Energy and Buildings 43 (2011) 853–863 Contents lists available at ScienceDirect Energy and Buildings journal homepage: www.elsevier.com/locate/enbuild Empirical validation and modelling of a naturally ventilated rainscreen façade building C. Marinosci a,∗ , P.A. Strachan b , G. Semprini a , G.L. Morini a a b DIENCA, Alma Mater Studiorum - Università di Bologna, Viale Risorgimento 2, 40136 Bologna, Italy Energy Systems Research Unit, University of Strathclyde, 75 Montrose Street, Glasgow, Scotland, UK a r t i c l e i n f o Article history: Received 15 September 2010 Received in revised form 7 October 2010 Accepted 7 December 2010 Keywords: Ventilated façade Rainscreen wall Modelling and simulation Test cell a b s t r a c t In this paper the thermal behaviour of a rainscreen ventilated façade has been investigated both experimentally and numerically. Field measurements were performed during the 2009/10 winter season in a test building located in San Mauro Pascoli (Italy) having a squared base of internal dimension of 2.89 m and a total internal height of 7.75 m. The external walls of this tower are rainscreen ventilated façades with a 24 cm air cavity and an external side composed of stoneware with open joints. Ventilation grills are located at the top and at the bottom of the tower. In this work the modelling of the test building using a dynamic thermal simulation program (ESP-r) is presented and the main results discussed. In order to study the rainscreen ventilated façade three different multi-zone models were defined and the comparison with the experimental results has been used in order to select the best ESP-r air flow network for the modelling of this kind of envelope component. The thermal analysis of this envelope component evidenced that the ventilated façade is able to reverse the direction of the heat flux through the envelope in regions characterized by large solar irradiation during the winter and moderate wind velocity, when the indoor–outdoor air temperature difference is small, thereby reducing the energy consumption required for indoor heating. © 2010 Elsevier B.V. All rights reserved. 1. Introduction The envelope of a building is the main element responsible for its energy demand. There are many types of envelope to improve thermal and energy performance of buildings: among the most complex and discussed solutions are ventilated structures. Ventilated façade design is not a new topic—the implementation of ventilated façades in buildings has been an object of broad application especially in recent years when the design of buildings having low energy consumption has become a priority. In addition, this kind of component attracts designers for aesthetic reasons, for good sound insulation and for improved indoor environment. In regions with high levels of solar radiation, ventilated structures maintain the temperature of the inner shell of single or double-skinned buildings at a temperature close to the ambient, significantly reducing the impact of incident radiation on the indoor environment. Ventilated walls, façades and roofs, if well designed, can help to reduce considerably the summer thermal loads due to direct solar radiation ∗ Corresponding author at: DIENCA, Facoltà di Ingegneria, Università di Bologna, Via Risorgimento 2, 40136 Bologna, Italy. Tel.: +39 051 2090549; fax: +39 051 2090544. E-mail address: cosimo.marinosci@unibo.it (C. Marinosci). 0378-7788/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.enbuild.2010.12.005 [1–4]. Moreover, ventilated structures can be extremely useful for the installation of photovoltaic panels in order to increase their cooling and, as a consequence, their efficiency [5,6]. In this paper the thermal behaviour of typical rainscreen ventilated façades is analysed. Typical rainscreen wall components consist of three main sublayers: the exterior layer (rainscreen), the ventilated cavity and the internal thermal insulated wall [7]. The exterior layer can be made of brick, precast-concrete, glass or sometimes aluminium. The main function of the rainscreen is to keep away most of the driving rain from the structural leaf. Using a pressure equalization mechanism, the rainwater penetration caused by wind-induced differential pressure can be also reduced. Further, it protects the internal layers from direct solar radiation, heavy wetting, sound, heat, pollutants, fire and other environmental factors. More details about the rainscreen principle can be found in [8]. However, the influence of this kind of component on the indoor environment and on energy consumption is very difficult to predict quantitatively. This is mainly due to the unsteady and complex air flows generated in the naturally ventilated skin façade cavities. In fact, the free convection flow rate depends on the temperature distribution, on the cavity geometry, on the fluid-dynamic head losses and on the external atmospheric conditions (especially wind velocity). In the rainscreen ventilated façade, from a fluid-dynamic 854 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 Nomenclature A area (m2 ) cp specific heat (J/(kg K)) Cd discharge coefficient DMIN minimum difference (K) DMAX maximum difference (K) hc convective heat transfer coefficient (W/(m2 K)) hr radiative heat transfer coefficient (W/(m2 K)) H height of a vertical surface (m) k thermal conductivity (W/(m K)) ṁ mass flow rate (kg/s) MEANDT average difference (K) Mt measured value at hour t M̄ mean values of the measured values n number of total hours in the period selected P pressure (Pa) qrad heat exchange with surrounding surfaces that are in longwave contact (W) qconv heat exchange between the room air and the solid surface (W) qint-zone thermal energy of air flowing into the control volume from other zones (W) qinf thermal energy of air flowing into the control volume from outdoors (W) qsolar solar radiation absorbed at node (W) qplant radiant plant input to node (W) qcasual radiant energy absorbed from casual sources (W) qI−1→I , qI→I+1 conductive heat flows across the faces of the control volume (W) RSQMEANDT root mean square difference (K) R2 Pearson coefficient t time (s) T temperature (K or ◦ C) Xt predicted value at hour t X̄ mean values of the predicted values Greek symbols  fluid density (kg/m3 ) x, y, z width, depth, and height of a control volume Subscripts i internal e external conv convection r radiation s surface plant plant interaction solar solar gain cas-rad radiant energy from casual sources cas-conv convective energy from casual sources int-zone internal zone I state at node under consideration I − 1, I + 1 state at neighbouring nodes J internal zone node inf infiltration Superscripts t value of a variable at the beginning of a simulation time-step t + t value of a variable at the end of a simulation timestep Fig. 1. (a) Test building and (b) lay-out of the rain-screen walls (dimensions in cm). point of view, the modelling can be complicated by the presence of several air by-passes due to specific opening joints. There are many published works on the thermal behaviour of ventilated façades, Trombe–Michel walls and solar chimneys. A large number of these works are concerned with the numerical simulation of these components by using computational fluid dynamics (CFD) and building energy simulation (BS). For example, Smolec and Thomas [9] determined the air temperature distribution through a Trombe wall by using a thermal model having a network structure in order to take into account all the existing heat fluxes which operate in this component and they compared the results with experimental data. Mootz and Bezian [10] performed a numerical study of a ventilated façade panel structured like a composite Trombe–Michel wall. Gan [11] carried out the numerical simulation of a Trombe wall for summer cooling by using the CFD technique and investigated the effect of the distance between the wall and glazing, wall height, glazing type, and wall insulation on the thermal performance of a Trombe wall. Economic evaluation of this kind of component using life cycle costing technique has been carried out by various researchers [12–15]. The analysis of the existing literature on the ventilated façade highlights that the rainscreen ventilated façade has received less attention than the other configurations. In order to fill this gap, in this paper a model for the thermal analysis of the rainscreen ventilated façade is presented. The results of the model are compared with a series of experimental data obtained during the winter in an Italian site. The comparison between the experimental data and the numerical results are used in order to select the best model for this kind of component and in order to demonstrate that it is possible to obtain detailed information on the thermal behaviour of the rainscreen ventilated façade by using BS tools such as ESP-r. 2. The test building The test building is located in Italy, San Mauro Pascoli—Forli Cesena (latitude 44.11◦ N, longitude 12.43◦ E). It was completed in November of 2009 with the purpose of detailed investigations of the rainscreen wall energy performance. The building has a squared base of internal dimension of 2.89 m and height of 2.67 m. There are three floors for a total internal height of 7.75 m. The door is on the north side and there are no windows (Fig. 1a). The structure is realized mainly with steel profiles and wood (floor). All external walls are three-layered with a middle layer composed of a ventilated 24 cm air cavity and an external layer composed of stone cladding (Fig. 1b). The stone cladding consists of porcelain stoneware tile systems with open joints that are separated from the inside wall by the ventilated air cavity. At the top and at the bottom of this 855 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 Table 1 Thermal properties of the layers of the external walls. Thickness (cm) Stoneware 1 Air cavity 24 Glass wool 6 Aluminium 0.2 EPS 4 Aluminium 0.2 Thermal conductivity (W/(m K)) Density (kg/m3 ) 1.30 – 0.04 160 0.04 160 2300 – 30 2800 70 2800 Specific heat (J/(kg K)) 840 – 1200 880 1200 880 Air cavity temperature (oC) Layer 30 25 20 15 10 5 0 0 tower are located the ventilation grills. The roof consists of a sandwich panel and is not ventilated. The ground floor is made of an upper layer of wood, a second layer comprising an unventilated cavity and a third layer of cement mortar in contact with soil. All the properties of the materials of the external wall of the test building are given in Table 1. A horizontal solar obstruction, about 70 cm over the external side, provides protect weatherproofing at the top opening grill. In order to investigate the winter thermal performance of the rainscreen wall component, a series of tests were performed, during the period from January 1st to March 31st 2010. The façade is exposed to real weather conditions, but only the South and West sides have been equipped with a series of sensors. The sensors comprise 72 thermocouples, 2 hotwire anemometers and one pyranometer. The thermocouples were positioned both inside and outside the test building. T-type thermocouples were used with a declared accuracy of ±0.5 C (class 1 according to EN 60584-2, range: −270÷400 ◦ C) using standard calibration. The accuracy of hotwire air speed transmitters was of ±0.04 m/s +2% of reading if the measurement ranged 0.05÷1 m/s. The pyranometer was a Second Class LP (PYRA 03) with a range of 0÷2000 W/m2 and an accuracy <|±2|%. Each wall is divided along the vertical direction of the cavity into three sections as shown in Fig. 2; sensors were located in each section along the cavity and outside. In more detail, the following measurement instruments were used during the experimental tests: - 7 temperature sensors fixed on the external side of the rainscreen. All the sensors were shielded with an aluminium cover from the 100 200 300 400 500 600 700 800 Solar radiation (W m-2) Fig. 3. Typical trend of the average value of the air cavity temperature as a function of the global solar vertical radiation on South side (period: March 19–25, 2010). direct solar radiation. - 17 temperature sensors to monitor the temperature distribution along the walls of the ventilated cavity. Thermocouples were fixed to the surface of the duct with silicone adhesive. - 17 temperature sensors used in order to determine the vertical distribution of the air temperature along the cavity. Sensors were positioned in the middle of the air duct. - 11 temperature sensors for the determination of the temperature on the external side of the thermal insulation (toward to the air cavity). - 17 temperature sensors on the inner surface of test building. - 3 temperature sensors for the determination of the indoor air temperature at the centre of each room. - 2 hotwire anemometers located at 7 m height on the South façade and on the West façade (in the middle of the ventilated air ducts). - 1 pyranometer (Second Class pyranometer according to ISO 9060) for the determination of the global solar radiation on the South wall. All sensors are connected to a dedicated Data Acquisition System (National Instruments). Acquisition and recording of the data were done by a computer station using Labview software. Data were recorded at intervals of 5 min and can be monitored remotely by using an Internet connection. Wind direction, wind speed, ambient temperature and air relative humidity were monitored every 5 min by a weather station installed close to the test rig. The test building was heated continuously to a room temperature of 20 ◦ C by means of a heat pump system. 3. Monitoring measurements Fig. 2. Three sections of the vertical ventilated duct. In this section typical results that were extracted from the set of measurements made on the test building are shown. The large number of temperature sensors enables a detailed study of the temperature distribution in the main regions of the building and its envelope, which can be used in order to calculate all the heat fluxes involved in the thermal balance of the building. In Fig. 3 the correlation between the vertical global solar radiation on the South side and the hourly average value of the air cavity temperature is shown. It can be noticed how the level of correlation between these parameters tends to improve for large values of the solar radiation, as also observed by Ong [16]. Conversely, for low values of solar radiation (<200 W/m2 ) the spread of the air cavity temperature is very large. In Fig. 4 the vertical gradient temperature along the air cavity of the South wall on a typical winter day (28th March 2010) as a function of the incident global solar radiation is depicted. It can be noticed that the typical vertical temperature distribution presents a maximum along the air cavity near the top of the channel. This 856 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 Fig. 5. Trend of the average values of the measured temperature and air cavity velocity of the South ventilated wall for different hours (14th February 2010). Global solar radiation and wind speed both are reported (prevailing wind direction W). Fig. 4. Vertical temperature gradients along the air cavity—South side. trend is shifted in the sense of the increasing temperature for larger values of the instantaneous global solar radiation. It is possible to see that when the vertical global solar radiation reaches its daily maximum values (733 W/m2 ) the air temperature is greater than 30 ◦ C when the external air temperature is equal to 19 ◦ C. The maximum value of the air temperature tends to decrease when the solar radiation decreases. The decrease of the temperature near the exit of the vertical cavity is due to the presence of the horizontal solar obstruction which shades the external layer close to the ventilation grille. In Figs. 5 and 6 the trend of the hourly values of the temperature measured in the wall are shown. These data show the variation of the outside wall temperature, of the air cavity temperature and of the inside wall temperatures during the central daily hours of two different days of the winter period characterized by a large and a low air temperature difference between the outdoor and indoor (18K in Fig. 5 and 2K in Fig. 6). By analyzing the data shown in Figs. 5 and 6 it is evident that during the winter the role of the outside layer is to absorb the solar radiation in order to increase the wall temperature of the cavity: this value influences the mean air cavity velocity due to the buoyancy effect. Of course, the air cavity velocity is related also to the wind velocity so the final value of the air flow rate through the cavity is obtained as the combined effect of the wind and the solar radiation which changes the wall temperature of the cavity. For this reason, in Figs. 5 and 6 the measured value of the average air cavity velocity is given together with the global solar radiation incident on the external surface of the rainscreen. It is evident in Fig. 5 that, even for low external air temperature (Text = 5 ◦ C), the outside wall of the cavity can reach a temperature near to 20 ◦ C for solar radiation values larger than 450 W/m2 . Greater temperature of the outside wall means greater temperature values on the inside cavity walls which implies a reduction of the heat flux from the room to the external environment. Moreover the temperature gradient inside the wall of the ventilated façade is progressively reduced when the incident solar radiation on the external surface of the rainscreen increases which results in a significant reduction of the heat losses through this component. For the same reason, during the winter season high values of the cavity air flow rate are not desirable, because the convective heat transfer between the air and the inside wall reduces the temperature on Fig. 6. Trend of the average values of the measured temperature and of the air cavity velocity of the South ventilated wall for different hours (28th March 2010). Global solar radiation and wind speed both are reported (prevailing wind direction SW). 857 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 the inside wall of the cavity and increases the heat flux through the inside wall. It is interesting to compare the temperature profile shown in Fig. 5 for a typical winter day with a low external air temperature with the profile depicted in Fig. 6 obtained for a day characterized by a reduced temperature difference between outside and inside and high solar radiation (a typical situation which can occur at the beginning and at end of the winter season). The most important difference between these two situations is related to the direction of the heat flux through the inside wall of the ventilated façade. In fact, Fig. 6 represents a typical situation in which the ventilated façade is able to reverse the direction of the heat flux through the wall and in which the façade become an active component for the heating of the indoor environment. For this reason, especially in regions characterized by large solar irradiation during the winter and moderate wind velocity, this kind of component can give a contribution to the reduction of the energy consumption related to the indoor heating during the winter. The data of Figs. 5 and 6 show that, since the direction and magnitude of the heat fluxes through the walls is an unsteady phenomena linked to the instantaneous values of the external air temperature, of the wind velocity and of the incident global solar radiation, a dynamic simulation of these components is necessary in order to obtain an “a-priori” evaluation of the energy saving due to the use of the ventilated façade. In the next section the problems related to the modelling of this kind of component are discussed. 4. Modelling setup A model of the test building was created by using the simulation program ESP-r, which is a transient simulation program based on the finite volume technique [17]. By using ESP-r, it is possible to model all the energy fluxes and the fluid flows within combined building and plant systems subject to dynamically varying boundary conditions. In ESP-r the building is discretized by using a finite number of nodes representing air volumes (such as rooms), opaque and transparent components (walls, windows, roofs, floors), solid–fluid interfaces (such as the internal and external surfaces of walls and windows), and plant components (such as boilers and heat exchangers and so on). Each node has an associated value of temperature and pressure. By knowing the values of the temperature and pressure linked to each node, the main mass fluxes and heat fluxes between two different nodes can be calculated by using the following general relationships: qrad,s→I = N  htr,s→I y z(Tst − TIt ) (1) s=1 t − TIt ) qconv,I = htc,I y z(TI+1 (2) M qint-zone,J→I =  mtJ→I cp (TJt − TIt ) (3) J=1 qinf,I = mte→I cp (Tet − TIt ) qI−1→I (4) kI−1 y z t ≈ (TI−1 − TIt ) xI−1 (5) kI+1 y z t t ) (TI − TI+1 xI+1 (6) qI→I+1 ≈ where the values of the heat fluxes (q) are a function of the temperature existing at the nodes (T), the thermal conductivity of the layers (k), the convective and radiative heat transfer coefficients (hc , hr ) and the thermophysical properties of the materials involved (cp , ). 10 qint −zone, 4→10 qsolar ,1 q plant + qcasual qrad ,5→3 qrad , s →1 e 1 2 3 qconv,1 5 4 qconv,3 qrad , s →9 qconv,5 qrad ,3→5 6 7 8 9 i qconv ,9 qinf, 4→e qint −zone, 4→11 11 Fig. 7. Heat fluxes for various kinds of nodes: nodes in homogeneous material layers (dashed circles), nodes at internal surfaces (filled circles) and nodes for air points (empty circles). The air mass flux between two nodes (m) can be calculated by using the difference of pressure existing between the nodes according to the following equation:  ṁJ→I = Cd A 2 PJ,I (7) in which Cd is the discharge coefficient of the opening and A is the total opening area. By using these equations is possible to write the heat and mass balance equations for the nodes located within homogeneous material layers (dashed circles in Fig. 7) and/or for internal surface nodes (filled circles in Fig. 7) and/or for the room air-point nodes (empty circles in Fig. 7): this procedure enables a nodal network model for a whole building or for some specific components of the envelope to be built. The resolution of the system of equations generated by the systematic application of the mass and energy balance at each node is rather complicated. More detailed information about the solution methods and the application of this type of tool can be found in Clarke [17]. Although this procedure can be useful in many cases, it is sometimes necessary to couple BS solvers with a CFD analysis, for example when flows in the airspace are complex [18]. In this work, the rainscreen ventilated façade of the test building was studied numerically using a specific nodal network model; an airflow network was integrated with the corresponding thermal network model so that the calculated airflows are based on nodal temperatures from the thermal model, and the resulting predicted airflows are used in the energy balances of the thermal simulation with an iterative procedure of solution. In order to model the whole test building, the rainscreen wall is divided into a stack of three zones adjacent to each of the floor levels. Three zones are divided by fictitious transparent surfaces with high conductivity, negligible thermal mass and high emissivity, and coupled by an airflow network which also includes the inlet opening at the bottom and the top outlet of the façade. As reported in [19] the number of nodes (zones) linked by an air flow network to the thermal network is of fundamental importance in order to take into account in the model the natural convection effects in a building. However, no rules have been developed up to now in order to know the number of nodes needed for an optimal simulation of these cases. In this specific case, the airflow and the thermal number of zones have been chosen according to the position of temperature sensors used in the experimental analysis of the test building. It has been verified that the numerical results are particularly sensitive to the number of zones used to model the airgap. 858 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 Table 2 Results for air cavity temperature of the West façade. Airflow network model DMAX (K) DMIN (K) MEANDT (K) RSQMEANDT (K) R2 Model A Model B Model C 18.09 5.94 5.13 −8.37 −5.66 −5.87 −0.25 0.02 0.04 2.50 1.06 1.03 0.8345 0.9680 0.9711 Since the air cavity is being treated as a set of thermal zones, the heat transfer coefficients used by ESP-r for the inner surface of the air channel are the same as for room inner surfaces. These coefficients have been computed by using the Alamdari–Hammond correlations [20] as a function of the temperature difference between the surface and the room air (T) and the surface height (H): hc,I = [(1.5(T /H) 0.25 6 ) (1.23(T /H) 1/6 0.33 6 ) ] (8) According to Poirazis [21] the air flow in the cavity is dependent on the wind pressure conditions on the building’s skin, on the stack effect and on the discharge coefficient (Cd ) of the openings. In the numerical simulation, it was verified that the results obtained by means of the airflow network model are very sensitive to the discharge coefficient Cd . In this analysis, the grilles at the bottom and the top of the ventilated façade were modelled as a standard orifice airflow component with a discharge coefficient equal to 0.65 and an opening area (A) equal to 50% of the geometric area. Also the open joints in the façade were modelled as standard orifice airflow components with a discharge coefficient equal to 0.65 and opening area equal to the geometric area. Some researchers have proposed alternative models in order to take into account the air exchange through the open joints [22,23] but during the set-up of the model of the rainscreen it was shown that the final results were weakly influenced by the model adopted for the open joints. In the thermal model of the test building a value of 0.2 for ground reflectivity was adopted. Thermal bridges of the test building were not included in the model as they were considered negligible. The cooling/heating system was introduced into the model by imposing a constant value of the room temperature. 5. Model validation First of all, a study was made to define the best airflow network model to adopt for the simulation of a rainscreen ventilated façade using ESP-r. In the definition of the appropriate airflow network model for the tested façade, it is necessary to determine if the joints and grills of the ventilated façade are important for the airflow model or if the role of these components can be considered negligible in the estimation of the thermal performance of the façade. In order to give an answer to this open question the tested façade was modelled in ESP-r by means of three different airflow network models: • The flow network considers joints and grills at the bottom and the top as closed (in the cavity there is only thermal buoyancy between the cavity and internal room and wind effects are not included in the analysis). This model is called model A. • The flow network considers all the joints closed; the grills at the bottom and the top are considered open. This model is called model B. • The flow network takes into account the real conditions of the ventilated façade: joints and grills are completely opened. This model is called model C. Furthermore, the comparison between results obtained by using models A and C can be useful in order to highlight the significance of the wind effect and the buoyancy driven flow on the thermal performance of the ventilated façade during the winter season. In fact, it is well known that forced circulation of the air in the cavity during the winter can reduce significantly the thermal insulation of the façade with negative effects on the building energy demand. To establish what is the best airflow network model for this kind of ventilated façade the measured values (M) and the ESP-r predicted ones (X) are compared by using as input for ESP-r the climate data recorded during the test period. For the purpose of the model validation, measured local weather data for three months (1st January–31st March 2010) were used. The measured local weather data include external air temperature and air humidity, wind speed and direction and total vertical (South side) global solar radiation. The external air temperature varied between −6 and 25 ◦ C with a high relative humidity (typically higher than 70%). The wind speed during the whole period was less than 5 m/s with a predominant direction between North and East (January–February) and West and North (February–March). By using these data as input data in the ESP-r software and solving the balance equations recalled above (Eqs. (1)–(7)) of the adopted airflow network model, the hourly average values of the temperature linked to the nodes of the network can be calculated. Assuming Xt is the predicted value at hour t from ESP-r and Mt the corresponding measured value at the same instant of time, it is possible to analyse the capability of the adopted model to predict the measured values by using the following statistical parameters: DMIN = min(Xt − Mt ) (9) DMAX = max(Xt − Mt ) (10) n MEANDT =  (Xt − Mt ) t=1 (11) n   n  (Xt − Mt )2 RSQMEANDT =  t=1 (12) n where n is the number of total hours in the period selected for the comparison. Tables 2 and 3 show the results for the air cavity temperature of the West and South façades by using different airflow network models for the ventilated façade (models A–C). It is interesting to Table 3 Results for air cavity temperature of the South façade. Airflow network model DMAX (K) DMIN (K) MEANDT (K) RSQMEANDT (K) R2 Model A Model B Model C 24.51 5.25 4.99 −11.88 −8.53 −7.85 −0.46 −0.42 −0.53 2.79 1.58 1.47 0.9103 0.9762 0.9795 859 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 R2 = ⎝ ⎞2 (X − X̄)(Mt − M̄) t=1,n t t=1,n (Xt − X̄) 2 t=1,n (Mt − M̄) 2 ⎠ (13) in which X̄ and M̄ are the mean values of the predicted (Xt ) and the measured (Mt ) values. If the model A is adopted for the ventilated wall the numerical results exhibit a large disagreement with the experimental results; the R2 coefficient of correlation quoted in Tables 2 and 3 between the measured values and the predicted ones of the air cavity temperature is less than 0.84 for the West-facing wall and 0.91 for the South façade. These values demonstrate that, when the model A is adopted, the agreement with the experimental data is strongly influenced by the orientation of the considered ventilated wall and increases for walls exposed to large values of the solar radiation. In fact, in this case the air circulation within the cavity is due to the temperature difference between the cavity walls and the air only and this difference increases with the incident solar radiation. It is interesting to observe that R2 increases by 16% between model A and the model C which reveals the best accuracy of this last approach. On the contrary, the variation of R2 between model B and model C is modest. One can conclude that during the simulation of the rainscreen ventilated façade the presence of the open joints can be ignored in the modelling without a significant decrease in the accuracy of the numerical results. Also, by ignoring the open joints, the airflow network model becomes simpler and the numerical solution of the balance equations faster. Results of Tables 2 and 3 confirm that the selection of the airflow network model is very important in order to predict the thermal performances of a ventilated façade, as it has been noted by many workers [24–27] in the past. On the basis of these results model C has been selected as the best method to model the tested rainscreen ventilated façade. For the West-facing façades, the cavity temperature monitored from January 1st to March 31st was compared with the values generated by ESP-r using the models A–C. The difference in the predictive capability of the models A and C can be observed by comparing the data of Figs. 8 (model A) and 9 (model C). For model A the deviation from the bisector (perfect solution) is larger than 50% while for the model C it is near to 15%. These data confirm the values of the R2 parameter given in Table 2. The good prediction capability of model C is confirmed also by the comparison between the experimental and the numerical data obtained by ESP-r for the South-facing façade: in fact the R-squared coefficients go from a value of 0.9711 for the West-facing façade to the value of 0.9795 for the South-facing façade. As underlined before, an interesting result of this comparison is that model B and model C have similar predictive capability for the air cavity temperature; in fact, if the value of the R2 param- 25 Air cavity temperature predicted (oC) ⎛ 30 20 15 10 5 0 -5 -5 0 5 10 15 20 25 30 Air cavity temperature measured (oC) Fig. 8. Correlation between measured and predicted cavity temperatures on the West side for three months for model A. eter linked to the West and the South-facing façade obtained by adopting the models C and B are compared, it was shown that this parameter increases only by 0.32–0.33% going from model B to model C. This result highlights that the adoption of a simpler air flow network in which all the joints of the rainscreen cover are considered closed can guarantee accurate results with reduced CPU time and therefore, especially for long dynamic simulations, model B is preferred to model C. 30 25 Air cavity temperature predicted (oC) observe that the difference between the predicted and the measured values of the air cavity along the West and the South façades change significantly with the model adopted. In particular, model A in which the effects of the wind through the grills and the joints are completely ignored tends to overestimate the temperature in the cavity with respect the measured values. The difference is very large especially in the South façade; this fact confirms the role of the forced air circulation due to the wind during the winter on the thermal conditions of the façade. On the contrary, it is interesting to observe that differences between the results obtained by using models B and C are quite limited. This fact indicates that the role of the open joints can be considered negligible in the tested conditions. In order to quantify the different performances of the models the Pearson coefficient (R-squared) can be calculated by using its classical definition: 20 15 10 5 0 -5 -5 0 5 10 15 20 25 30 Air cavity temperature measured (oC) Fig. 9. Correlation between measured and predicted cavity temperatures on the West side for three months for model C. 860 a 35 External surface temperature (oC) C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 30 Measured Predicted 25 20 15 10 5 0 19/02 19/02 20/02 20/02 21/02 21/02 22/02 22/02 23/02 23/02 24/02 24/02 25/02 25/02 b 1000 16 Ambient temperature Ambient temperature (oC) 14 Solar radiation 12 10 800 600 8 400 6 4 200 2 0 0 Vertical global solar radiation (W m -2) Time 19/02 19/02 20/02 20/02 21/02 21/02 22/02 22/02 23/02 23/02 24/02 24/02 25/02 25/02 Time Fig. 10. (a) Comparison of measured and predicted results of the external surface temperature on South side. (b) Vertical global solar radiation during the tests. 6. Comparison with monitored data The previous analysis showed that model B gave the best compromise between accuracy and model complexity for general modelling purposes. However, for the detailed comparisons between measured and predicted results presented in this section, model C has been used. Thus the adopted air flow network takes into account the real conditions of the ventilated façade with joints and grills opened. The numerical results obtained by means of the ESP-r model of the rainscreen ventilated façade in terms of external surface temperature and air cavity velocity have been compared with the measured corresponding quantities during the whole test period. As stated before, three parameters are primarily responsible for the thermal performance of the ventilated façade: (i) the solar radiation on the external surface, (ii) the external air temperature and (iii) the wind velocity and direction. The comparisons that are presented in this section aim to demonstrate that the model of the ventilated façade is able to take into account the combined effects of these quantities. Fig. 10a shows the comparison between the measured and predicted values of the external rainscreen surface temperature during the week from 19 to 25 February on the South side of the façade. In Fig. 10b the measured values of vertical global solar radiation on the South side and external air temperature for the same period are shown. As evidenced by Fig. 10b, the selected period was characterized by a large variation of the measured values of solar radiation ranging from 27 to 868 W/m2 . It was selected in order to test the capability of the model to predict the thermal behaviour of the ventilated façade exposed to highly variable incident solar radiation. The simulated rainscreen surface temperature agrees very well with measured data and this fact confirms that the model is able to correctly take into account the role played by the solar radiation and by the external air temperature. During this period the wind velocity was variable with high value about 5 m/s (19/02) and less than 3–4 m/s on other days, with a predominant direction South-West. However, the data of Fig. 10a show that the model underestimates the external rainscreen surface temperature when solar radiation is absent. A possible reason for this is the uncertainty associated with estimating long-wave radiation transfer to the sky at night. The comparison between the numerical and the experimental data for the whole test period (January–March) showed that the role played by the wind velocity on the air cavity velocity was generally the most difficult effect to predict. Fig. 11a shows the comparison between the experimental and the numerical hourly average data of the air cavity velocity during three days characterized by low external air temperatures (−0.6 ◦ C), low solar radiation (40 W/m2 ) and wind velocities between 0 and 1.5 m/s for the West façade with a variable direction between West and East (Fig. 11b). It is interesting to observe that typical air cavity velocities have a low value during this period (less than 0.12 m/s) and this fact is beneficial in order to reduce the heat losses through the façade during the winter. The low values of the velocity, confirmed also by the data quoted in Figs. 5 and 6, are generally due to the large thickness of 861 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 a 0.3 Air cavity velocity (m s-1) Measured Predicted 0.2 0.1 0.0 03/01-20.30 04/01-2.30 04/01-8.30 04/01-14.30 04/01-20.30 05/01-2.30 Time 1.5 NW Air velocity (m s-1) 1.3 W 1.0 SW S 0.8 SE 0.5 Wind speed 0.3 Wind direction NE N 0.0 03/01-20.30 E Wind direction b 04/01-2.30 04/01-8.30 04/01-14.30 04/01-20.30 05/01-2.30 Time Fig. 11. (a) Comparison of measured and predicted results of the air cavity velocity—West façade. (b) Wind speed and direction during the tests. the air cavity adopted in this ventilated façade (24 cm). The data of Fig. 11a show that during a night time without wind the measured velocity tended to be strongly variable, due to the thermal buoyancy, although measured and predicted results are in agreement in their order of magnitude. During the day, the model and the experimental data showed some small discrepancies. It is important to note that comparisons are difficult at low air velocities, particularly in view of the large uncertainties of the hot wire anemometer measurements. However, the level of agreement was considered acceptable. Finally, a test was conducted of the capability of the model of the whole building to predict the trend of the indoor air temperature during a period in January (two weeks) in which the heating plant of the tower was turned off. Fig. 12 shows the trend of the experimental values of the indoor air temperature of the test building compared to the numerical data obtained by using ESP-r. Some discrepancies between the measured and the numerical data can be seen: in particular the ESP-r simulation (in this specific case) seems to overestimate the thermal insulation and the thermal inertia of the building; in fact the predicted indoor air temperature is generally larger than the measured values both in steady-state and unsteady periods. Those small discrepancies can be due to some inaccurate information on thermophysical properties of the materials involved in the structural elements of the building such as thermal conductivity, specific heat and also the surface properties such as emissivity. A similar trend has been evidenced by other authors [19] who used ESP-r for the analysis of the thermal behaviour of ventilated building components. It is important to remember when processing these results that the tested envelope was built with light materials and for this reason it is characterized by a lower thermal capacity with respect to the typical envelope components used in real buildings. As a consequence, the time constant of this test building is low and the indoor air temperature is much more sensitive to the changes in the boundary conditions than in a building with more massive construction. However, the data shown in this paper demonstrate that the reliability of the numerical model of the rainscreen ventilated façade built with ESP-r is considered sufficient for design studies of such building components. 7. Limitations and future work Results presented in this paper about the thermal behaviour of the rainscreen skin envelope were limited to the winter period. However, in order to estimate the heat loads through this kind of envelope during the whole year an extended thermal analysis for typical summer conditions is needed. Further experimental data will be collected during the summer 2010. In addition, the effects of the indoor air re-circulated through the cavity on the energy consumption of a building will be studied in the future by using ESPr and other dynamic software. Additional simulations with a CFD code for the analysis of the air flow patterns through the ventilated cavity, particularly if coupled with the BS models, would be useful in providing results comparable with this study. It is intended that the numerical and experimental results will be used in order to define an index to represent the energy performance of the rainscreen ventilated façade. Although the comparisons have evidenced a good capability of the model to predict the thermal behaviour of the ventilated façade, the experimental campaign has shown that the measurement sys- 862 C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 Internal air temperature (oC) 25 Measured Predicted 20 15 10 5 0 01/01 02/01 03/01 04/01 05/01 06/01 07/01 08/01 09/01 10/01 11/01 12/01 13/01 Time Fig. 12. Comparison between the measured and predicted results of the internal air room temperature. tem of the air velocity within the cavity must be improved by using a larger number of sensors distributed along the cavity and more accurate sensors for low values of velocity. 8. Conclusions Ventilated façades can play a fundamental role in the thermal performance of buildings. It has been demonstrated that they act as barriers between external and internal conditions and help to reduce the energy consumption for cooling, ventilation and air conditioning. Moreover, they can help to produce healthy and comfortable indoor conditions. The analysis of the existing literature on the ventilated façade highlights that the rainscreen ventilated façade has received less attention than the other configurations. In order to fill this gap, in this work an analysis of the thermal performance of the rainscreen ventilated façade has been presented. The behaviour of this envelope component has been investigated both experimentally and numerically. Field measurements were performed during the 2009/10 winter season in a test building located in San Mauro Pascoli (Italy). The modelling of the test building was made using ESP-r. Three different multi-zone models were defined and the comparison with the experimental results was used in order to select the best ESP-r air flow network for the modelling of this kind of envelope component. The main results obtained can be summarized as follows: • If the thermal model of the ventilated wall does not include the open grills at the bottom and at the top of the wall and the presence of the open joints on the external rainscreen surface, the numerical results exhibit, as expected, a large disagreement with the experimental results. In this case, the agreement with the experimental data is strongly influenced by the orientation of the considered ventilated wall and the temperature increases for walls exposed to large values of the solar radiation. The air circulation within the cavity is due to the temperature difference between the cavity walls and the air only and this difference increases with the incident solar radiation. • In the modelling of the rainscreen ventilated façades the presence of the external open joints can be ignored without decreasing the accuracy of the numerical results significantly. Conversely, by ignoring the open joints, the airflow network model becomes simpler and the numerical solution of the balance equations is faster. For this reason, this kind of model can be suggested as more appropriate for long-term dynamic simulations in which the CPU time saving can be important. • The ESP-r model of this component underestimates the external rainscreen surface temperature during the night when the solar radiation is absent; on the contrary, in the presence of high lev- els of solar radiation, when the effect of this parameter becomes predominant with respect to the effects due to the external air temperature and to the wind, the agreement with the experimental data is very good. • The typical air cavity velocities predicted by the ESP-r model are very low (less than 0.12 m/s) and these values are confirmed the experimental data. The low velocity is due to the large thickness of the air cavity adopted in this ventilated façade (24 cm). • It was shown that the model of the whole building could predict the trend of the indoor air temperature during the period in which the heating plant of the tower was turned off, although the ESPr simulation overpredicts the degree of thermal insulation and the thermal inertia of the building, probably due to uncertainties in thermophysical properties of the materials involved in the structural elements of the building. More generally, the data shown in this paper demonstrate that the reliability of the numerical model of the rainscreen ventilated façade built with ESP-r can be considered good enough for engineering design. The dynamic thermal analysis of this component enables the designer to know under which conditions the ventilated façade is able to reverse the direction of the heat flux through the envelope during the year as a function of the local weather conditions. In this way the designer can estimate the possible contribution of this kind of envelope component to the reduction of the energy consumption related to the indoor heating. Acknowledgments The author gratefully acknowledges ESRU, University of Strathclyde for the ESP-r training. References [1] M. Ciampi, F. Leccese, G. Tuoni, Cooling of buildings: energy efficiency improvement through ventilated structures, in: Proceedings of the 1st International Conference on Sustainable Energy, Planning & Technology in Relationship to the Environment, EENV 2003, Halkidiki, Greece, 2003, pp. 199–210. [2] C. Afonso, A. Oliveira, Solar chimneys: simulation and experiment, Energy and Buildings 32 (2000) 71–79. [3] C. Balocco, A simple model to study ventilated façades energy performance, Energy and Buildings 34 (2002) 469–475. [4] C. Bartoli, M. Ciampi, G. Tuoni, Ventilated walls: airspace positioning and energy performance, in: Proceedings of 3rd International Thermal Energy & Environment Congress, ITEEC 97, vol. II, Marrakesh, Morocco, 1997, pp. 528–533. [5] L. Mei, D.G. Infield, U. Eicker, V. Fux, Thermal modelling of a building with an integrated ventilated PV façade, Energy and Buildings 35 (2003) 605–617. [6] J.A. Clarke, P.A. Strachan, Simulation of conventional and renewable building energy systems, Renewable Energy 5 (1994) 1178–1189. [7] K.S. Kumar, Pressure equalization of rainscreen walls: a critical review, Building and Environment 35 (2000) 161–179. C. Marinosci et al. / Energy and Buildings 43 (2011) 853–863 [8] A. Baskaran, Review of design guidelines for pressure equalized rainscreen walls, in: Internal Report No. 629, IRC, NRCC, Ottawa, Canada, 1992. [9] W. Smolec, A. Thomas, Some aspects of Trombe wall heat transfer models, Energy Conversion and Management 32 (1991) 269–277. [10] F. Mootz, J.-J. Bezian, Numerical study of a ventilated façade panel, Solar Energy 57 (1996) 29–36. [11] G. Gan, A parametric study of Trombe walls for passive cooling of buildings, Energy and Building 27 (1998) 37–43. [12] I. Cakmanus, Optimization of double skin façades for buildings: an office building example in Ankara-Turkey, in: Proceedings of Clima, 2007. [13] L. Gu, D. Gu, B. Lin, M. Huang, J. Gai, Y. Zhu, Life cycle green cost assessment method for green building design, Proceedings of Building Simulation (2007) 1962–1967. [14] D. Stribling, B. Stigge, A critical review of the energy savings and cost payback issues of double façades, CIBSE/ASHRAE Conference, (2003). [15] K. Roth, T. Lawrence, J. Brodrick, Double-skin façades, ASHRAE Journal (2007) 70–73. [16] K.S. Ong, A mathematical model of a solar chimney, Renewable Energy 28 (2003) 1047–1060. [17] J.A. Clarke, Energy Simulation in Building Design, 2nd ed., ButterworthHeinemann, Oxford, 2001. [18] H. Manz, T. Frank, Thermal simulation of buildings with double-skin façades, Energy and Buildings 37 (2005) 1114–1121. 863 [19] V. Leal, E. Maldonado, E. Erell, Y. Etzion, Modeling a reversible ventilated window for simulation within ESP-r—the solvent case, in: Eighth International IBPSA Conference, Eindhoven, Netherlands, 2003, pp. 713–720. [20] F. Alamdari, G.P. Hammond, Improved data correlations for buoyancy driven convection in rooms, Building Services Engineering Research & Technology 4 (1983) 106–112. [21] H. Poirazis, Double-skin façades for office buildings-literature review, in: Report ECBCS Annex 43, Division of Energy and Building Design, Department of Construction and Architecture, Lund Institute of Technology, Sweden, 2006. [22] P. Fazio, T. Kontopidis, Cavity pressure in rain screen walls, Building and Environment 23 (1988) 137–143. [23] R.I. Kimura, Scientific basis of air conditioning, Applied Science (1977) 175–186. [24] P.A. Strachan, Simulation support for performance assessment of building components, Building and Environment 43 (2008) 228–236. [25] P.A. Strachan, L. Vandaele, Case studies of outdoor testing and analysis of building components, Building and Environment 43 (2008) 129–142. [26] A. Pedrini, F.S. Westphal, R. Lamberts, A methodology for building energy modelling and calibration in warm climates, Building and Environment 37 (2002) 903–912. [27] D. Bronson, S.B. Hinchey, J.S. Haberl, D.L. O’Neal, A procedure for calibrating the DOE-2 simulation program to non-weather-dependent measured loads, ASHRAE Transactions 98 (1992) 636–652.