Printer Friendly

Investigation of thermal and airflow conditions near glazed facades using particle image velocimetry and cfd simulation--eliminating the need for secondary perimeter heating systems.


Occupants near windows often experience thermal discomfort. During the winter, window interior temperatures usually fall below indoor air temperature, causing thermal discomfort due to radiant temperature asymmetry, low operative temperature, and downdraughts. High-performance windows provide solutions to these problems and improve glazing interior surface temperatures in winter. The main HVAC system, which is usually variable air volume, provides ventilation, cooling, and often some heating as well. In cold climates, perimeter heating is traditionally used in office buildings during cold periods as a secondary heating system near fenestration in order to improve thermal comfort near facades and to avoid oversized HVAC systems. However, as this paper will show, by utilizing a high-performance envelope, the main HVAC system can supply the necessary heating without a secondary heating system while ensuring good comfort conditions.

Perimeter heating systems can be convective or radiative. Electric baseboard heaters are most common; they are usually placed under the windows and exchange heat with the air by convection and with the person, glass, and room surfaces by radiation. The warm air rises and increases window interior surface temperatures, thus reducing thermal discomfort. Radiant ceiling/floor panels are sometimes used instead. These electrically-heated panels exchange heat by radiation with the person and the surfaces of the windows, walls, floor, and furniture. Another variation is electrically-heated windows. In this case, the window surface itself is heated with a current provided on a film deposited on the interior glass between the glass panes. Therefore, the glass surface is directly heated, and thermal discomfort due to radiation exchange can be eliminated.

The first part of this paper investigates and compares the performance of three different perimeter heating systems and presents air velocity and temperature gradients near a glazed facade with different perimeter heating variations (baseboard heaters, radiant ceiling panels, and heated windows) in a controlled test chamber. Although air turbulence affects occupant thermal comfort, this impact was not considered in the present study. The temperature field was monitored using a grid of thermocouples, while air speed near the facade was measured using a particle image velocimetry (PIV) system mounted on a traverse (Karava et al. 2007). The PIV technique was applied for the first time for measurements near a full-scale facade. Previous work by Hosni and Jones (2002) considered application of PIV in a large-scale room but focused on the evaluation of air velocity field around a standing person. It is the first time that this type of configuration is used for measurements near full-scale facades. The effect of the presence of a roller shade on the window was also studied. Following the experimental measurements, a three-dimensional numerical CFD model was developed in order to validate the software and be able to generalize temperature and air velocity modeling near facades.

In commercial buildings, the air-conditioning system near the perimeter provides heating to the zones close to cold facades, with outlets often positioned under the windows or nearby ceiling diffusers. It is quite common in Canada to have both a perimeter heating system and an air-based HVAC system; however, the presence of two systems increases capital and operational costs. Larsson and Moshfegha (2002) investigated the effect of thermal performance, window bay, and displacement ventilation on the downdraught and showed that well-insulated windows may result in reduced air velocity and turbulence intensity. In another study, Rueegg et al. (2001) measured draughts and thermal comfort near a facade with different window sizes and insulation characteristics under variable outdoor temperatures. They found that the window frame is a more critical parameter than the glazing itself and that highly insulated window frames will ensure acceptable draughts. In the presence of solar radiation, glass temperatures may be still low, depending on the type and properties of glazing. In the winter, shading devices can act as a radiant barrier, absorbing solar radiation and decreasing the radiation heat exchange between the cold window surface and the occupant. Heat transfer through windows with shading devices has been extensively studied experimentally (Collins et al. 2001; Duarte et al. 2001; Collins & Harrison 2004) and numerically (Collins et al. 2000; Collins 2004).

The second part of the paper explores the possibility of employing a high-performance envelope to eliminate the need for perimeter heating and utilize the main HVAC system to provide both ventilation and all air-conditioning required. Measurements in the test chamber and near a glazed facade in a real office space were, therefore, conducted without any perimeter heating to examine thermal and airflow conditions in the space under winter conditions. The experiments also considered the impact of solar radiation and shading attachments on thermal comfort (Bessoudo 2008). The experimental results were used to validate a numerical simulation model, which was developed based on a finite difference thermal network approach. Furthermore, a CFD model was utilized to investigate the conditions required to eliminate secondary perimeter heating--namely glazing properties, diffuser location, and supply velocity.


Experimental Facilities

The first part of the study consists of experimental measurements of air velocity and temperature gradients near a glazed facade with different perimeter heating configurations. The experiments were performed in the Hydro-Quebec research laboratories (Laboratoire des Technologies de l'Energie) in Shawinigan, Quebec. An environmental test chamber with a 4.2 m glazed facade that separates a cold room from the test space (3.65 m deep X 2.3 m high) was used for the measurements (Figure 1). The temperature in the cold room was controlled (using a cooling system) and could be decreased to -18[degrees]C to imitate cold (winter) outdoor conditions.


The test space is equipped with three controllable perimeter heating devices: (1) a 690 W baseboard heater under the windows, (2) electrically-heated windows (108-216 W/[m.sup.2] power), and (3) a radiant ceiling panel right above the windows. Only one heating source was used during each measurement in order to study them separately and compare their performance. The ventilation system was turned off so that air movement due to natural convection only would be measured during the experiments. The window properties are shown in Table 1. The windows were also furnished with interior retractable standard commercial roller shades, which were used to study the impact of a shading layer on airflow and temperature patterns. The experimental equipment consisted of the following:
Table 1. Window Properties of the Test Facade

Window thickness, m               0.025
Power for heating, W/[m.sup.2]  108-216
Visible transmittance, %             73
U-value, W/[m.sup.2]*K            1.873
SHGC                               0.67

* A PIV system for air velocity field measurements

* A two-dimensional automated traverse system for moving the PIV system (controlled by a computer)

* A grid of thermocouples (constant and movable) for air and surface temperature measurements

* An infrared camera for surface temperature measurements

* A data acquisition system (connected to the computer) for automatic air velocity and temperature data collection and processing

The experiments included air velocity and temperature field measurements near the facade for the following cases:

* Baseboard heater on

* Baseboard heater on and windows covered with roller shade

* Heated windows ON

* Radiant ceiling panel ON

* No heating (cold windows)

The following sections present the experimental method and results for each case.

Air Velocity Measurements Using Particle Image Velocimetry

PIV is a measurement technique for obtaining instantaneous whole field velocities. Seeding particles, which are good light scatterers, are suspended in the flow of interest, a two-dimensional plane of the flow is illuminated by a strobo-scopic light source (using a pulsed laser), and a charge coupled device (CCD) camera placed perpendicular to the light sheet records scattered light from the seeding particles. A pair of light pulses freezes two consequent position fields, time t apart. The camera images are divided into rectangular regions called interrogation areas (or interrogation regions) and, for each of these interrogation areas, the image from the first and the second pulse of the light sheet are spatially correlated (cross-correlation using Fast Fourier Transform) to produce an average particle displacement vector. When this process is completed for all the interrogation regions, a vector map of average particle displacements is obtained. For general information regarding the PIV technique, see Raffel et al. (1998).

Two-dimensional PIV measurements were performed to obtain the time-averaged air velocity field due to natural convection on a vertical plane perpendicular to the window surface. Figure 2 shows the experimental set-up for PIV measurements. Olive oil particles (seeding) of about [10.sup.-6] m diameter (probability density function of size has a peak at 1 mm) were generated using Laskin nozzles (atomizer). In the present experiment, two Laskin nozzles operating at 10 psi (69 kPa) were sufficient to provide uniform seeding in the flow of interest. Seeding particles were illuminated with a 532 nm wavelength laser sheet generated using a two-cavity (0.12 J each) Nd:YAG laser. The thickness of the laser sheet at the measurement plane was about 0.0015 m, and it was obtained using a combination of spherical and cylindrical lenses. A pair of light pulses--1800 us apart--was produced at a frequency of 8 Hz (the pulse frequency can be adjusted in the range between 1 and 15 Hz). The time between pulses was selected in order to achieve the best temporal and spatial resolution. It should be noted that due to low energy (0.12 J at 532 nm) and small pulse duration (9 ns), laser-induced thermal effects are insignificant. A CCD camera (1600 X 1186 pixels) placed perpendicular to the light sheet was equipped with a 105 mm lens to cover an image (field of view) of 0.152 m (width) X 0.112 m (height). A synchronization board was used to control the timing between the laser pulses and the camera shutter open time by using software installed in a PC desktop computer.


The camera and the laser were placed on a two-dimensional motorized traverse system, as shown in Figure 2, in order to map the air velocity at different heights and distances from the window. The traverse system was controlled from a computer to move in a vertical and a horizontal (closer/further away from the facade) direction. PIV measurements were performed in 12 successive vertical positions, as shown in Figure 3. The minimum measurement distance from the glass varied between 0.098-0.5 m. The lowest measurement point (position 1) starts at 0.43 m from the floor, while the highest point is located at 1.78 m from the floor. For each position, 70 pairs of images (sample size) were recorded to capture the flow for 9 s. The duration of the signal (and the sample size) is relatively small, but this option was limited by the data storage capacity of the system. The camera images were divided into rectangular regions--i.e., interrogation areas--and a 32 X 32 pixel vector field with 50% overlapping was used, as it was found to be the best compromise between spatial resolution and velocity dynamic range (maximum displacement within interrogation to result in successful correlations). With this overlapping, the resolution of the velocity field was increased to 16 X 16 pixels. Typical errors in PIV are the interrogation error (produced from random correlations during the image FFT-processing that do not represent true particle displacements), which is attributed to poor particle seeding, strong flow velocity gradients, or strongly three-dimensional motion in the flow, the mean-bias error, and the root-mean-square (RMS) error. Measurements in the present study produced more than 98% valid vectors. Valid vectors are subject to errors and/or measurement uncertainties, the detailed estimation of which is rather complex. Typically, in carefully conducted PIV experiments for turbulence levels up to 20% and sample size equal to 70, the anticipated error in the mean flow field is in the order of 3%.


With the baseboard heater (690 W) operating below the windows, natural convection results in pronounced upward airflow near the facade. The average glass temperature (above the heater) was 28[degrees]C. The velocity contour plots are shown in Figure 4a. Away from the window and up to 0.2 m away from the glass (100 mm in Figure 4), the velocities are small (0.05 m/ s), and the flow is more-or-less horizontal. Closer to the facade (and above the baseboard), the velocity increases and the flow turns upwards, reaching a maximum velocity (0.48 m/s) at 1.65 m from the floor (1200 mm on z-axis) and 0.1 m from the glass (the camera cannot see closer to the window because of frames and other obstructions). The study considered the impact of shading on air velocity and temperature fields. The opaque roller shades were closed, covering the entire glass area and extending directly above the baseboard heater (therefore they were easily heated). The velocity contour plots are shown in Figure 4b. In this case, the origin is 6 cm from the shade surface, which is warmer than the glass--the average shade surface temperature was 32[degrees]C. The measured velocities are, therefore, higher and reach a maximum of 0.54 m/s at 1.75 m from the floor. Compared to the previous case (without shading), the local velocity differences vary with height, reaching a maximum difference of 0.3 m/s at 1.3 m from the floor (the origin of the y-axis is at 0.4 m from the floor). It should be noted that, in Figure 4, the origin is at 0.45 m above the floor and 0.1m from the window.


The next experiment considered heated windows for the perimeter heating source. The glazing was heated and stabilized at 36[degrees]C. The velocities are very small and become noticeable (>0.05 m/s) only higher than 1.7 m from the floor--the flow direction is almost horizontal. Similar results were observed for the case of the radiant ceiling panel. Finally, experiments without any heating source showed that velocities remain quite low (<0.1 m/s), even with an outdoor temperature of -18[degrees]C.

Temperature Field Measurements

The thermal environment near the facade was monitored using thermocouples placed at different heights and distances from the windows. Some of them were placed on a bar mounted on the traverse system, in order to move automatically and measure air temperature and velocity simultaneously (at different heights and distances from the glass). A data acquisition system recorded the measured temperatures every 15 s. Therefore, 3-4 temperature readings were obtained for each frame (PIV position), and then they were averaged over the velocity measurement time (approximately 60 s). In this way, we managed to have a dense grid of continuous temperature readings near the facade. At the same time, the data acquisition system recorded the readings of other thermocouples further away from the windows and at different heights, as well temperatures of all room surfaces. In some cases, an infrared camera was also used.

The temperature data were processed and converted into contour plots in order to see the impact of different perimeter heating configurations (baseboard heater, heated windows, and radiant ceiling panel) on thermal stratification and air temperature increase near the facade. Figure 5 shows the air temperature field as a function of distance from the windows (x-axis) and distance from the floor (y-axis) for each studied case. In all cases, the air temperature near the center of the room remained near 24[degrees]C (controlled heating setpoint). Table 2 summarizes the results.
Table 2. Summary of Temperature Field Measurements

Heating System Used            Maximum       Maximum      Maximum
                             Temperature,    [DELTA]T     [DELTA]T
                              [degrees]C   (Horizontal)  (Vertical)

Baseboard heater                   32          7.3             3

Baseboard heater and closed      33.5           10           4.8
roller shades

Heated windows                   29.5          4.4           2.2

Radiant ceiling panels             29          4.4           3.5

The highest temperature variations occur for the case of baseboard heater with shaded windows; the temperature reaches 33[degrees]C near the shade (1.75 m from the floor), while it remains at 24[degrees]C 0.5 m from the glass. The temperature difference between floor and ceiling (2.3 m) reaches 4.8[degrees]C right above the baseboard heater. For the baseboard heater without shading, the stratification is smaller, although the horizontal temperature variation is higher than 7[degrees]C within 0.5 m from the glass. For heated windows, which act as uniform vertical radiators, the stratification is clearer (horizontal contours) up to 0.4 m from the windows, but small compared to the baseboard cases (2.2[degrees]C). Finally, the impact of the radiant ceiling panel, which is located 0.5 m from the glass and at 2.3 m height, is evident; the air temperature changes significantly only near the panel and above 1 m from the floor.

Temperature field measurements without any perimeter heating system (the only heat sources were the people and the computer in the test room) were performed in order to quantify a reference thermal stratification for comparison purposes. The cold room temperature was kept at -18[degrees]C, and the glass temperature dropped to 13[degrees]C. A 2[degrees]C temperature difference was observed between the floor and the ceiling (as expected, there are minimal horizontal differences since there is no perimeter heating operating). The fact that air temperature is not very low, even under very cold outdoor conditions, led to further investigation of the possibility of employing a high-performance envelope to reduce/eliminate the need for secondary perimeter heating.


Air temperature and velocity near the facade greatly affect both the thermal comfort of the occupants and the heat losses to the exterior. For example, it is uncomfortable for people to sit near cold windows or near high air velocities (higher than about 0.2-0.3 m/s). Therefore, it is important to be able to predict drafts and low surface temperatures in order to provide the required perimeter heating, as well as to provide input on sizing fenestration areas for different types of glazings. For generalization of results, reliable simulation models need to be developed. It was decided that the same space (used for the experiments described above) would be simulated in order to have a validated CFD model and compare simulated and measured results. The model includes examination of the three different heating scenarios (baseboard heaters, heated windows, and radiant ceiling panels) and considers the impact of internal shading devices on the thermal environment near the facade (Candanedo et al. 2007).

There are several commercial programs available to solve the momentum, mass, and energy equations. Airpak[C] (Fluent 2007) works with the FLUENT CFD solver engine for thermal and fluid-flow calculations. The first step in creating the model was to replicate the dimensions and features of the test chamber. Usually, as the numerical model approaches real conditions, the numerical solution provides results closer to the measured values for the velocity and temperature gradients in the room. For this reason, a three-dimensional simulation was preferred.

Model Description

The dimensions of the walls and windows were set equal to those of the actual room. Special care was taken to reproduce the geometry of the framing between the windows and the baseboard heater located under them, as well as to reproduce the position of the roller shades (vertical layers, 0.04 m from the glazing surface). In order to reproduce the internal gains in the room during the experiments, the model included two people (58.2 W/m each), a computer (120 W), one LCD computer screen (75 W), and the data acquisition system (12 W).

Airpak solves the Navier-Stokes equations for mass, momentum, and energy transport by calculating laminar flow with heat transfer. In the case of turbulent flow and radiative heat transfer, additional transport equations are solved (Fluent 2007). All the simulations are run for the steady-state case. The main equations used in the Airpak model are listed below. The equation continuity or mass equation is given by the following:

[[partial derivative][rho]/[partial derivative]t] + [nabla]*([rho]*[[vector].[upsilon]]) = 0 (1)

The transport of momentum is given by

[[partial derivative]([rho]*[[vector].[upsilon]])/[[partial derivative]t] + [nabla]*([rho]*[[vector].[upsilon]]*[[vector].[upsilon]]) = - [nabla]p + [nabla]*[??] + [rho]*[[vector].g] + [[vector].F], (2)

where p is the static pressure, and [tau] is the stress tensor calculated using the following expression:

[??] = [mu][([nabla][[vector].[upsilon]] + [nabla][[[vector].[upsilon]].sup.T]) - [2/3][nabla]*[[vector].[upsilon]]*I] (3)

where [mu] is the molecular viscosity, I is the unit tensor, and the last term represents the effect of volume dilation.

The energy equation for a fluid region is given by

[[partial derivative]([rho]*h)/[partial derivative]t] + [nabla]*([rho]*h*[[vector].[upsilon]] = [nabla][(k + [k.sub.t])[nabla]T] + [S.sub.h], (4)

where k is the molecular conductivity, [k.sub.t] is the conductivity due to turbulent transport ([k.sub.t] = [c.sub.p] * [[mu].sub.t]/P[r.sub.t]), and [S.sub.h] includes the volumetric heat sources defined in the model.

Turbulence Model. For turbulence, the Indoor Zero-Equation Turbulence model was employed since the convergence was faster. This model was developed for indoor airflow simulation (Chen and Xu 1998). Simulations using a k-[epsilon] model were also attempted; a considerable amount of time to reach convergence was required, and air velocities were generally underpredicted, especially near heat sources. The Indoor Zero-Equation uses the following equation to calculate the turbulent viscosity, [[mu].sub./t]:

[[mu].sub.t] = 0.03874 [rho][upsilon]L (5)

where v is the local velocity magnitude, [rho] is the fluid density, and L is defined as the distance from the nearest wall, and the constant is empirical.

Buoyancy and Natural Convection. The Rayleigh number is a parameter employed to categorize the nature of the buoyancy-induced flow. This number represents the qualitative ratio of buoyancy-to-viscosity effects (White 2003). This number in pure natural convection is calculated as follows:

Ra = [g[beta][DELTA]T[L.sup.3][rho]/[mu][alpha]] (6)

where the thermal expansion coefficient, [beta], is defined as

[beta] = -[1/[rho]][{[partial derivative][rho]/[partial derivative]T}.sub.p], (7)

and the thermal diffusivity, [alpha], is defined as

[alpha] = [k/[rho]*[c.sub.p]]. (8)

When the Rayleigh number is less than [10.sup.8], the flow is buoyancy-induced laminar flow for the case in front of a vertical wall. The transition to turbulence occurs over the range of [10.sup.8] < Ra < [10.sup.10]. Using the software, it is possible to obtain an estimation of the Rayleigh and Prandlt numbers: Ra = 8.2 X [10.sup.10] and Pr = 0.74. Thus, we are working in the transitional region, which also makes the solution of the equations more cumbersome.

In the Airpak model, the Boussinesq model was selected, although it was also possible to define the density using the incompressible ideal gas law. The density in the Boussinesq model is treated by the solver as a constant in all solved equations, except for the buoyancy term in the momentum equation:

([rho] - [[rho].sub.0])g[approximately equal to] - [[rho].sub.0]*[beta](T - [T.sub.0])g (9)

where [[rho].sub.0] is the constant density of the flow, [T.sub.0] is the operating temperature, and [beta] is the expansion coefficient. By using the approximation [rho] = [[rho].sub.0](l -[beta][DELTA]T), the [rho] is eliminated from the buoyancy term, as shown in Equation 9.

Radiation Modeling: Airpak has three models for radiation, a surface to surface radiation model, a discrete ordinates (DO) radiation model, and the solar model. For this simulation, the surface-to-surface radiation model was chosen since there was not shortwave radiation, only longwave radiation. The Airpak 3.0 User's Guide (Fluent 2007) recommends including the radiation heat transfer model for natural convection problems. The surface-to-surface model employs view factors that are calculated for all the surfaces in the room. All the surfaces were included in the radiative exchange. Radiation heat transfer rate is defined as follows:

q = [sigma]*[epsilon]*F([T.sub.surface.sup.4] - [T.sub.remote.sup.4]) (10)

where [[rho].sub.0] is the Stefan Boltzmann constant, [epsilon] the emissivity of the surface (assumed equal to 0.9 for all the surfaces), F is the view factor, [T.sub.surface] is the temperature of the object surface, and [T.sub.remote] is the temperature of the surface for which the view factor was calculated. Airpak has two methods of computing the view factors, a hemicube method and adaptive method. The method employed here was the hemicube. For more details, see the Airpak 3.0 User's Guide (Fluent 2007).

Geometry Meshing. In the Airpak environment, each object is meshed individually. The type of the mesh is unstructured. The number of nodes on the model mesh is 773,784 and contains 741,137 elements. The maximum admitted size for x, y, and z was 0.05 m. It is important to avoid elements with high skew (distortion), since they reduce the accuracy and take longer to converge. The mesh employed in our simulations is hexahedral.

Boundary Conditions. The temperature boundary conditions applied in the model were taken from the experimental measurements. These are presented in Table 3 for the case with the baseboard heater and for the case of no heating source.
Table 3. Temperature Boundary Conditions Used in Simulation Model

Temperature, [degrees]C  Baseboard Heater  No Heating System

Left wall                      25.6              17.5
Right wall                     25.4              16.4
Front wall                     22.3              14.3
Back wall                      25.6              16.3
Floor                          22.7              15.5
Ceiling                        26.7              17.4
Windows                        27                12

CFD Simulation Results

The air velocity and temperature simulation results are presented in Figure 6a-b for the case of baseboard heater and Figure 6c-d for the case of no heating source. For the baseboard case, temperatures are generally in good agreement with the experimental results (1[degrees]C difference) except close to the heater; near this location, the model overpredicts air temperature. This is partly due to the fact that constant boundary conditions were used for the surfaces near the heater (in reality, these surface temperatures change rapidly with time when the baseboard heater turns on). The air velocity simulation results are close to the measured values: average differences vary around 0.1 m/s; the maximum simulated speed was somewhat less than 0.6m/s at 1.7m from the floor.

For the case without any heating system (Figure 6c-d), simulated temperatures match the measured ones (1[degrees]C difference). The model predicted higher temperatures (up to 2[degrees]C) near the top window sections (>2m from the floor), but these are not important for thermal comfort considerations. Finally, calculated air velocities for this case are exactly the same as the measured ones and remain at low levels (<0.1m/s).

The white rectangles in the graphs denote the PIV and temperature measurement areas for comparison with the experimental results (Figures 4 and 5). Overall, comparing the simulation results with the experimental measurements, it was found that the CFD model can be considered reliable and reasonably accurate--especially for air velocity, the agreement is excellent; calculated temperatures were somewhat higher than measured ones. Therefore, CFD modeling can be used to generalize results and investigate the performance of an office facade without a perimeter heating system.


The second part of the study explores the possibility of employing a high-performance envelope to eliminate the need for secondary perimeter heating systems (e.g., baseboard heaters). Instead, the main HVAC system may be utilized to provide both ventilation and all air conditioning required.

Experiments Near a Real Office Facade Section

An experimental setup was established in Montreal in a perimeter office with a large glass facade. The office is located on the 16th floor of an unobstructed building, and its facade is facing 20[degrees] west of south. The width of the facade in the office is over 9 m (the space is large enough to also be used as a research laboratory); therefore, smaller experimental sections were made by isolating six different sections using curtain partitions from ceiling to floor, perpendicular to the facade. Each isolated section is 1.5 m wide, 3.4 m high, and 2.3 m deep. The facade is composed of a 0.8 m high spandrel and two sections of 1.3 m high glazing (sample shown in Figure 7). During the measurement period, perimeter heating (baseboard heaters under the windows) was turned off. Ventilation air supply and space conditioning is provided in the adjacent core zone with air exchange occurring with the perimeter zone via infiltration between the curtain partitions.


Climatic data were collected every minute from exterior sensors placed outside next to the office. Incident solar radiation and daylight levels were recorded using a Licor LI-200 pyranometer and a LI-210 photometric sensor. The pyranometer has a spectral response from 280-2800 nm and it is pre-calibrated against an Eppley precision spectral pyranometer under natural conditions. It has a linear response up to 3000 W/ [m.sup.2], a cosine correction for up to 80[degrees] angle of incidence, and the absolute error is 3%; the response time is 0.01 ms. Several type T thermocouples (error = [+ or -]0.5[degrees]C) were used to record the exterior air temperature, as well as the temperatures of all the interior surfaces (glazing, shading device, floor) in the air gap between the glazing and the shade, and the indoor air temperature.

Indoor environmental conditions were measured using a Bruel & Kjaer Indoor Climate Analyzer type 1213, a collection of instruments that provides a way to measure individual basic environmental parameters (air velocity, humidity, air temperature, and plane radiant temperature in two directions). In addition, a Bruel & Kjaer Thermal Comfort Meter type 1212 was used to measure the operative temperature. This transducer is able to evaluate the thermal effect that the surrounding objects and surfaces have on the human body. The transducer's size has been designed to closely match the ratio of radiative heat loss to convective heat loss of a human body. Its ellipsoid shape is determined by the need to obtain the same angle factors to the individual room surfaces as for a human body. The transducer can be set at different angles depending on the posture of the human body in question. For the case of a seated person, it is set at a 30[degrees] angle. The indoor climate analyzer and thermal comfort meter were placed 1.5 m from the facade at a height of 1.1 m.

For testing the impact of shading, a roller shade with an average reflectance of 55%, absorptance of 40%, and transmittance of 5% was used. The windows of the facade are double glazed (6 mm/12 mm air space/6 mm) with a low-? coating outside the interior pane and argon filling. At normal incidence, the glazing has a total solar transmittance of 36% and a visible transmittance of 69%; the U-value is 1.59 W/[m.sup.2] K, and the SHGC is 0.37.

Measurements were taken for a period of three months, from January to March 2007. The experiments included the impact of variable weather conditions (outdoor temperature and solar radiation) and shading attachments on the air temperature field near the facade, as well as the impact on human comfort indices (Bessoudo 2008). The experimental results showed that, during sunny winter days, not only is perimeter heating not required, but overheating and thermal discomfort could occur if shading protection is not provided (Figure 8). Operative temperature reaches 30[degrees]C under the presence of solar radiation. Keeping the roller shade closed results in improved thermal comfort conditions. During cold, cloudy days, perimeter heating might be needed in some cases (Bessoudo et al. 2007). However, if a high-performance glazing is used, the need for turning on perimeter heating could be eliminated.


In the framework of improving thermal comfort conditions near windows with and without shading, and also studying the possibility of eliminating perimeter heating, a generalized numerical transient thermal model was developed using a finite difference thermal network approach (Tzempelikos et al. 2007). The space is divided into a series of nodes with interconnecting paths through which energy flows. Each surface, mass layer, and air component is represented by a node that is connected by one or more thermal resistances to other adjacent nodes. The thermal resistances in the network represent the three heat flow mechanisms between two nodes: convection, radiation, and conduction. An energy balance is applied at each node for each time step to determine the temperature of each node as a function of time. Starting from a set of initial conditions, the general form of the explicit finite-difference model corresponding to node i and time-step p is (Athienitis & Santamouris 2002):

[T.sub.i.sup.p + 1] = [[DELTA]t/[C.sub.i]]*{[q.sub.i] + [SIGMA] [([T.sub.j.sup.p] - [T.sub.i.sup.p])/[R.sub.ij]]} + [T.sub.i.sup.P] (11)

where T is the temperature, p + 1 represents the next time-step, j represents all nodes connected to node i, [R.sub.ij]: is the thermal resistance connected to nodes i and j, [C.sub.i] is the capacitance of node i, and q is a heat source at node i. Using this equation, temperatures of all nodes are calculated at each time step. The heating-and-cooling load is computed using appropriate proportional and integral control constants.

Components of the indoor thermal environment--mean radiant temperature (MRT), operative temperature, and radiant temperature asymmetry--are also computed at each time-step. Mean radiant temperature for a person exposed to solar radiation is calculated using the algorithm developed by La Gennusa et al. (2005):

[T.sub.r.sup.4] = [N.summation over (i)][F.sub.p - i]*[T.sub.i.sup.4] + [1/[[epsilon].sub.s]*[sigma]]([[alpha].sub.irr,d] [M.summation over (j = 1)][F.sub.p - j]*[I.sub.d,j] + [[alpha].sub.irr, b]*[f.sub.p]*[I.sub.b]) (12)

where [[epsilon].sub.s] is the emissivity of the person, [[alpha].sub.irr,d] is the absorptivity of a person for diffuse solar radiation, [[alpha].sub.irr,b] is the absorptivity of a person for direct solar radiation, [F.sub.p - j] is the view factor from the person to any nonopaque element of the building envelope, and [f.sub.p] is the so-called "projected area factor." This equation takes into account the effect of three separate components on the MRT: low-temperature surfaces, absorbed diffuse solar radiation, and absorbed direct beam solar radiation. Radiant temperature asymmetry, the difference between the plane radiant temperatures of two opposite sides of a small plane element, is determined using the calculation method outline in ANSI/ASHRAE Standard 55-2004, Thermal Environmental Conditions for Human Occupancy (ASHRAE 2004).

The model was validated using the experimental results (sample comparison in Figure 8) and was used to predict temperatures and thermal comfort indices near glazed facades. It also includes detailed thermal comfort calculations based on the transient two-node model of Gagge et al. (1971), which considers the human body as composed of two concentric thermal compartments: core and skin. This model calculates the indices of thermal sensation and thermal discomfort using a recommended 60 s time step (ASHRAE 2004). Gagge et al. (1986) and Grivel et al. (1989) compared the two-node model with other numerical models and experimental measurements, respectively. An overview of the basic model and a discussion of experimental techniques used to assess the perception of the indoor environment are described by Zmeureanu and Doram-ajian (1992). The simulation results (operative temperature and asymmetry indices) showed that perimeter heating might not necessarily be needed if a high-performance building envelope is used (Bessoudo et al. 2008).

CFD Simulation for Further Investigation

Although the finite-difference network model is adequate for performing simulation of transient thermal environments, it handles air temperature as a single node and air velocity as an input variable and is, therefore, unable to model the complex airflow and temperature stratifications that can occur within the space. Thus, it cannot be used to predict local discomfort caused by temperature stratification or air speed. A CFD model was further developed to perform a more detailed analysis and investigate the conditions required to eliminate the secondary perimeter heating system. The objective was to examine the possibility of using the main HVAC system to heat the space during the winter without any other assistance.

Model description. A representative perimeter zone office with dimensions 3.4 X 3.0 X 3.0 m was modeled. The exterior wall has a window of dimensions 2.4 X 2.8 m above a spandrel of height 0.8 m. The part of the exterior facade on both sides of the window has a width and thickness of 0.2 m. Both the spandrel and exterior facade section have a thermal conductivity of 0.07 W/m*K. The glazing is modeled as having a U-value of 2 W/[m.sup.2]*K (double-glazed, low- [epsilon], Argon-filled).

An airflow and thermal analysis was completed in Airpak for the case of a heated office under winter conditions with the diffuser for supply heating at different locations. The most interesting case was when the diffuser was placed below the windows, where the baseboard heaters are supposed to be. Heating was supplied to the space through a diffuser located on the floor parallel to the facade, as shown in Figure 9. The air is coming out at 28[degrees]C and 0.5 m/s. The opening area of the diffuser was determined based on the required heating supply. Different shapes were tried; however a long strip (0.15 X 2 m long) was finally used. Minus twenty degrees Celsius was used as a boundary condition for the exterior wall (glass and spandrel sections) to investigate the space performance under extremely cold conditions. The rest of the surfaces (side walls, floor, and ceiling) were assumed adiabatic (23[degrees]C surface temperature).


The indoor zero-equation turbulence model was selected for the flow regime since it was developed specifically for indoor airflow simulations and is ideally suited for predicting indoor airflows that consider natural convection, forced convection, and displacement ventilation (Fluent 2007). Under-relaxation values for pressure (0.2) and momentum (0.1) were selected based on convergence criteria for flow and energy. After refining the grid size for the mesh, it was determined that a 0.095 X 0.095 m control volume size was sufficient to generate adequate results. A more refined mesh was generated for components such as diffuser and return air vent. In total, the model contained 83,790 elements and 89,700 nodes.

CFD Simulation Results. Figure 9 presents the air velocity results and the space configuration. Higher velocities are only observed right above the diffuser (air is supplied at 0.5 m/s). The hot air rises and the flow is maintained upward, thus avoiding downdraughts. The velocity magnitude in the rest of the room is very low (0.1-0.2 m/s) and will not cause any discomfort to the occupants sitting near the facade (0.5 m away or further). The results suggest that, using this type of heating configuration (air coming from the main HVAC system), secondary perimeter heating is not needed to mitigate cold downdraft or to increase operative temperature near the facade.

The operative temperature results (in [degrees]C) are shown in Figure 10. The hot air coming out of the floor diffuser (using an underfloor air distribution system) results in an impressively uniform temperature distribution in the room. Very close to the diffuser (placed 0.3 m from the facade), the temperatures are high (air is provided at 28[degrees]C), but this is limited only to a strip of air volume extending vertically [+ or -]0.1 m from the opening. Further than 0.5 m from the facade (where people usually sit), the temperatures are all in the comfortable range (23[degrees]C-24[degrees]C) without any spatial variations. The warm air near the glass prevents thermal discomfort caused by colder window surface temperatures.



In cold climates, perimeter heating is often used as a secondary heating system near the windows in order to prevent thermal discomfort due to cold window temperatures. In this study, air velocity and temperature gradients near a glazed facade with three different perimeter heating variations--baseboard heater, radiant ceiling panel, and electrically--heated windows--were initially studied in a test chamber with a glass facade separating the test space from a controlled cold room.

For the air velocity measurements, a PIV system was mounted on a two-dimensional automatic traverse system for scanning entire planes close to the facade--the first time this type of configuration was used for measurements near full-scale facades. The study also included the impact of interior shading on airflow and thermal conditions near the windows; the impact of air turbulence was not studied. It was found that the baseboard heater induces upward flow with velocities that reach 0.48 m/s. When the windows were covered with roller shades (extending above the heater), the maximum velocity increased, reaching 0.54 m/s. Radiant ceiling panels and heated glazings did not generate high velocities (< 0.1 m/s), as they mainly acted as radiant heat sources.

The air temperature field near the facade was monitored using a thermocouple grid, part of which was moving together with the PIV system. When the baseboard heater was on, high temperature gradients were measured both vertically and horizontally, especially when shading attachments were used. The thermal stratification (2.3 m height) above the baseboard was 4.8[degrees]C; air temperature near the glass (above the heater) reached 33[degrees]C, while 0.5 m from the glass it was only 24[degrees]C. Even without the shading layer, the temperature difference between the air near the windows and 0.5 m away was 7[degrees]C. Heated windows and radiant panels resulted in smaller temperature differences near and away from the facade (2-4[degrees]C). Also, the fact that measured air temperatures were not very low under very cold outdoor conditions (without any perimeter heating) initiated a further investigation of potential elimination of secondary heating systems.

After the experimental measurements, a three-dimensional numerical CFD model (steady-state) was developed to validate the software and be able to generalize temperature and air velocity modeling near facades. The results were in good agreement with the experimental measurements. However, since the flow regime is transitional, it causes the problem to become more unstable and it takes longer to converge. Another interesting approach would be to run a transient simulation and acquire solutions at different times for further examination.

The second part of this paper explores the possibility of employing a high-performance envelope to eliminate the need for secondary perimeter heating and utilize the main HVAC system to provide both ventilation and all air conditioning required. Measurements near a glazed facade in a real office space were conducted without any perimeter heating to examine thermal and airflow conditions in the space under winter conditions. The experiments also considered the impact of solar radiation and shading attachments on thermal comfort, and a transient finite difference thermal model was developed and validated using the experimental results. It was found that, during cold sunny days, perimeter heating is not required; shading protection is necessary to protect from overheating. During cold, cloudy days, perimeter heating might be needed in some cases depending on the thermal properties of the glazing.

Finally, another CFD model was further developed in order to perform a more detailed analysis and investigate the conditions required to eliminate the secondary perimeter heating system. It was found that, for a facade covered by 60% glass (double-glazed, low-?, Argon-filled), the main HVAC system could provide all the necessary heating through an underfloor air-distribution system, even under extremely cold outdoor conditions (-20[degrees]C). By providing hot air at 28[degrees]C and 0.5 m/s, and by placing the diffuser close to the facade (0.3 m away), uniform temperature distribution in the comfortable range and low air velocities (0.1 m/s) were achieved inside the entire room (0.5 m from the facade and deeper). Therefore, a secondary perimeter heating system (e.g., baseboard heaters) is not necessarily needed.


The authors would like to thank Hydro-Quebec and, especially, Dr. Louis Handfield from the Laboratoire des Technologies de l'Energie (LTE) in Shawinigan, Quebec, for their kind support and use of facilities. This study is part of a research project in the Solar Buildings Research Network (, mainly funded by the Natural Sciences and Engineering Research Council of Canada (NSERC) under the strategic research networks program. The input from Professor R. Zmeureanu (Concordia University) is also acknowledged with thanks.


ASHRAE. 2004. ANSI/ASHRAE Standard 55-2004, Thermal Environmental Conditions for Human Occupancy. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc.

Athienitis, A.K., and M. Santamouris. 2002. Thermal Analysis and Design of Passive Solar Buildings. London: James & James, Ltd.

Bessoudo, M. 2008. Thermal comfort conditions near highly glazed facades: an experimental and simulation study. Masters thesis, Department of Building, Civil and Environmental Engineering, Concordia University, Montreal, Canada.

Bessoudo, M., A. Tzempelikos, A.K. Athienitis, and R. Zmeureanu. 2007. The impact of shading on thermal comfort conditions in perimeter zones with glass facades. Proceedings of 2nd PALENC and 28th AIVC Conference: Building Low Energy Cooling and Ventilation Technologies in the 21st Century, Crete, Greece.

Candanedo, L.M., P. Karava, M. Bessoudo, A. Tzempelikos, A.K. Athienitis, and L. Handfield. 2007. Airflow and thermal simulation in a controlled test chamber with different heating configurations using CFD modeling. Proceedings of Joint 32nd Solar Energy Society of Canada and 2nd Solar Buildings Research Network Conference, Calgary, Canada.

Chen Q., and W. Xu. 1998. A zero-equation turbulence model for indoor airflow simulation. Energy and Buildings Journal 28(2):137-44.

Collins, M., and S.J. Harrison. 2004. Calorimetric analysis of the solar and thermal performance of windows with interior louvered blinds. ASHRAE Transactions 110(1):474-86.

Collins, M. 2004. Convective heat transfer coefficients from an internal window surface and adjacent sunlit venetian blind. Energy and Buildings Journal 36:309-18.

Collins M., S.J. Harrison, D. Naylor, and P.H. Oosthuizen. 2000.Numerical modeling of radiative and convective heat transfer from an irradiated complex window assembly. Proceedings of IMECE, Orlando, FL, pp. 1-12.

Collins, M., S.J. Harrison, D. Naylor, and P.H. Oosthuizen. 2001. An interferometric study of convective heat transfer from an irradiated complex window assembly. Proceedings of IMECE, New York, NY, pp. 1-13.

Duarte, N., D. Naylor, P.H. Oosthuizen, S.J. Harrison. 2001. An interferometric study of free convection at a window glazing with a heated venetian blind. HVAC & Research, 7(2):169-84.

Fluent, Inc. 2007. Airpak 3.0 User's Guide. Lebanon, NH: Fluent, Inc., pp. 853-74.

Gagge A.P., A.P. Fobelets, L.G. Berglund. 1986. A standard predictive index and human response to the thermal environment. ASHRAE Transactions, 92(2):709-31.

Gagge, A.P., J.A.J. Stolwijk, Y. Nishi. 1971. An effective temperature scale based on a simple model of human physiologicalresponse.ASHRAETransactions 77(1):247-62.

Grivel, F., A. Hoeft, V. Candas. 1989. Body temperatures and thermal perceptions during warm and cool periodic transients: Influence of clothing. Proceedings of the Annual International Industrial Ergonomics and Safety Conference, pp. 217-224.

Hosni, M.H, and B.J. Jones. 2002. Development of a PIV for measuring air velocity in large-scale room airflow applications. ASHRAE Transactions 108(2).

Karava P., A. Tzempelikos, L.M. Candanedo, M. Bessoudo, A.K. Athienitis, and L. Handfield. 2007. Airflow and heat transfer measurements near a facade with different heating configurations. Proceedings of Joint 32nd Solar Energy Society of Canada and 2nd Solar Buildings Research Network Conference, Calgary, Canada.

La Gennusa, M., A. Nucara, G. Rizzo, and G. Scaccianoce. 2005. The calculation of the mean radiant temperature of a subject exposed to solar radiation--A generalised algorithm. Building and Environment Journal 40:367-75.

Larsson, U., and B. Moshfegha. 2002. Experimental investigation of downdraught from well-insulated windows. Building and Environment Journal 37:1073-82.

Raffel, M., C. Willert, and J. Koupenhans. 1998. Particle Image Velocimetry: A Practical Guide. Heidelberg, Germany: Springer.

Rueegg T., V. Dorer, and U. Steinemann. 2001. Must cold air down draughts be compensated when using highly insulating windows? Energy and Buildings Journal 33:489-93.

Tzempelikos, A., M. Bessoudo, R. Zmeureanu, and A.K. Athienitis. 2007. Simulation of thermal comfort conditions in highly-glazed perimeter zones with shading devices. Proceedings of Joint 32nd Solar Energy Society of Canada and 2nd Solar Buildings Research Network Conference, Calgary, Canada.

White, F.M. 2003. Fluid Mechanics, 5th Edition. New York: University of Rhode Island and McGraw-Hill Higher Education.

Zmeureanu R., and A. Doramajian. 1992. Thermally acceptable temperature drifts can reduce the energy consumption for cooling in office buildings. Building and Environment Journal 27:469-81.

Athanassios Tzempelikos, PhD

Associate Member ASHRAE

Panagiota Karava, PhD

Associate Member ASHRAE

Luis Miguel Candanedo

Mark Bessoudo

Andreas K. Athienitis, PhD, PE


Athanassios Tzempelikos is an assistant professor at Purdue University, West Lafayette, Indiana. Panagiota Karava is an assistant professor at the University of Western Ontario, London, Ontario, Canada. Luis Miguel Candanedo and Mark Bessoudo are graduate students and Andreas K. Athienitis is a professor and research chair at Concordia University, Montreal, Quebec, Canada.
COPYRIGHT 2009 American Society of Heating, Refrigerating, and Air-Conditioning Engineers, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2009 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Tzempelikos, Athanassios; Karava, Panagiota; Candanedo, Luis Miguel; Bessoudo, Mark; Athienitis, And
Publication:ASHRAE Transactions
Article Type:Report
Geographic Code:1USA
Date:Jan 1, 2009
Previous Article:Improving efficiency in a campus chilled water system using exergy analysis.
Next Article:Modeling for predicting frost behavior of a fin-tube heat exchanger with thermal contact resistance.

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters