Wind induced airflow through lightweight pitched roof ... - CiteSeerX

Wind induced airflow through lightweight pitched roof constructions: Test roof element – measurements and model validation Christoph Deseyve, Dipl.-Ing., Vienna University of Technology, Institute of Building Construction and Technology, Center of Building Physics and Building Acoustics; email: [email protected], http://www.bph.tuwien.ac.at Thomas Bednar, ao. Univ. Prof., Dr. techn., Dipl.-Ing., Vienna University of Technology, Institute of Building Construction and Technology, Center of Building Physics and Building Acoustics; email: [email protected], http://www.bph.tuwien.ac.at KEYWORDS: compact pitched roof, wind tightness, windproof, airflow, glass fibre insulation, air tight, wind washing. SUMMARY: In the case of non ventilated compact roofs the wind tightness of the construction is usually warranted by a windproof underlay membrane and the flow resistance of the thermal insulation, as well as sealed eave and ridge details. Because of the current construction practice of wind tight layers in Austria there are numerous small leakages in the eaves, the ridge and the underlay area. Because of this and the low density and length of the thermal insulation in common Austrian constructions the wind induced pressure differences between the eaves and loft area cause an air flow which cancels the thermal insulating effect of the mineral wool partially or completely. During periods of high wind speed this leads however to uncomfortable low operative temperatures caused by an increased heat loss and lower surface temperatures. Based on different in-situ measurements (thermal performance and air propagation) a test roof element was constructed and implemented in a common slope of 45° in a climate chamber (hot box – cold box) to investigate different tightening measures and to validate a simulation model. In the paper the results from air flow and heat flow measurements of the test roof element variants are presented. Heat fluxes rise up to 900% of the stationary calculated values at the eaves with a ventilation duct at the warm side and wind speed of about 5 m/s (ΔP=33Pa) and up to 300% in fully insulated elements. The results lead to the validation of the 3 dimensional simulation tool HAM3D-VIE. HAM3D-VIE allows variation studies of the hydrothermal performance and comfort regarding to durability and construction practice auf lightweight constructions.

1. Introduction As already shown in Deseyve & Bednar (2005 and 2006) pitched, light weight roofs with thermal insulation should be insulated air- and wind tight sandwich constructions. Otherwise the construction could cause inconvenient low surface and operative temperatures and uncommon high heat losses caused by wind washing. Wind washing means infiltration of the building envelope with cold outdoor air forced by temperature and pressure differences. Heat conduction seems to be just one of the factors causing heat transport, especially wind washing and wind and stack induced air flow through the building envelope insulated with mineral wool with low density are important factors. The most common building practice for light weight pitched roof constructions presently in Austria is the described compact system with no intended air flow in or above the thermal insulation. The air flow through and within the building envelope is a complex combination of different flows with some dominant factors. Riesner (2003) showed that there are some constrictive rotational flows in completely insulated roof constructions caused by the high air permeability of the thermal insulation. In Deseyve & Bednar (2006) is shown that standard Austrian light weight roof constructions have no significant effect of natural convection and that the dominant effect for the rising heat flux is the wind induced air flow. Following current construction practice different test roof elements were built to investigate the in-situ measured effects of the wind induced airflow and air patterns and to validate a simulation model to assess different building scenarios with different quality standards.

2. Test roof element 2.1 System description The climate chamber is constructed out of 10 cm extruded polystyrene (XPS) 1.15 m wide, 3.95 m long and 3.80 m high as shown in figure 1. The temperature difference of up to 25 K is warranted by a refrigerator at the cold side and a heater mat at the warm side. The roof element is fixed on a wooden construction on the warm side with a slope of 45°. On the warm side as well as on the cold side ventilators ensure that there is no thermal layering. At the eaves and the ridge the airflow through the roof element is controlled by fans assembled in pipes. The air flow is recorded by hot wire anemometers. Pressure differences were measured between the airspace at the “eaves” and the airspace at the “ridge”. The roof element was constructed following Austrian building practice for light weight roof constructions defined in table 1. The hot and the cold side are separated by the roof element and a vertical and a horizontal XPS element.

FIG.1: Cross section of the test roof element box & view of the test box - enclosure xps 10cm.

3 gaps 2cm high 24 cm wide (3x2/24)

ventilation duct 5 x 2/17cm – cold side

FIG.2: Cold side of the test roof element – eave variation and air flow channel above the thermal insulation.

TABLE. 1: Construction definition.* layer

d [m]

λ [W/mK]

ρ [kg/m³]

c [J/kgK]

kL [m²]

Acryl glass

0.010

0.190

1180

1440

0

Glass wool

0.200

0.038

14.5

1030

9x10-9

0.015

0.130

600

1700

0

OSB W

* U = 0,193 /m²K, kL = air permeability in m² The properties of air are assumed to η = dynamic viscosity in Pas = 17,1 x 10-6; ρL = air density in kg/m³ according to the Boussinesq-Approximation, β = isobar volume expansion coefficient in 1/K = 3,7 x 10-3.

2.2 Measurement equipment During the reference period air temperature (PT 1000, M-FK 222, Heraeus) and relative humidity (humichip, Vaisala) were measured at a few points in the air volume on both sides of the roof element. Surface temperature (PT1000) at different points, heat flows (heat flux measuring film, RdF, typ 20457) at the interior surfaces and pressure differences (halstrup & walcher, Delta_p) between the eaves and the ridge air space were monitored. The positions of the sensors are shown in Figures 1, 2, 3.

FIG.3: Hot side of the test roof element – heat flux measuring films and PT 1000.

2.3 Case specification To be able to validate the simulation model according to different air flow patterns 7 different variations of a roof element were built. Thereby means “ventilation duct” a 20 mm wide air flow channel above or beneath the thermal insulation layer. The tightness variation of the eaves is secured by an underlay frame with 3 gaps 20 x 240 mm at the warm or the cold side as shown in figure 2. TABLE. 2: Case specification. Roof element

Thermal insulation material

1

Ventilation duct

Eaves

Gap

size [cm]

situation

size [cm]

variation

size [cm]

glasswool 14.5kg/m³

20

no duct

-

open

-

open

2


18

warm side

2

open

-

open

3


18

warm side

2

cold side

3 x 2/24

open

4


18

warm side

2

warm side

3 x 2/24

open

5


18

cold side

2

open

-

open

6


18

cold side

2

cold side

3 x 2/24

open

7


18

cold side

2

warm side

3 x 2/24

open

Ridge/Attic

2.4 Test Results Based on the insitu measured data (concerning the correlation between pressure distribution and wind speed) different pressure differences were applied to the roof element. At the stage of an almost steady state condition the heat fluxes were recalculated to U-values and compared with the insitu measured with different roof element variations. The dynamic heat loss coefficient U* in W/m²K was calculated according to Deseyve & Bednar (2005) to an hourly mean value [U*(t)=q(t)/(Ti(t)-Te(t))] (measurement error 5.4 %). 2.4.1 Basics

cp = pressure coefficient (-) FIG.4: Correlation between wind speed and pressure difference – Test roof and insitu – measurements Deseyve & Bednar (2006) Figure 4 shows the correlation between the wind speed and the insitu measured pressure differences between the eaves and the attic (ridge) air space. For the further testing series regarding the pressure differences across the test roof element the values of the highest curve (c=2.5) where used to calculate the worst case. 2.4.2 First Results

FIG.5: Test roof element 1- correlation between wind speed and current heat loss coefficient


Test roof element 1 (full filled thermal insulation) has an increase of the heat flux according to the wind speed as expected and shown in figure 5. The heat flux rose mostly at the eaves and the fewer the nearer to the ridge. The maximum U-value at a wind speed of 4.5 m/s (equals about 30 Pa pressure difference) was measured at the eaves up to 0.7 W/m²K.





FIG.11: Test roof element 7- correlation between wind speed and current heat loss coefficient As shown in figure 6 the duct at the warm side raised the heat flux and the U-value in correlation to the wind speed of 4.5 m/s up to 1.80 W/m²K. If we compare U1 of figure 6 to the run of U1 in figure 7 and 8 there is clearly shown that the routing of the air flow has an important impact too. Especially if the whole thermal insulation section is protected with the underlay and the air flow is directed in the duct as shown in figure 8. As we can see in figure 9 there is hardly any difference of the heat flux run whether there is no duct or the duct at the cold side. Figure 10, where the air is directed into the duct at the cold side, shows a better behaviour than the fully filled variant. Figure 11 shows the impact of the air flow routing in the same way as shown for elements 3 and 4 in figure 7 and 8. Most of the air is directly routed into the thermal insulation layer. This causes higher heat fluxes especially in the eaves area at higher wind speeds. The impact of the ventilation duct on the cold side is very small at the eaves and higher the longer the air is transported through the mineral wool (realation of resistance).

2.4.3 Relation to former insitu - measurements

Correlation between wind speed and current heat loss coefficient – laboratory and insitu –measurements FIG.12 according to Deseyve & Bednar (2005)

FIG.13: according to Deseyve & Bednar (2006)

Figure 12 shows the correlation between the rising U-value and the wind speed depending on different variations of duct and thightness level. The instu measured data of a light weight roof construction in different detached houses in Austria fit well to the laboratory measured U-values of the worst case test roof element. As we expected the values for element 4 fit very well to the insitu measured system There was a wind induced airflow under the thermal insulation form the eaves to the attic caused by the duct effect at higher pressure differences between the thermal insulation and the polyethylene foil (AFVR). The values are based on the situation of U2 as shown in figure 1. Based on the reference data of the test roof figure 13 shows the wind dependent U-values of element 5, 6, 7 and 1 well fitting to the measured data according to Deseyve & Bendnar (2006). All in all we can assume that the data obtaind out of the test roof element are comparable to the insitu measured date and consequential we can use the measured data of the test roof element to validate HAM3d-VIE to calculate real buildings under real climate conditions.

3. Simulation Model 3.1 Basics The mathematical modelling for the three dimensional heat and moisture transport is based on the prEN 15026 (2004) and Bednar (2005) and considers moisture dependent heat conductivities, vapour diffusion and capillary forces but no air flow. The already existing program HMS (Bednar 2005) was adapted as described in 3.2 according to Deseyve & Bednar (2006) to simulate the combined heat, moisture and air transport (intrusion, infiltration or convection). For a first step only the pitched roof area was simulated with HAM3D-VIE to get data for the validation. During the simulation the measured values where used for the boundary conditions of the interior (hot box) and the ridge as well as the outdoor climate (cold box).

3.2 Air flow For the calculation of air flow effects equation (1) according to Wang (2003) has been used. The air volume is simulated with an effective permeability of 3.5e-7 m² and a thermal conductivity of 0.025 W/mK.

qa = ρa

k ⎡∇Pa + ρagβ ( T − Tref ) ⎤⎦ η⎣

(1)

where q a = mass flow rate in kg/(m²s), ρa = air density in kg/m³, k = air permeability in m², η = dynamic viscosity in Pas, ∇Pa = pressure gradient in Pa/m, g = gravity constant in m/s², ß = volume expansion coefficient in 1/K, T = temperature in K and Tref = reference temperature in K. At the inlet and outlet the measured mass flow rate is used as a boundary condition.

3.3 Results

FIG.14: Test roof element 1- correlation between measured and simulated U-values – 0 Pa


As shown in figures 14 to 17 the simulated heat fluxes fit well to the measured values of the test roof element.



FIG.18: Test roof element 5,6 - correlation between measured and simulated U-values – 33 Pa

The simulation for the duct at the cold side (figure 18) correlate very well with the measured values for the eave variation of element 5 (open eave) near to the eave and ongoing to the ridge the simulated values fit to the measured data of element 6 (only a gap on the cold side – no airflow through the thermal insulation), so that we can assume that the flow resistance of the air duct is lower as we expected in relation to the flow resistance of the thermal insulation.

4. Conclusions It is important to ensure that the thermal performance and the durability of light weight building envelopes. The heat fluxes of the fan forced air flow variants through the test roof element correlate with the insitu measurements of light weight roof constructions of detached houses. Heat fluxes rise up to 900% of the stationary calculated values at the eaves with a ventilation duct at the warm side and wind speed of about 5 m/s (ΔP=33Pa) and up to 300% in fully insulated elements. The measurements could be reproduced by the simulation tool HAM3D-VIE. HAM3D-VIE allows variation studies of the hydrothermal performance and comfort regarding to durability and construction practice auf lightweight constructions.In the next step the results of the case study with the validated simulation model should lead to construction recommendations to achieve a moisture tolerant roof construction without relevant wind induced heat losses as well as to coefficients for building codes concerning the additionally heat losses and the decrease of surface temperatures in relation to wind incidence.

5. Acknowledgements This paper has been written within the ongoing Translational Research Programm L 233 “Hygrothermal Performance of Windthight Roof Constructions” funded by the Austrian Science Fund (FWF).

6. References Bednar, T. (2005). Hygrothermische Gebäudesimulation. Modellentwicklung und Anwendung in der Neubauund Sanierungsplanung. Professorial dissertation. Vienna: Vienna University of Technology. Deseyve, C. & Bednar, T. (2005). Increased Thermal Losses caused by Ventilation through compact Pitched Roof Constructions – In Situ Measurements. Proceedings of the 7th Symposium on Building Physics in the Nordic Countries. Vol. 2: 881-887. Reykjavik: The Icelandic Building Research Institute. Deseyve, C. & Bednar, T. (2006). Wind induced thermal losses through compact pitched roof constructions – Test roof – measurements, simulation model und validation. 3rd International Building Physics Conference. in Research in Building Physics and Building Engineering. 459 - 464. Taylor & Francis Group. Montreal: Concordia University. prEN 15026 Draft. (2004). Hygrothermal performance of building components and building elements – Assessment of moisture transfer by numerical simulation. Brussels: European Committee for Standardization Riesner, K. (2003). Natürliche Konvektion in losen Außendämmungen – Untersuchungen zum gekoppelten Wärme-, Luft- und Feuchtetransport. Dissertation, Rostock: Ingenieurwissenschaftliche Fakultät, Fachbereich Baukonstruktionen und Bauphysik, Universität Rostock. Wang, J. (2003). Heat and Mass Transfer in Built Structures – Numerical Analyses. Dissertation, Göteburg: Department of Building Physics, Chalmers University of Technology.