanting roofs with various designs. Stream contours and temperature distribution in some roofs are also presented.

1. Introduction
Two typical roof designs are common in Norway; the compact flat unventilated roof, and the slanted ventilated roof. The compact roof is common for larger commercial buildings, and the ventilated roof is common for smaller buildings, and in particular for residential buildings. Some typical ventilated roof constructions are shown in Figure 1, with the cold attic and the roof with insulation in the roof deck being the most common designs. In this paper the focus is on the roof with insulation in the roof deck (the hygrothermal performance of the ventilated attic is studied in detail by Uvslkk, 2005). For the roof with insulation in the roof deck, a ventilation channel is located above the insulation layer and wind barrier (roofing underlay) but below the roofing layer, see Figure 2. Ventilation of the roof has two purposes; to transport any moisture that has penetrated the roof construction out of the roof and to keep the outer roof surface cold enough to avoid snow melting. Snow melting on the roof can cause pools of water freezing near the eave construction, which again can cause mechanical degradation and leakage. Formation of icicles may also occur which again may be of danger to passing pedestrians. In Norway there have been some discussions regarding how large the air gap in ventilated roofs should be in order to provide sufficient ventilation for ensuring low temperatures and for allowing transport of moisture out of the roof. Previous studies have mostly been based on simplified analyses where the driving pressure and pressure loss coefficients have been used to calculate the air flow through the roof (Blom, 1990). In this paper a Computational Fluid Dynamic (CFD) tool is used to study the air flow rates, temperature distribution and flow pattern in the air cavity of some ventilated roofs.

2. Geometric Model
The different roofs simulated are named H1 to H8. Two-dimensional cases were simulated, see below. Details of the ridge and the eave constructions are shown in Figure 3, with dimensions shown in Table 1 and Table 2. Table

FIG. 1: Examples of typical ventilated roof designs. (Figure from SINTEF Building Research Design Sheets.)

FIG. 2: Figure of a typical ventilated roof with ventilation directly above the roofing underlay (wind barrier). The numbers refer to: 1. roofing tiles or sheets, 2. roofing battens, 3. counterbattens, 4. roofing underlay, 5. example of a specific underlay system installation, 6. thermal insulation, 7. vapour barrier, 8. ceiling, 9. noggings, 10. cautiously clamped overlap joints, and 11. cross ventilation between underlay and roofing. (Figure from http://tjenester.byggforsk.no/prodok/ntg/dok/2375/2375ge.pdf.) 1 presents common dimensions for all roofs, while Table 2 presents dimensions that are different for the various configurations. All roofs were of the type where the wind barrier and roof underlay is combined into one single layer. This is different to the other common practice where the wind barrier is below a separate underlay layer. Because the roof was simulated in two dimensions a two-dimensional representation of the rather complex threedimensional ventilation channel had to be made. That is, highly three-dimensional objects like profiled roofing tiles were converted to a flat plane sheet, while the air gap openings between the battens and roofing tiles were retained by using an equivalent distance between roofing tiles and roofing battens. In real Norwegian roofs the rafters (and counterbattens mounted on the rafters) are spaced 600 mm apart. The height of the channel, which equals to the sum of the height of the battens and counterbattens, is usually smaller than 100 mm. The flow in the middle of each section may therefore be close to two dimensional. The ridge and the eave of the roofs are also simplified compared to a real three-dimensional roof configuration. The eave is simulated without an eave box, which is used quite a lot in Norway. The eave box limits the amount of snow that may penetrate into the ventilation channel (Thiis et al. 2007). The ridge is simulated without a ridge band that usually is installed to avoid rain penetration into the roof. Such a thin band would be difficult to model. The main difference between the eight houses can be seen from Table 2. H1 to H5 may be thought of as real roof constructions, while the others are included to see the effect of the ridge and eve details by themselves. House H1 is the reference: H1 is about 10 m wide, and has 18 battens in each side of the roof. The height of the counter-battens and battens are 36 mm and 30 mm, respectively. The angle of the roof is 30. The difference between roof H1 and the roofs H2 to H5 are: H2 has a house width that is double the width of H1, H3 has a roof angle that is 20, compared to 30 for H1 (this results in a slightly shorter roof ventilation channel and a reduction in number of roofing battens from 18 to 17),

H4 has counterbattens that are 48 mm tall, compared to 36 mm for the other roofs. H5 has battens that are 36 mm tall instead of the 30 mm that is common for the other roofs. Roofs H6-H8 are variants of roof H1, where H6 has a roofing channel without battens, but where the ridge and eave details are included H7 has roofing battens but is without eave and ridge constructions, H8 has no battens or eave and ridge constructions.

3. Numerical Simulations
To examine the air flow in the ventilation channel of the roof, numerical simulations with a CFD program (Fluent, 1998) was carried out. Only air flow by natural convection was considered, that is there was no external wind acting on the building. This is the case with the least air flowing through the roof.

3.1 Natural convection simulations


Only the ventilation channel of the roof was considered in these simulations. A pressure inlet boundary condition was used on the inlet (the eave) and a pressure outlet was used at the opening in the ridge. Because of symmetry, only half the roof was simulated. That is, a symmetry boundary condition was used at the center of the ridge.

FIG. 3: Ridge for house H1 to the left and figure of the entire house to the right. Refer to Table 1 and Table 2 for dimensions (Hofseth 2003). TABLE 1: Common dimensions for the various roofs. Description Roofing thickness Equivalent distance between roofing and roofing battens Width of roofing battens Size (horizontal) of eave Height of eave board Size of opening between ridge tile and ridge board Height of ridge board Height of opening between ridge board and lath Size of opening between ridge tile and roofing tile Size of opening between ridge board and roofing tile Size of opening between roofing battens and ridge board Distance from underlay to eave board Distance between roofing battens Symbol tt dtl bl br tfb dm1 hmb dm2 dm3 dm4 dm5 drl Value 20 10 48 400 148 10 61 30 10 30 30 18 350 Dimension mm mm mm mm mm mm mm mm mm mm mm mm

TABLE 2: Dimensions separating the various roofs. Case Height of counter-battens, ds [mm] 36 36 36 48 36 36 Height of roofing battens, dl [mm] 30 30 30 30 36 30 House width [m] 10 20 10 10 10 10 10 10 Channel length on each side of ridge[m] 5.998 11.598 5.321 5.998 5.998 5.998 5.998 5.998 Angle of roof [] 30 30 20 30 30 30 30 30 Number of roofing battens 18 42 17 18 18 0 18 0

H1 H2 H3 H4 H5 H6 H7 H8

The exterior air temperature was set equal to 0 C. Thus, air entering the ventilation channel had this temperature. A heat resistance of 0.04 m2K/W was assumed on the outside surface of the roof (fluid flow was not simulated there). Inside the building a temperature of 20 C was assumed. Heat resistances of 4.5 and 6.7 m2K/W were assumed between inside of the building and the exterior wall surface and the roof underlay surface, respectively. These resistances are based the U-value requirements in the Norwegian Building Code from 1997 (NBC 1997). Incompressible and steady laminar flow was assumed. Further, viscous dissipation was not addressed, and all thermophysical properties were assumed to be constant except for the buoyancy term of the y-momentum equation where the Boussinesq approximation was used. The Semi-Implicit Method for Pressure-linked Equations Consistent (SIMPLEC) was used to model the interaction between pressure and velocity. The energy and momentum variables at cell faces were found by using the First order upwind scheme. In addition, the CFD code used central differences to approximate diffusion terms and relied on the PREssure Staggering Option scheme (PRESTO) to find the pressure values at the cell faces. PRESTO is similar to the staggered grid approach described by Patankar (1980). Radiant heat transfer was included in the simulations through use of the discrete transfer radiation model (DTRM), which relies on a ray-tracing technique to calculate surface-to-surface radiation. The internal cavity walls were assumed to be diffuse gray and to have an emissivity of 0.9, and air did not interact with the radiative process. The double precision solver of the CFD code was used. In the Boussinesq model, which is used for finding the buoyancy effects in y-momentum equation, the operating temperature is among the parameters. Correct determination of this temperature is important in order to find the right flow through the channel. Because the operating temperature is equal to the mean air temperature of the ventilation channel, an iterative process was used. Fist, an operating temperature was guessed, then after some iterations the operating temperature was adjusted to reflect the new temperature in the channel. This procedure was repeated until the temperature hardly changed any more. The air properties used in the simulations were calculated at the mean temperature of air in the ventilation channel and at atmospheric pressure, P = 101325 Pa. A standard acceleration of gravity of -9.81 m/s2 was used in all calculations. The conductivity of the wooden battens and the roofing material was set equal to 0.12 and 1.7 W/mK, respectively. As the results section will show, only transient solutions of the specified problems were found. That is, first a stationary solution procedure was tried (with under-relaxation to avoid divergence), but the residuals did not decrease. Therefore a transient solution procedure was selected. A time step of 0.1 s was chosen (Fluent, 1998). A small part of the wall was included in the numerical simulations. Thus, there were two inlet boundaries at the eave; the opening between the lower batten and the roofing layer and the horizontal opening between the bottom part of the eave board and the wall of the building. This was done to allow some air flow along the wall against the eave openings.

4. Results and Discussion


Transient solutions were found for all the various cases in Table 1. Below some snapshots of the velocity steam function and temperature plots are shown. Air flow rates are also reported and discussed.

4.1 Velocity results


Figures 4 and 5 show the stream contours for the ridge and the eave for roof configuration H1, respectively. Both figures show that air circulation occurs between the battens. Although there are some exceptions, the air flows upward on the right side (upper side) and downward on the left side (lower side), between each pair of battens. Figures 4 and 5 also show that more irregular patterns may develop between the battens. The reason for this air circulation is that the lower surface is warmer than the upper surface. As we will see below, similar flow also occurs in roof H6, where there are no battens. Another feature that can be seen from the figures is that most of the air flows along the lower surface of the roof, between the battens and the roof underlay. Here the flow direction is upward, from the eave toward the ridge. Above the battens and beneath the roofing layer (the equivalent spacing that that takes into account the flow between profiled tiles and battens) the main air flow direction is downward. At the eave, where the air enters the roof, Figure 5 shows that most of the air enters through the opening between the batten and the roofing layer. Less air enters the ventilation channel between the batten and the roof underlay. The reason for the small amount of air entering the roof from the opening below the batten, may be that the air flow direction outside the opening is away from the opening. This movement is caused by the rising heated air along the wall. Figures 6 and 7 display stream contours for roof configuration H6. This roof does not have any battens between the eave and the ridge, and was included in order to see the effect of the battens versus the eave and ridge details on the air flow through the roof. Both figures show that a complex flow pattern also develops inside this roof; even if there are no roof battens. Similar flow patterns have been reported by e.g. Corcione (2003), but then for rectangular horizontal closed cavities. Two typical cell sizes are found inside the ventilation channel, large ones and small ones. The size of the different cells seems to be quite constant except for the larger cell close to the eave. The larger cells circulate clockwise and the small cells circulates counter clockwise. Studies of whether the cells move upward was not performed, but this is expected because there is some air flow through the roof, from the eave to the ridge.

FIG. 4: Stream contours for the ridge of roof H1.

FIG. 5: Stream contours for the eave of roof H1.

FIG. 6: Stream contours for the ridge of roof H6.

FIG. 7: Stream contours for the eave of roof H6.

4.2 Roof temperature


The temperature distribution in roof H1 (close to the ridge) is shown in Figure 8. As seen from the colour bar to the left, the temperature difference between the warmest and the coldest part of the roof is not large. The coldest temperature (about 273.15 K) can be found close to the eave while the warmest temperature (about 274.3 K) can be found in the ridge. The effect of the air circulation on the temperature distribution of the air in the ventilation channel can easily be seen. Where the air rises the temperature is high, and vice versa. A similar temperature pattern was observed for roof H6, even if there were no battens inside the roof. Figure 9 displays the outside surface temperature of roof H1. The horizontal axis shows the distance from the eave in millimetres (mm), while the vertical axis shows the temperature in K. The figure shows that the temperature of the roofing material is close to the outside air temperature of 273.15 K close to the eave. Further, the figure shows that the temperature increases quite rapidly, from the eave, to about 273.25 K, where it oscillates for the remaining part of the roofing layer, except for a rather small area close to the ridge. In the middle part of the roof, where the oscillations take place, the lowest temperature is found close to the battens. This is because the battens act as thermal radiation shields. The average surface temperatures for all the various roofs were practically equal (to 273.25 K).

274.32 274.25 274.17 274.09 274.01 273.93 273.85 273.78 273.70 273.62 273.54 273.46 273.38 273.31 273.23 273.15

FIG. 8: Temperature contours for roof configuration H1. Temperatures are in Kelvin.
273.4 273.35 273.3 273.25 T [K] 273.2 273.15 273.1 273.05 0 567 1134 1701 2268 2836 3403 3970 4537 5104 5671 Distance from eave [mm]

FIG. 9: Figure showing outside surface temperature for roof H1.

4.3 Mass flow rates


Table 3 shows the mass flow rate and the mean velocity in the ventilation channel. The results reported are time averaged data for the outlet of the ventilation channel. That is, the simulation ran for several cycles, and the average during these cycles was reported. If a repeating pattern was found in the mass flow rate through the exit opening, the average is reported for a complete cycle. By looking at the table the following results can be observed: 1) Increasing the height of the counter-battens from 36 mm (H1) to 48 mm (H4) increases the flow rate from 2.6110-4 kg/s to 3.7110-4 kg/s. 2) Increasing the height of the battens from 30 mm (H1) to 36 mm (H5) also increases the flow rate (from 2.6110-4 kg/s to 3.510-4 kg/s). The latter is because the height of the ventilation channel (height of battens and counter-battens) as a whole increases. Increasing the width of the building (going from configuration H1 to H2) also increases the flow rate (keeping the same roof angle of 30). What is more surprising is that reducing the roof angle from 30 (H1) to 20 (H3) also increases the mass flow rate. This may partly be explained by the shortening of the ventilation channel (the channel has less battens), but also by a changing flow pattern that changes with roof angle and roof configuration. That the latter is the case, may be shown by looking at the results for cases H6 (no battens but with contractions in the eave and ridge), H7 (battens, but no contractions at the eave and ridge) and H8 (open channel no battens and no contractions at the eave and ridge). From looking at the stream contours of roof H6 (see Figure 6), it can be seen that there hardly is any air movement through the roof, but that air circulates within the ventilation channel. This is the reason for the low mass flow rate through this roof.

TABLE 3: Mass flow rates for the various roofs. Case H1 H2 H3 H4 H5 H6 H7 H8 Mass flow rate [kg/s] 2.6110-4 3.9110-4 3.0010-4 3.7110-4 3.5010-4 4.6910-5 4.7510-4 2.9710-3 Average velocity at outlet [m/s] 0.020 0.030 0.023 0.029 0.027 0.0036 (two exit openings between battens) 0.0349

These results should be validated by experiments, which to some extent were tried, but controlling and measuring the flow rate through the roof at these small temperature differences was difficult. The CFD code has however been used to predict similar flow patterns in windows and frames which have been verified by experiments (Gustavsen et al. 2001).

5. Conclusions
The numerical simulations show that a complex flow pattern exists within the roof ventilation channel. Changing one parameter, i.e. the roof angle or removing the battens, may not always give the results that seems obvious in the fist place. The reason for this is that a circular flow pattern occurs, and that this pattern changes with changing geometry and roof angle. Still, there seem to be a relationship between the height of the ventilation channel (height of battens and counter-battens) and the mass flow rate through the channel. The mass flow rates may for instance be used in moisture simulation software to analyze whether the various designs produce large enough air flow rates in the ventilated cavity of the roof for transporting moisture, which has penetrated the roof construction from the interior space of the building, out of the roof.

6. Acknowledgement
This paper has been written within the SINTEF strategic institute projects Impact of Climate Change on the Built Environment and Climate 2000 - Weather Protection in the Construction Process. The authors gratefully acknowledge the Research Council of Norway. Further, this work would not have been possible without the contributions by Vidar Hofseth (Hofseth 2003).

7. References
Blom, P. 1990. Ventilation of insulated slanting roofs. Ph.D. dissertation, Department of Building and Construction Engineering, Norwegian University of Science and Technology. Corcione, M. 2003. Effects of the thermal boundary conditions at the sidewalls upon natural convection in rectangular enclosures heated from below and cooled from above, International Journal of Thermal Sciences, Vol. 42 (2), pp. 199-208. Fluent (1998). FLUENT 5 Users Guide, Fluent Inc., UK. Gustavsen A., Griffith B.T., and Arasteh, D. 2001. Three-Dimensional Conjugate Computational Fluid Dynamics Simulations of Internal Window Frame Cavities Validated Using Infrared Thermography, ASHRAE Transactions, Vol. 107(2), pp. 538-549 Hofseth, V. 2003. Air flow in ventilated insulated roofs (in Norwegian). Project report, Department of Civil and Transport Engineering, Norwegian University of Science and Technology. NBC 1997. Norwegian building code 1997 (in Norwegian). Patankar, S.V. 1980. Numerical Heat Transfer and Fluid Flow. Washington: Hemisphere. Thiis, T.K., Barfoed, P., Delpech, P., Gustavsen, A., Hofseth, V., Uvslkk, S., and de Virel, M.D. 2007. Penetration of snow into roof constructionsWind tunnel testing of different eave cover designs, Journal of Wind Engineering and Industrial Aerodynamics, Vol. 95(9-11), pp. 1476-1485. Uvslkk, S. Roofs with a cold attic (in Norwegian), Project report 396, SINTEF Building and Infrastructure.

Vous aimerez peut-être aussi