Sensitivity analysis of the thermal performance of radiant and convective terminals for cooling buildings

Energy and Buildings 82 (2014) 482–491

Contents lists available at ScienceDirect

Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild

Sensitivity analysis of the thermal performance of radiant and convective terminals for cooling buildings J. Le Dréau, P. Heiselberg Department of Civil Engineering, Aalborg University, Sohngaardsholmsvej 57, DK-9000 Aalborg, Denmark

a r t i c l e

i n f o

Article history: Received 22 September 2013 Received in revised form 27 April 2014 Accepted 1 July 2014 Available online 9 July 2014 Keywords: Floor cooling Cooled wall Cooled ceiling Active chilled beam Emitters Sensitivity analysis Cooling need Convective heat transfer coefficient Air temperature stratification Comfort

a b s t r a c t Heating and cooling terminals can be classified in two main categories: convective terminals (e.g. active chilled beam, air conditioning) and radiant terminals. The mode of heat transfer of the two emitters is different: the first one is mainly based on convection, whereas the second one is based on both radiation and convection. In order to characterise the advantages and drawbacks of the different terminals, steadystate simulations of a typical office room have been performed using four types of terminals (active chilled beam, radiant floor, wall and ceiling). A sensitivity analysis has been conducted to determine the parameters influencing their thermal performance the most. The air change rate, the outdoor temperature and the air temperature stratification have the largest effect on the cooling need (maintaining a constant operative temperature). For air change rates higher than 0.5 ACH, differences between terminals can be observed. Due to their higher dependency on the air change rate and outdoor temperature, convective terminals are generally less energy effective than radiant terminals. The global comfort level achieved by the different systems is always within the recommended range, but differences have been observed in the uniformity of comfort. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Differences can be observed between offices built nowadays and the ones built in the eighties or before. First of all, the level of insulation and air tightness of buildings has increased due to strengthening of the different building regulations. A better treatment of daylight by architects and the development of new products have led to an increase of the glazed area of buildings; fully glazed fac¸ades are becoming more widely installed. The use of buildings has also changed with the emergence of computers and other electronic devices, thus increasing internal heat loads. For these reasons and also due to a raised focus on thermal comfort, more cooling systems are installed in offices. In the European Union, the cooled area in non-residential buildings has increased by 45% between 2000 and 2010, resulting in an electricity consumption of 95 TWh for the EU-15 members [2]. This situation creates serious supply difficulties during peak load periods, especially in southern European countries such as Spain or Italy [3]. Convective terminals are the most widely installed cooling system, despite their high initial costs, high energy use and often unacceptable indoor climate. Occupants sometimes complain

E-mail addresses: [email protected] (J. Le Dréau), [email protected] (P. Heiselberg). http://dx.doi.org/10.1016/j.enbuild.2014.07.002 0378-7788/© 2014 Elsevier B.V. All rights reserved.

about the noise or the draught of this type of system [4]. Switzerland and the state of Hamburg in Germany have even restricted the installation of full air conditioning systems for buildings [5]. Radiant technology is an alternative to air-based emitters. Contrary to convective terminals, which transfer heat mainly by convection, radiant terminals transfer heat partly by radiation to (or from) the neighbouring surfaces, and partly by convection to (or from) the indoor air [6]. The first radiant cooling system was installed after the First World War, in the Bank of England [7]. In the 1990s, European offices were increasingly equipped with cooled radiant ceilings because of longer overheated periods during summer time [8]. In 2004, a cooled radiant floor was installed in the humid climate of Bangkok airport [9]. More recently, radiant walls have been introduced to the market. Most of the studies comparing radiant and air-based systems conclude to the lower energy use of radiant systems [10–16]. Radiant systems are an efficient way of transporting energy [4], mainly due to the higher heat capacity of water and the reduced fan usage. Moreover, the large surface of exchange of radiant systems allows the use of source temperature closer to the room temperature, increasing the efficiency of production systems. The total energy savings oscillate between 10 up to 60%, depending on the climate, the source considered, the area of the radiant system and the efficiency of the different components. Fabrizio et al. [16] have

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

483

Nomenclature A ACR Dh Cp Fp-i g H h Q q R RRCB T Ti –j V

surface area (m2 ) air change rate (ACH) hydraulic diameter (m) heat capacity (J/kg K) view factors from the plane to the surface i (calculated according to [1]) total solar energy transmittance height of the surface (m) convective heat transfer coefficient (W/m2 K) heat flow (W) heat flux (W/m2 ) thermal resistance (m2 K/W), surface resistances not included recirculation rate, only applicable for the active chilled beam (h−1 ) temperature (◦ C) temperature difference Ti − Tj (K) volume (m3 )

Greek symbols density (kg/m3 )   mean cooling need (W/m2 ) standard deviation of the cooling need (W/m2 )  Subscripts cond conduction conv convection pr plane radiant radiant rad rad LW long-wave radiation rad SW short-wave radiation i surface i

compared numerically the performance of radiant floor and ceiling systems versus all-air and fan coil systems. Dynamic simulations of a typical office building showed that the cooling energy use is greatly reduced for warm climates, whereas the reduction is smaller for cold climates. In addition to the total energy use, some studies compare the energy need in the space. Differences in the heat balance are noted in several publications [11,15–20], but the effect on the cooling need is not clearly defined: some studies show a higher demand [17] for radiant terminals, whereas some others conclude to a lower [18,19] or similar demand [11,16]. As stated by Djunaedy et al. [20] and Feng et al. [17], “no research can be found that fundamentally studies the differences of the heat transfer process in zones conditioned by an air and a radiant system”. In most of the studies, the dynamic simulations do not highlight the sensitivity of one specific parameter on the cooling need. Parameter variation is needed to emphasize the influencing factors. In this paper, four terminals have been selected (active chilled beam, radiant floor, radiant wall and radiant ceiling) and their thermal performances have been compared. This paper focuses on describing the heat transfer within the space. Therefore, the source and the type of energy used to remove heat have not been taken into consideration. Humidity control has also not been considered, as the problem of humidification or dehumidification has to be treated in the plant, before the air enters the space. The main objective is to identify the case(s) in which the different technologies achieve the best performance in maintaining a constant operative temperature of 26 ◦ C. The robustness of the different cooling systems will be evaluated by performing sensitivity analyses and

Fig. 1. Geometry of the room (dimension in mm). Red dots indicate the different positions considered for the occupant. (For interpretation of the color information in this figure legend, the reader is referred to the web version of the article.)

parameter variations. The parameters varied are related to the outdoor conditions (outdoor temperature, part of direct to total solar radiation), the type of ventilation system (air change rate, air temperature gradient, convective flow in the room), the room properties (emissivity and absorptivity of the internal surfaces) and the position of the person/sensor in the room. A typical European office building has been chosen as the base case and numerical simulations have been performed under steady-state conditions. 2. Case study 2.1. General parameters The case of an office room located in Europe has been chosen to study the influence of the type of terminal on the cooling need (Fig. 1). The internal dimensions of the room have been chosen similar to the PASSYS test cell: 5 × 2.76 × 2.75 m (length × width × height), resulting in a floor area of 13.8 m2 [21]. A window facing south is providing daylight to the room (g-value of 0.6). The thermal characteristics of the building components are given in Table 1. The insulation level of the floor and the roof is relatively high, in order to model the case of a multi-storey building. Internal heat loads are equal to 20 W/m2 [22]. The outdoor conditions have been selected from the weather data of the Design Reference Year (DRY) in Copenhagen, Denmark. It corresponds to a hot summer day (Table 2). The total heat load in the room (internal and solar) is equal to around 40 W/m2 . Table 1 Thermal properties of the construction elements. Surface

R (m2 K/W)

Walls Window Floor Roof

6.66 0.71 10.00 10.00

Table 2 Definition of the base parameters (13th of July, 3 pm). qsolar normal, horizontal (W/m2 ) qsolar diffuse, horizontal (W/m2 ) Solar azimuth (◦ ) Solar height angle (◦ )

296 389 234 47

484

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

2.2. Description of the terminals Four terminals have been tested: radiant wall, radiant floor, radiant ceiling and active chilled beam. The performance of the three radiant systems can be compared to each other, as the activated areas are similar. Radiant terminals are embedded close to the surface, so that their radiant effect applies directly on the internal side of the construction. The cooling effect of these terminals is assumed to be constant over the entire surface, thereby not taking into account the inhomogeneity due to the pipes layout. All four types of terminals are associated with a ventilation system (Dedicated Outdoor Air System) to ensure an acceptable indoor air quality. This system is not equipped with heat recovery. The inlet is assumed to be located at the ceiling, either in the middle or close to a side wall. The type of terminal influences the inlet air temperature (in the room) and thus the convective heat transfer. For radiant systems, the inlet air is equal to the outdoor air temperature. The active chilled beam combines both the supply of fresh air and the cooling unit. Therefore, the inlet air temperature is decreased as a function of the cooling effect and is calculated according to Equation (1), taking into consideration the recirculation rate. The outlet is assumed to be located in the upper part of the room.



Tinlet =

3600 Troom air + Qconv terminal × RRCB × Vroom × air × Cp air

RRCB ACR · + Toutdoor × RRCB + ACR (RRCB + ACR)



(1)

where RRCB = 3.8 ACH = 40 L/s. Two models for air mixing have been simulated: fully mixed or with a temperature gradient of 1.5 K/m (i.e. 4 K difference between the floor and the ceiling). This second model can, for example, simulate the effect of displacement ventilation. The temperature gradient chosen ensures that the local thermal comfort is achieved, i.e. that the vertical air temperature difference between 0.1 m and 1.1 m is lower than 3 K [23]. 2.3. Method of analysis In order to compare the different types of terminals, all simulations have been performed maintaining the operative temperature in the room equal to 26 ◦ C. This parameter ensures that the global thermal comfort is identical for all cases. The operative temperature is defined as the mean value of the radiant and the air temperature. This evaluation of the operative temperature is valid for relative air velocity below 0.2 m/s and for asymmetry between radiant and air temperature below 4 K [1]. For a person sitting, the operative temperature is calculated at a height of 0.6 m, whereas the reference height is 1.1 m for a person standing. The mean radiant temperature (Trad ) is evaluated by the following expressions [1]:

Table 3 Local thermal discomfort limits for cold surfaces (category II - normal level of expectation) [23]. Surface

Radiant temperature asymmetry

Minimum surface temperature

Wall Floor Ceiling

<10 K

≈17 ◦ C (dew point) 19 ◦ C ≈17 ◦ C (dew point)

<14 K

Fig. 2. Subdivision of the room surfaces into 61 sections. Window indicated by hatching.

Therefore, it is assumed that the building has been designed so that such a situation does not occur (by using solar shading). As the operative temperature is the same for all cases (26 ◦ C), the cooling need of the different terminals can be compared (expressed in W/m2 of internal floor area). In addition to the energy use, different comfort parameters will be evaluated based on EN ISO 7730 [23] (similar to ASHRAE 55 [25]). The predicted percentage dissatisfied (PPD) and the uniformity of the comfort level inside the occupied zone will be assessed. Two local thermal comfort parameters will also be evaluated: the surface temperatures and the radiant temperature asymmetry. The limit values for cold surfaces are given in Table 3. 3. Physic of the room 3.1. Room heat balance The conduction through the walls (Qcond ) is calculated using the thermal resistances (Table 1). The heat transfer is assumed to be one-dimensional, and thermal bridges are not considered. The internal heat gains from equipment, lights and people are equal to 20 W/m2 and are modelled as sensible heat load, assuming 40% of this heat load convective (Qconv internal sources ) and 60% radiative (Qrad internal sources ) [26]. The radiant heat loads are distributed on the different surfaces according to their relative area. The cooling

Trad = 0.13(Tpr[up] + Tpr[down] ) + 0.185(Tpr[right] + Tpr[left] + Tpr[front] + Tpr[back] )

(2)

for a person sitting (orientation not fixed) Trad = 0.06(Tpr[up] + Tpr[down] ) + 0.220(Tpr[right] + Tpr[left] + Tpr[front] + Tpr[back] )

(2bis)

for a person standing (orientation not fixed)



4 = where Tpr T 4F i i p-i The factors used in the calculation of the radiant temperature correspond to the projected area factors of a sitting or standing person in the six directions. In the first part of the study, eight positions of the occupant have been tested, either in the middle of the room or 0.6 m away from the walls (Fig. 1). The effect of solar radiation on the radiant temperature has been disregarded due to the large discomfort generated by direct solar radiation [24].

power of the terminal is expressed by Qconv terminal for the active chilled beam or by Qrad terminal for radiant terminals (negative values). In order to model the heat transfer in the room, the space is discretized in a total of 62 nodes: 61 nodes correspond to the different sub-surface temperatures (Fig. 2) and one node to the mean room air temperature. Even though the air temperature is represented by only one node, the model can also account for an air

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

485



Table 4 Model 1 for convection [22].

hhorizontal downward = 0.60

Tsurface−room air Dh2

hconv (W/m2 K) Vertical surface Horizontal upward Horizontal downward

2.5 4.0 0.7

temperature gradient. The 62 unknowns are determined by simultaneously solving the heat balance at the construction level (Eq. (3)) and at the air-node level (Eq. (3bis)). As the objective is to achieve an operative temperature of 26 ◦ C, the heat absorbed by the different terminals (Qrad terminal or Qconv terminal ) is changed iteratively until convergence is achieved (i.e. 26 ± 0.005 ◦ C). Heat balance at each sub-surface (61 in total): Qcond (i) + Qrad SW (i) + Qrad internal sources (i) + Qrad terminal (i) = A(i) × hconv (i) × Tsurface−room or inlet air (i) + Qrad LW (i)

(3)

0=



i

A(i) × hconv (i) × Tsurf−room or inlet air (i)

+ Qventilation + Qconv internal sources + Qconv terminal

(3bis)

(4ter)

- Model 3: mixed convection is modelled by combining the correlations of Alamdari and Hammond for natural convection (Eqs. (4)) and Fisher for forced convection [29]. Fisher investigated the convective heat transfer due to a free horizontal jet, with cold air supplied in the middle of a side-wall (Eqs. (5)–(5ter)). Due to convergence issues, these correlations have been redeveloped to obtain a forced convective flow equal to zero when there is no air circulation. It has to be noted that forced convective heat transfer coefficients are correlated to the inlet air temperature, meaning that the convective heat flux is obtained by multiplying hconv by (Tsurface –Tinlet ), and not by (Tsurface –Troom air ) as for natural convection. hceiling = 0.704 + 0.168(ACR + RRCB )0.8

(5)

0.8

(5bis)

hfloor = 0.064 + 0.00444(ACR + RRCB )0.8

(5ter)

hwalls = −0.110 + 0.132(ACR + RRCB ) Heat balance at the air node:

1/5

- Model 4: mixed convection is modelled by combining the correlations of Alamdari and Hammond for natural convection (Eqs. (4)) and Fisher and Pedersen for forced convection [30]. Fisher and Pedersen studied the convective flow with a radial ceiling diffuser, supplying cold air between 3 up to 12 ACH (Eqs. (6)–(6ter)).

3.2. Models for convection

hceiling = 0.49(ACR + RRCB )0.8

In order to evaluate the influence of convection on the cooling need, four correlations for modelling convection at the internal surfaces of the room (hconv ) have been tested. Two of these models assume natural convection, meaning that the flow is driven by buoyancy forces resulting from surface-to-air temperature differences. The last two models assume mixed convection, which means that both mechanical and buoyancy forces can be important. It has to be noted that the combination radiant panels and forced convection regime is not an incompatible case. For example, Novoselac et al. [27] observed forced convection in a test room equipped of cooled ceiling panels with high aspiration diffuser.

hwalls = 0.19(ACR + RRCB )0.8

(6bis)

0.8

(6ter)

- Model 1: natural convection is modelled with constant coefficients selected according to the surface slope and the type of convective flow observed (Table 4). If the room is cooled down with a radiant floor, the flow at the surface will be stably stratified and lead to a low convective heat transfer (horizontal downward). On the contrary, the flow will be buoyancy-driven in case of cold ceiling (horizontal upward). - Model 2 (base model): natural convection is modelled with dynamic coefficients. Alamdari and Hammond [28] analysed different experimental results from the literature (free plates experiments) and derived correlations for convective heat transfer for both laminar and turbulent cases (Eqs. (4)–(4ter)). hvertical surface

⎡  1/4 6 Tsurface−room air = ⎣ 1.50 H

+(1.23 × Tsurface−room air 1/3 )

⎡ hhorizontal upward =



⎣ 1.40

Tsurface−room air Dh



6 1/6

(4)

1/4 6

+(1.63 × Tsurface−room air 1/3 )



6 1/6

(4bis)

hfloor = 0.13(ACR + RRCB )

(6)

When using mixed correlations, the dominating effect (forced or natural convection) is determined by selecting the model, which gives in the highest convective heat transfer. Iterations are performed to achieve convergence of the solution, which is assessed by a change in the surface temperature lower than 0.01%. 3.3. Radiative heat transfer The building energy balance is affected by two different types of radiative exchange: short-wave radiation, also called solar radiation, and long-wave radiation. Due to different temperatures of the emitting bodies, these two types of radiation have different spectrums of emission. Solar radiation comes from a body heated at ≈5800 K, and 75% of the flux is emitted at a wavelength lower than 1 ␮m. On the contrary, long-wave radiation is emitted by objects at ambient temperature and is concentrated in the infra-red region (wavelength around a few micrometres): for example a black body heated at 300 K has a peak emission at 10 ␮m. The material properties change with the type of radiation considered, i.e. the type of spectrum. 3.3.1. Long-wave radiative heat transfer Long-wave radiative heat transfer between surfaces (Qrad LW ) has been modelled assuming diffuse grey surfaces. The window is taken to be opaque to long-wave radiation. The non-linearity of radiative exchange implies the use of iterative techniques to solve the radiosity equations. Due to its robustness and its short execution time, the method proposed by Blomberg [31] has been chosen. This technique linearizes the radiosity equations around a reference temperature, and an additional term (called excess radiation temperature) is then used to obtain the exact solution. The use of this model implies several iterations before convergence is reached (assessed by a change in the surface temperature smaller than 10−10 %).

486

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

Table 5 Comparison of the temperatures obtained with the two models.

Commercial software Model developed

Operative temperature (◦ C)

Air temperature (◦ C)

Radiant temperature (◦ C)

21.84 21.92

20.41 20.50

23.26 23.34

Table 6 Definition of the variation range for the selected parameters (base case indicated by bold characters). Parameter

Variation range

Weather

Outdoor temperature (◦ C) Part of direct to total solar radiation (–)

18; 20; 23; 26; 30 0; 0.2; 0.4; 0.6; 0.8; 1

Ventilation

Air change rate (ACH) Air temperature gradient (K) Type of convective flow

0; 0.5; 1; 1.5; 2; 2.5; 3 0; 4 Constant coefficients, Alamdari & Hammond, Mixed Fisher, Mixed Fisher & Pedersen

Others

Emissivity of the internal surfaces (–) Absorptivity of the internal surfaces (–) Position of the person/sensor in the room

0.2; 0.5; 0.85

3.3.2. Short wave radiative heat transfer One of the main issues, when calculating the cooling need of a terminal, is the computation of the incoming solar radiation (Qrad SW ). There are three steps in the calculation of the solar load: - Step 1: Determination of the irradiation on the south fac¸ade. The irradiation on a vertical surface is calculated according to [32]. It takes into consideration direct, diffuse and reflected solar radiation, and also the shading due to the window opening. - Step 2: Separation of solar radiation (absorbed or reflected by the glazing, or transmitted to the room). The part of solar radiation absorbed by the glazing and transmitted to the room is calculated using the software Window 6 [33]. In this program, the detailed calculation of reflection, absorption and transmission of each pane is performed hemispherically for diffuse radiation and in steps of 10◦ incidence angle for direct solar radiation. It is therefore possible to define the part of solar radiation directly transmitted (Tsolar ) and the part absorbed and then reemitted towards the room (SHGC–Tsolar ), with SHGC Solar Heat Gain Coefficient. In the studied case, around 5% of the solar radiation is absorbed and reemitted and 50–55% is transmitted to the room (these values vary with the portion of direct to total solar radiation). - Step 3: Calculation of the solar repartition in the room. The part of transmitted solar radiation is then distributed on the room surfaces. Two calculation methods are used depending on the type of solar radiation (direct or diffuse). For direct solar radiation, the method proposed in BESTEST [26] is used to calculate the distribution depending on the solar absorptivity of surfaces. This technique assumes that all incident solar radiation initially hits the floor and is then reflected over the other surfaces according to their view factors. For the considered case, the distribution is 63% at the floor, 10% at the ceiling, between 4 up to 9% at each wall, and 1% lost through the window. For diffuse radiation, a similar technique is used: the diffuse solar radiation initially hits the room surfaces according to the view factors from the window to the considered surfaces. The remaining amount of original solar radiation is then assumed to be absorbed by all surfaces in proportion to their area-absorptance products. For the considered case, the distribution is around 22% at the floor, ceiling, and side walls, 9% at the back wall and 3% at the front wall. 3.4. Validation of the model A commercial software [34] has been used to check the validity of the model developed. Due to restrictive access to some input parameters in the commercial software, it has been chosen not to model the window and only convective terminals have been evaluated. In order to emphasize differences, the outdoor temperature has been defined to 40 ◦ C. In both models, the cooling power has been set to - 58 W/m2 . A good agreement between the two models can be observed (Table 5). The temperature difference, which is lower than 0.1 K, can be due to different parameters such as the value of combined heat transfer coefficient on the external side, the model used for radiation, the type of surface meshing, the calculation of the operative temperature, etc.

0.2; 0.4; 0.6; 0.8 Sitting in the middle of the room + 7 other positions (Fig. 1)

4. Sensitivity analysis of the cooling need 4.1. Definition of the parameter variation Parameters such as the amount of solar radiation, the level of insulation or the internal heat loads are primary factors for determining the cooling need. The first stage of the building design consists in optimising these parameters (e.g. by changing the orientation of the building or the type of solar shading). Once the cooling load has been determined (around 40 W/m2 in this case), it is of interest to study the parameters influencing the cooling performance of the different terminals. A list of eight factors has been defined: outdoor temperature, part of direct to total solar radiation, air change rate, air temperature gradient, convective flow in the room, emissivity and absorptivity of the internal surfaces and the position of the person in the room. These parameters are related to the outdoor conditions, the type of ventilation system or the room properties. Using the Morris Elementary Effect Method [35], the influence of these different parameters has been analysed for the four terminals. The objective of this method is to determine, which factors may be considered to have effects, which factors are negligible, linear and additive, or non-linear or involved in interactions with other parameters. It is a one-at-a-time method, in which the impact of changing the value of each of the chosen factor is evaluated in turn [36]. This method combines therefore the advantages of local and global methods, but it does not quantify the difference between the factors. The variation range of the different parameters is presented in Table 6. Discrete distributions with equal weighting are defined for all the parameters. Simlab [36], software developed for uncertainty and sensitivity analysis, is used to generate 90 Morris samples with 8 levels of executions. 4.2. Results The results of the sensitivity analysis for the four terminals are presented in Fig. 3. The air change rate and the outdoor temperature are the parameters that have the highest effect on the cooling need. Moreover, both of them have a high standard deviation, which indicates that these factors interact with other factors or have a non-linear behaviour. The air temperature stratification has also a relatively high influence on the cooling need. The absorptivity and part of direct to total solar radiation influence slightly the cooling need. For some parameters, the influence is limited to some terminals. For example, the correlation used for describing convection is the

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

487

Fig. 3. Estimated mean () and standard deviation () of the elementary effects of different parameters on the cooling need of terminals.

fourth most influencing parameter for the active chilled beam and the radiant floor. The influence of the position of the sensor can only be observed for the radiant wall. In the following part, the analysis will focus on the parameters, which influence the most the cooling need. 5. Detailed analysis of the influencing parameters 5.1. Influence of the air change rate Using the parameters of the base case (Table 6), the influence of the air change rate has been investigated (Fig. 4). Under these conditions, the ventilation assists the cooling process due to the relatively low outdoor temperature (23 ◦ C). At low air change rates, the active

chilled beam is slightly more effective, but the differences are rather small. The advantage of radiant terminals can be observed for high air change rates because the heat can be removed directly from surfaces to surfaces, without using the air as a transport medium. It results in a higher room air temperature, thereby maximising the ventilation losses. The radiant floor is the most efficient system despite the low convective heat transfer coefficient. This is due to the large view factor between the person sitting and the floor (the view factors to the floor, wall and ceiling are respectively 0.35, 0.20 and 0.12). If the person is standing, the performance of the radiant floor becomes closer to other radiant systems (and the view factors become respectively 0.22, 0.24 and 0.15). Radiant ceiling is the least efficient radiant terminal despite the high convective heat transfer coefficient: it can be explained by the low amount of solar radiation reflected on this surface and by the location of the operative temperature sensor. Differences have also been observed in the terminal heat balance. The convective part varies greatly according to the type of terminal: for floor cooling, it represents only 7% of the heat exchange, whereas it represents around 30–40% for the other radiant terminals. Similar result has been obtained experimentally by Novoselac et al. [27] in their study about cooled ceiling.

5.2. Influence of the outdoor temperature

Fig. 4. Comparison of the cooling need to maintain 26 ◦ C as a function of the air change rate and terminal type.

Simulations with different outdoor temperatures have been performed using the parameters of the base case. The energy needed to cool down a building is highly dependent on the ambient conditions (Fig. 5). If associated with a proper control strategy, the outdoor air

488

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

downward convective flow from the panels tends to homogenise the air temperature. For the chilled beam, the recirculation flow mixes the air in the room. 5.4. Influence of the type of convective flow

Fig. 5. Comparison of the cooling need to maintain 26 ◦ C as a function of the outdoor temperature and terminal type.

Fig. 6. Influence of a temperature gradient on the cooling need of a radiant wall.

can be used as a source of free cooling. Nevertheless, this characteristic is more profitable for radiant terminals than for convective terminals, as colder air is needed for the active chilled beam. 5.3. Influence of the temperature stratification Fig. 6 presents the effect of an air temperature gradient of 1.5 K/m on the cooling need of a radiant wall. Due to the warmer outlet air, the cooling need to achieve thermal comfort is reduced: at 3 ACH, the required power decreases by 30%. Similar results have been obtained with the other types of terminal. However, in practise, such a temperature gradient can probably be achieved only with the radiant wall and floor. With the floor cooling, temperature gradients up to 5 K have been observed experimentally by Tomasi et al. [37] under similar boundary conditions (i.e. floor height and cooling effect). When floor cooling is associated with displacement ventilation, the temperature gradient can even rise up to 7.5 K [38]. The air temperature gradient will be much smaller with a radiant ceiling or with the active chilled beam. With a radiant ceiling, the

In order to evaluate the influence of the type of convective flow, the cooling need has been calculated with the four models presented in Section 3.2, using the parameters of the base case. All the algorithms used in this paper are experimentally derived and can be used in building simulations, but their ranges of application are limited to specific cases. This means that the convective algorithm has to be chosen according to the design and location of the ventilation system. If the air velocity close to the surfaces is relatively low, correlations of natural convection should be used (model 1 or 2). If the building is ventilated mechanicaly and the Richardson number is lower than one [39], model 3 should be used in case of wall diffuser and model 4 in case of radial ceiling diffuser. The active chilled beam and the radiant floor are affected by the type of convective flow because of the relatively large temperature difference between the air and the surfaces (Fig. 10); therefore only these results are presented in Fig. 7. At low air change rates, the type of convection has a minor impact on the calculation of the cooling need because natural convection is the dominating effect. Similar results have been obtained by De Carli et al. [40], in their numerical study about the cooled floor. When the air change rate is higher than 1 ACH, the efficiency of the different terminals will vary depending on the design of the ventilation system: the difference is up to 12.5% at 3 ACH. Due to the low inlet air temperature, convective terminals are more dependant on the surface convection than radiant terminals. The efficiency of convective terminals is improved when forced convection is involved: due to the cold air blown close to the ceiling, the ceiling temperature decreases and acts as a radiant ceiling. Therefore, an efficient air conditionning system would be a terminal, which has characteristics similar to a radiant terminal. Because of its low natural convective heat transfer coefficient, cooled floor is also affected by the type of convection in the room. The stable convective flow results in a high ceiling temperature. When cold air is supplied close to the ceiling, its temperature decreases, resulting in a lower radiant temperature and a lower cooling need. 5.5. Influence of the level of insulation Even though the influence of the level of insulation was not included in the sensitivity analysis, this parameter needs to be assessed in order to identify, to which extend these results can be applied. Therefore, two additional simulations have been performed to assess the effect of varying the insulation level either

Fig. 7. Influence of the model for convection on the cooling need of the active chilled beam (on the left) and the radiant floor (on the right).

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

Fig. 8. Comparison of the cooling need to maintain 26 ◦ C as a function of the air change rate and terminal type (R-value of all surfaces decreased by a factor two).

Fig. 9. Predicted percentage dissatisfied in the plane z = 0.6 m with a radiant wall (30 ◦ C outdoor). (For interpretation of the color information in this figure legend, the reader is referred to the web version of the article.)

at the fac¸ade or for the entire room. From the first set of simulations, it has been observed that changing the level of insulation at the fac¸ade does affect the relative difference between radiant and convective terminals. In the second set of simulations, the insulation level of all construction parts has been decreased (Fig. 8). By comparing with Fig. 4, a large effect can be observed on the relative difference between terminals. When decreasing the thermal resistance of the constructions, the back losses of radiant terminals increase and air-based terminals become more advantageous.

489

Fig. 10. Difference between air and radiant temperature as a function of the air change rate and terminal type (30 ◦ C outdoor).

therefore slightly exceeding the range of category II (normal level of expectation). Similar studies have been performed with the other terminals in order to compare the uniformity of the comfort in the space. The PPD for both sitting and standing persons has been calculated in the occupied zone and the average, minimum and maximum values and standard deviations are presented in Table 7. The active chilled beam achieves theoretically the most uniform environment. Nevertheless, this evaluation does not account for the possible high air velocity in the room (over 0.08 m/s), which can be both a drawback (risk of draught) and an advantage (compensate for high operative temperature). The cooled ceiling is the radiant system, which achieves the most uniform environment. On the contrary, radiant floor has the largest mean PPD. One reason is the relatively large variation of comfort between the sitting position and the standing position as mentioned in Section 5.1. The other reason is the air temperature, which is higher than the one obtained with other systems. For the same operative temperature and moisture content, the PPD increases with the air temperature because of differences in the evaluation of convective and radiative heat losses from the human body [41]. In addition to the uniformity of comfort, Table 7 and Fig. 9 show the importance of the location of the control sensor, especially for the radiant floor and radiant wall. This sensor should be placed close to the occupied zone and far from warm or cold surfaces (Fig. 9). 6.2. Air temperature

6. Level of comfort achieved with the different terminals All combinations presented in Table 6 achieve an acceptable indoor climate, but local comfort conditions differ with the type of terminal. In order to account for warm conditions, the level of comfort has been evaluated using the parameters of the base case, but with a higher outdoor temperature (30 ◦ C). 6.1. Uniformity of the comfort level in the room In the sensitivity analysis (Fig. 3), it has been observed that the position of the sensor influences the cooling need of radiant walls, but this analysis did not give the sensitivity on thermal comfort. Therefore, the operative temperature has been set to 26 ◦ C in the middle of the room (at 0.6 m high) and the predicted percentage of dissatisfied (PPD) has been calculated over the working plane (Fig. 9) using the following assumptions: office worker (0.5 clo, 1.2 met), air velocity below 0.08 m/s and moisture content of 12.8 g/kg (corresponding to 60% relative humidity at 26 ◦ C) [23]. The occupied zone is defined as the area 0.6 m away from the walls. Despite the low temperature of the cooled wall (20 ◦ C), no discomfort can be observed close to the activated surface (PPD around 8.5%). Further from the radiant wall, the PPD increases up to 12.5% and is

When comparing the air and the radiant temperature, it can be observed that radiant terminals have a smaller asymmetry (Fig. 10). The difference is especially small for the cooled ceiling and wall, due to the relatively large convective exchange at these surfaces. The higher air temperature of radiant cooling systems can be a disadvantage, as the perceived air quality decreases with increasing enthalpy [42]. Nevertheless, this effect is relatively small: at constant air moisture content, the percentage dissatisfied increases by a maximum of 2 p.p. This asymmetry between air and radiant temperature can also affect the control of terminals. In the case of a radiant floor or an airbased cooling system, it can lead to a poor control if the operative temperature is not correctly measured. 6.3. Surface temperatures The risk of discomfort or condensation due to cold surfaces has been evaluated for different cooling power (Fig. 11). The minimum surface temperature is always above the minimum values stated in Table 3, except for the radiant floor, where the limit is reached at a cooling capacity of 50 W/m2 . Similar value has been obtained by Olesen [4].

490

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491

Table 7 Comparison of the predicted percentage of dissatisfied over the occupied zone (persons sitting and standing, 30 ◦ C outdoor).

Active chilled beam Radiant floor Radiant wall Radiant ceiling

Minimum PPD (%)

Mean PPD (%)

Maximum PPD (%)

Standard deviation PPD (%)

9.24 10.46 7.30 9.54

9.65 12.64 10.06 10.23

9.93 15.96 12.63 11.30

0.17 1.55 1.26 0.47

Fig. 11. Comparison of the minimum surface temperature with different radiant terminals as a function of the cooling power.

Fig. 12. Percentage dissatisfied due to the down-draught from the cooled wall.

The temperature asymmetry has also been assessed and is within the recommended range. The maximum asymmetry has been obtained for the cooled floor, with a radiant asymmetry between the floor and the ceiling equal to 6.5 K. 6.4. Risk of down-draught with the radiant wall When using radiant walls to cool down a building, the cold natural convective flow from the wall can create discomfort due to the down-draught. Correlations between surface temperature and air velocity have been developed by Heiselberg [43] to estimate the risk of draught from cold vertical surfaces. The percentage dissatisfied (PD) increases with the cooling power from the active surface and reaches 15% for a cooling power of 70 W/m2 (Fig. 12): this value is below 20%, the limit of category II [23]. The risk of down-draught from the radiant wall is therefore not the factor limiting the cooling capacity of radiant walls. 7. Conclusion Different cooling strategies have been applied in a typical office room, in order to evaluate the influence of the terminal type on the cooling power needed to achieve a fixed operative temperature. The thermal performances of four terminals (active chilled beam, radiant floor, radiant wall and radiant ceiling) have been compared

in term of delivered energy for different boundary conditions. The most influencing parameters have been identified by performing a sensitivity analysis. The objective of this analysis was to identify the advantages and drawbacks of the different technologies and also find out the parameters that are important to include in the evaluation of terminals. It has been observed that the interaction between the ventilation system and the terminal (e.g. air change rate, outdoor temperature, air temperature stratification) plays an important role in the room heat balance. At low air change rates (lower than 0.5 ACH), the performances of the different terminals are similar. But differences between the terminals can be observed at higher air change rates. The air temperature is warmer with radiant cooling terminals, resulting in higher air temperature, thus increasing the ventilation losses and decreasing the cooling need. The higher the air change rate and the warmer the outdoor air, the larger the savings achieved with a radiant cooling terminal. It has also been observed that the effectiveness of the active chilled beam depends on the type of flow in the room, i.e. on the design of the cooling system. Finally, the positive effect of a vertical air temperature gradient has been observed. At 1 ACH, the cooling need of the terminal decreases by 10% if a temperature gradient of 1.5 K/m is achieved. Such a temperature gradient can be achieved by a cooled floor or wall or by using displacement ventilation. Among radiant cooling terminals, the cooled floor has the lowest cooling need due to the large view factor between a sitting person and the activated surface. When considering people standing, the three terminals have a similar effectiveness for the considered geometry. These conclusions are valid for well-insulated buildings (R > 5 m2 K/W) and also for multi-storey buildings. For single-storey buildings with a low level of insulation, air-based terminals might be more energy-efficient than radiant terminals due to the lower conduction losses. Additionally, the variation of comfort with the different terminals has been evaluated. The radiant ceiling achieves the most uniform comfort conditions in the space, whereas the least uniform conditions have been obtained with the cooled floor. The active chilled beam achieves uniform comfort conditions in theory, but specific studies (e.g. CFD simulations) are needed to validate these results and account for the draught risk. Finally, it has been observed that the cooling power from a radiant wall is not limited by the risk of down-draught from the cold surface.

Acknowledgments This work is collectively sponsored by the Danish Agency for Science, Technology and Innovation and the Ministry of Science and Technology of P.R.China in the Sino-Danish collaborated research project: “Activating the Building Construction for Building Environment Control” (Danish International DSF project n◦ . 09-71598, Chinese international collaboration project n◦ . 2010DFA62410).

References [1] EN ISO 7726, Ergonomics of the thermal environment - Instruments for measuring physical quantities, CEN Brussels, 2001.

J. Le Dréau, P. Heiselberg / Energy and Buildings 82 (2014) 482–491 [2] J. Adnot, P. Waide, EECCAC (Energy Efficiency and Certification of Central Air Conditioners) Project for the Directorate General Transportation-Energy of the Commission of the European Union Final Report, 2003. [3] L. Pérez-Lombard, J. Ortiz, C. Pout, A review on buildings energy consumption information, Energy and Building 40 (2008) 394–398. [4] B.W. Olesen, Radiant heating and cooling by embedded water-based systems, in: Proceedings of Congreso Climaplus, Madrid, Spain, 2011. [5] G. Hauser, C. Kempkes, B.W. Olesen, D.F. Liedelt, Computer simulation of the performance of a hydronic heating and cooling system with pipes embedded into the concrete slab between each floor, ASHRAE Transactions 106–1 (2000) 702–710. [6] C. Roulet, J. Rossy, Y. Roulet, Using large radiant panels for indoor climate conditioning, Energy and Building 30 (1999) 121–126. [7] R. Bean, B.W. Olesen, K.W. Kim, History of radiant heating and cooling systems, ASHRAE Journal 52 (2010). [8] M. De Carli, A. Zarrella, R. Zecchin, Comparison between a radiant floor and a radiant wall on heating and cooling energy demand, ASHRAE Transactions 115-2 (2009) 563–572. [9] W. Kessling, S. Holst, M. Schuler, Innovative design concept for the new Bangkok international airport NBIA, in: Symposium on Improving Building Systems in Hot and Humid Climates, Richardson, USA, 2004. [10] M. Brunk, Cooling ceilings-an opportunity to reduce energy costs by way of radiant cooling, ASHRAE Transactions 99 (1993) 479-479. [11] J.L. Niu, L.Z. Zhang, H.G. Zuo, Energy savings potential of chilled-ceiling combined with desiccant cooling in hot and humid climates, Energy and Building 34 (2002) 487–495. [12] Z. Tian, J.A. Love, Application of radiant cooling in different climates: assessment of office buildings through simulation, in: Proceedings of the 11th International IBPSA Conference, Glasgow, Scotland, 2009, pp. 2220–2227. [13] G.P. Henze, C. Felsmann, D.E. Kalz, S. Herkel, Primary energy and comfort performance of ventilation assisted thermo-active building systems in continental climates, Energy and Building 40 (2008) 99–111. [14] J.-W. Jeong, S.A. Mumma, W.P. Bahnfleth, Energy conservation benefits of a dedicated outdoor air system with parallel sensible cooling by ceiling radiant panels, ASHRAE Transactions 109–2 (2003) 627–636. [15] T. Imanari, T. Omori, K. Bogaki, Thermal comfort and energy consumption of the radiant ceiling panel system: Comparison with the conventional all-air system, Energy and Building 30 (1999) 167–175. [16] E. Fabrizio, S.P. Corgnati, F. Causone, M. Filippi, Numerical comparison between energy and comfort performances of radiant heating and cooling systems versus air systems, HVAC&R Research 18–4 (2012). [17] J. Feng, S. Schiavon, F. Bauman, Cooling load differences between radiant and air systems, Energy and Building 65 (2013) 310–321. [18] S. Oxizidis, A.M. Papadopoulos, Performance of radiant cooling surfaces with respect to energy consumption and thermal comfort, Energy and Building 57 (2013) 199–209. [19] T. Kim, S. Kato, S. Murakami, J. Rho, Study on indoor thermal environment of office space controlled by cooling panel system using field measurement and the numerical simulation, Building and Environment 40 (2005) 301– 310. [20] E. Djunaedy, J. Hensen, Q. Chen, M. Loomans, Simulating radiative cooling/heating using BES-CFD coupled simulation, in: Proceedings Indoor Air, 2005. [21] L. Vandaele, P. Wouters, The PASSYS summary report, in: European Commission Publication, Publication No. EUR 15113 EN, 1994.

491

[22] EN 15265, Energy performance of buildings. Calculation of energy needs for space heating and cooling using dynamic methods. General criteria and validation procedures, CEN Brussels, 2007. [23] EN ISO 7730, Ergonomics of the thermal environment - Analytical determination and interpretation of thermal comfort using calculation of the PMV and PPD indices and local thermal comfort criteria, CEN Brussels, 2006. [24] M. La Gennusa, A. Nucara, G. Rizzo, G. Scaccianoce, The calculation of the mean radiant temperature of a subject exposed to the solar radiation - a generalised algorithm, Building and Environment 40 (2005) 367–375. [25] Standard ASHRAE 55. Thermal environmental conditions for human occupancy, 2010. [26] R. Judkof, J. Neymark, International Energy Agency Building Energy Simulation Test (BESTEST) and Diagnostic Method, National Renewable Energy Laboratory (NREL), 1995. [27] A. Novoselac, B.J. Burley, J. Srebric, New convection correlations for cooled ceiling panels in room with mixed and stratified airflow, HVAC&R Research 12 (2) (2006) 279–294. [28] F. Alamdari, G.P. Hammond, Improved data correlations for buoyancy-driven convection in rooms, Building Services Engineering Research and Technology 4 (1983) 106–112. [29] D.E. Fisher, An Experimental Investigation of Mixed Convection Heat Transfer in a Rectangular Enclosure, PhD Thesis, University of Illinois at UrbanaChampaign, USA, 1995. [30] D.E. Fisher, C. Pedersen, Convective heat transfer in building energy and thermal load calculations, ASHRAE Transactions 103 (1997) 137–148. [31] T. Blomberg, Heat Conduction in Two and Three Dimensions - Computer Modelling of Building Physics Applications, Report No. TVBH-IO08, Lund, Sweden, 1996. [32] ASHRAE, Handbook of Fundamentals, Chapter 14 - Climatic Design Information, Atlanta, GA, 2009. [33] University of California. Window 6.3, 2012. [34] Danish Building Research Institute. SBi-BSim, Version 6,11,1,14, 2012. [35] M.D. Morris, Factorial sampling plans for preliminary computational experiments, Technometrics 33 (2) (1991) 161–174. [36] N. Giglioli, A. Saltelli. SimLab 1.1, Software for Sensitivity and Uncertainty Analysis, tool for sound modelling, http://arxiv.org/ftp/cs/ papers/0011/0011031.pdf (2000). [37] R. Tomasi, M. Krajˇcík, A. Simone, B.W. Olesen, Experimental evaluation of air distribution in mechanically ventilated residential rooms: Thermal comfort and ventilation effectiveness, Energy and Building 60 (2013) 28–37. [38] F. Causone, F. Baldin, B.W. Olesen, S.P. Corgnati, Floor heating and cooling combined with displacement ventilation: Possibilities and limitations, Energy and Building 42 (2010) 2338–2352. [39] S. Leenknegt, R. Wagemakers, W. Bosschaerts, D. Saelens, Improving the modelling of surface convection during natural night ventilation in Building Energy Simulation models, in: Proceedings of the 12th International IBPSA Conference, Sydney, Australia, 2011, pp. 2233–2240. [40] M. De Carli, M. Scarpa, R. Tomasi, A. Zarrella, DIGITHON: A numerical model for the thermal balance of rooms equipped with radiant systems, Building and Environment 57 (2012) 126–144. [41] P.O. Fanger, Thermal Comfort: Analysis Applications of Environmental Engineering, Danish Technical Press, 1970, pp. 244. [42] P.O. Fanger, Human requirements in future air-conditioned environments, International Journal of Refrigeration 24 (2) (2001) 148–153. [43] P. Heiselberg, Draught risk from cold vertical surfaces, Building and Environment 29 (1994) 297–301.