- Email: [email protected]
Prediction of hybrid ventilation performance using two simulation tools
Solar Energy 80 (2006) 908–926 www.elsevier.com/locate/solener
Prediction of hybrid ventilation performance using two simulation tools M. El Mankibi
a,*
, F. Cron b, P. Michel a, C. Inard
b
a
b
LASH, ENTPE, Rue Maurice Audin, 69518 Vaulx en Velin, France LEPTAB, University of La Rochelle, Avenue Michel Cre´peau, 17042 La Rochelle Cedex 1, France Received 22 October 2004; received in revised form 3 June 2005; accepted 29 August 2005 Available online 3 October 2005 Communicated by: Associate Editor Matheos Santamouris
Abstract This paper describes the results of a collaboration between ENTPE-LASH and LEPTAB within the framework of the IEA Annex 35 ‘‘Hybrid Ventilation in New and Retrofitted Buildings’’. The aim of the work is to carry out a cross-simulation study and to identify optimal control strategies for ventilation systems in order to provide a comfortable thermal indoor environment and a good indoor air quality with energy efficiency. Two models were developed by ENTPE-LASH and LEPTAB in order to carry out hybrid system simulations taking into account air flows, heat transfers and CO2 concentrations, and numerical results are compared in this study. The models were first adjusted to an experimental cell, HYBCELL, created in the same project. The simulations were carried out using a fictive classroom. The test room was assumed to be in Copenhagen and to be equipped with a natural ventilation system (two inlet grilles and an exhaust chimney) or with mechanical ventilation systems (fans with or without heat recovery). This work also reveals what are the differences in results between the two tools and outlines some conclusions on relative performance of the specific control strategies chosen in this study. 2005 Elsevier Ltd. All rights reserved. Keywords: Hybrid ventilation; Control strategies; Comfort; Energy efficiency
1. Introduction Nowadays, there is a growing interest in taking into account the ventilation at the early design stage of a building. Designers are interested in ventilation features as well as in indoor air quality, thermal comfort and energy consumption (Heiselberg, * Corresponding author. Tel.: +33 04 72 04 77 46; fax: +33 04 72 04 70 41. E-mail address: [email protected] (M. El Mankibi).
2000). One way to reach these targets is to combine both natural and mechanical ventilation advantages into a hybrid system. In addition, reliable design tools are needed to generate guidelines on control strategies. Examples of existing hybrid ventilation and control strategy systems in office and educational buildings (Aggerholm, 2002) show that control strategies have to be carefully designed to take into account the building design, the ventilation system, the occupant behaviors and their indoor climate expectations.
0038-092X/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.solener.2005.08.003
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
909
Nomenclature t Dt V b Si T tint T text T tz;i T ts;i T ti Tv T trm T tair hint
vent
hext qas qes cpas cpes ci ki ai ei cc r0 Fi,j Fvault
time (s) step time (s) zone volume (m3) slop () area of surface i (m2) indoor air temperature at time t (C) outdoor air temperature at time t (C) or (K) indoor air temperature at time t for zone i (C) surface i surface temperature at time t (C) or (K) node i temperature at time t (C) sky vault temperature (K) mean radiative temperature at time t (C) ventilation temperature at time t (C) internal convection coefficient (W m2 K1) external convection coefficient (W m2 K1) air sec density (kg m3) pollutant density (kg m3) specific heat of air sec at constant pressure (J kg1 K1) specific heat of pollutant at constant pressure (J kg1 K1) specific heat of material in layer i (J kg1 K1) thermal conductivity of material in layer i (W m1 K1) absorbance of material in layer i () emissivity () nebulosity indicia () Stefan–Boltzman constant (W m2 K4) view factor from i to j () view factor between fac¸ade and the sky vault ()
This paper describes the work done by both ENTPE/LASH and LEPTAB laboratories within the framework of IEA Annex 35. The main work objective is to identify optimal ventilation systems within a simple school building in Denmark. Two mechanical ventilation systems were compared with a hybrid ventilation system during three seasons. Analytical solutions are available to study natural driving forces (Li and Delsante, 2001) as well as numerical tools to perform a ventilation perfor-
FEnv Dxi W z Zb Zs ZNL Po,i Pz,i DPz,i Cp Cd v qmasji qmesji Ses uiConv uiSW uiLW Qventil QBelow QAbove Qik Uheating Ugains Urad
view factor between fac¸ade and the environment () sub-layers thickness (m) opening width (m) height (m) bottom of the opening (defined with reference to the floor) (m) top of the opening (defined with reference to the floor) (m) neutral level (m) reference pressure on the ground for the zone i (Pa) pressure at height z for the zone i (Pa) pressure difference across an exterior opening at a height z (Pa) pressure coefficient () discharge coefficient (m3 s1 Pa2) wind speed at the height of the building (m2 s1) pure air mass flow rate from the zone j to the zone i (kgas s1) CO2 mass flow rate from the zone j to the zone i (kges s1) pollutant source (kges s1) convective heat for surface i (W m2) short-wave radiative heat for surface i (W m2) long-wave radiative heat for surface i (W m2) ventilation flow (kg s1) volumetric flow below ZNL (m3 s1) volumetric above ZNL (m3 s1) volumetric flow from zone i to zone k (m3 s1) heating power (W) gains power (W) radiative gain power (W)
mance analysis (Frascatoro and Perino, 2002). Even if a hybrid-ventilated school has already been simulated with one of these tools (Jeong and Haghighat, 2002), some concerns remain about • the difficulties encountered to simulate the problem, • the robustness of the tools, • the possibility to generate general guidelines on control strategies to be developed.
910
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Each enclosure (opaque or transparent) has been meshed into insolated and shaded surfaces thanks to a sun patch model (Clarcke, 1985). The surface temperature T ts;i is assumed constant over the whole area of a surface.
Thus two simulation tools were developed for this study using coupled airflow and thermal models: HYBCELL1.0 and one implemented in the object-oriented solver SPARK. We propose here to compare results of simulations performed with the two different tools and to see how reliable these results can be use in order to conclude about the potential of a hybrid ventilation system compared to more traditional ones.
2.1.1. Heat transfer through enclosure structures Explicit finite difference method has been used to calculate the wall surface temperature. Each surface j is considered as one-dimensional transient heat conduction problem and is discretized into sub-layers of equal thickness (see Fig. 1), with temperatures being calculated at each time step. For sub-layers in contact with interior or exterior air, the following equations apply:
2. Description of the tools 2.1. HYBCELL1.0 HYBCELL1.0 (El Mankibi, 2003) has been developed at LASH/ENTPE. Simulations were carried out by coupling a thermal model based on finite differences and a pressure air flow model using the onion approach (Hensen, 1999). This tool has been developed under the Mathworks MATLAB/SIMULINK environment, and several control strategies for hybrid ventilation based on CO2 and temperature were implemented in the model. Schedule and occupation patterns were also taken into account in the model. Indoor air temperature is calculated from various heat transfers phenomena such as heat transfer through the walls and other enclosure structures, air infiltration and ventilation, internal heat gains and auxiliary heating or cooling. The air temperature Tint change in time dt is calculated as: dT int qas V cpas dtffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl
T tþdt nþ1 ¼
T tþdt ¼ 1
2 hint Dt T tint 2 k1 Dt T t2 þ q1 Dx1 c1 q1 ðDx1 Þ2 c1 2 hint Dt 2 k1 Dt 1 q1 Dx1 c1 q1 ðDx1 Þ2 c1
þ
2 Dt a ujSW int þ ujLW þ q1 Dx1 c1
Indoor air temperature change
¼
2 hext Dt T text 2 kn Dt T tn þ qn Dxn cn qn Dx2n cn 2 hext Dt 2 kn Dt þ 1 T tnþ1 qn Dxn cn qn Dxn cn 2 Dt a ujSW ext þ ujLW ext þ ð2Þ qn Dxn cn
X
S i hint ðT ts;i T tint Þ Surfaces |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
For interior sub-layer:
þ
T tþ1 ¼ i
Heat transfer through enclosures
UHeating þ Ugains |fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl} Auxiliary heating or cooling and internal gains þ Qventil cpas T tair vent T tint |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
ð1Þ
ki Dt 2
Ventilation
ρi , λ i , c i
ρn, λn, cn Δxn
Exterior Tn+1
… Tn
Δxi Ti+1
ρ1, λ1, c1
… Ti
Interior
Δx1 T2
T1 (Tint, hint)
(Text, hext) Fig. 1. Enclosure discretization used in HYBCELL1.0 model.
ð3Þ
ki Dt
ϕ SW_int ϕ LW int
ϕSW_ext ϕ LW ext
T t1
int
qi ðDxi Þ2 ci ! 2 ki Dt 1 T ti 2 qi ðDxi Þ ci
qi ðDxi Þ ci T ti1 þ
T tiþ1 þ
!
ð4Þ
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
For interior surfaces of adjoining constituent wall layer: T itþDt ¼
ki1 Dt Dxi1 ðqi1 ci1 2 þ qi ci Dx2 i Þ Dxi1
þ
T ti1
ki Dt T tiþ1 ðqi1 ci1 Dx2i1 þ qi ci Dx2 i Þ Dxi
1
ki1 Dt Dxi1 ðqi1 ci1 2 þ qi ci Dx2 i Þ Dxi1
! ki Dt þ T ti ðqi1 ci1 Dx2i1 þ qi ci Dx2 i Þ Dxi with h ci ki qi uSW uSL ai Dt T ti Dx
ð5Þ
convection coefficient (hint = 4) for inside and hext = 15 for outside (W m2 K1) specific heat of material in layer i (J kg1 K1) thermal conductivity of material in layer i (W m1 K1) density of material in layer i (kg m3) short-wave radiative heat (W m2) long-wave radiative heat (W m2) absorbance () step time (s) node i temperature at time t (C) sub-layers thickness (m)
The stability of the solution requires a` step time value so that the expressions (6) and (7) remain true. Therefore time interval depends on enclosure construction materials. The time step chosen for this problem is 1 min. Stability conditions: 2
Dt 6
ðDxÞ q c 2k
Dt 6
q ðDxÞ c 2 ðh Dx þ kÞ
for internal sub-layers
911
To compute the internal distribution of the shortwave radiation used in HYBCELL1.0, multiple reflections, direct retransmission and reflection to the outside has been taken into account (Pfrommer, 1995). Reflection has been assumed to be diffuse on the interior surfaces. Internal net long-wave radiation exchanges have been carried out using a radiosity method (Sacadura, 1980) and view factors are based on an area weighting technique as following: 8 S > PN j if j 62 Xi > > Sk > > < k62Xi if j 2 Xi F i;j ¼ 0 > > Xi ¼ set of facets located > > > : on the wall containing S i ð8Þ Radiative heat exchange of the fac¸ade (slop = b) with the outside environment (at the air temperature) and with the sky vault at the temperature Tv is calculated according the following view factors: 1 þ cosðbÞ 1 cosðbÞ and F Env ¼ ð9Þ 2 2 Thus, the following relation has been used in HYBCELL1.0
F vault ¼
uiLW ¼ ei r0 F vault ðT 4v T 4s;i Þ þ F Env ðT 4ext T 4s;i Þ ð10Þ where r0 ei
Stefan–Boltzman constant (W m2 K4) emissivity ()
ð6Þ
Sky vault temperature has been computed using esp-r (clarck) model:
ð7Þ
T v ¼ 9:365574 106 ð1 ccÞ T 6ext 4 þ T ext cc ð1 0:84 ccÞ 0:527 þ 0:84 cc
2
for boundary sub-layers
2.1.2. Radiative heat exchange The short-wave radiation calculation is based on a sun patch model (Clarcke, 1985). This model includes the calculation of shades generated by the surrounding masks at the fac¸ade, the determination of the sunny parts of glazed areas and their projection on the interior surfaces following the direction of the solar beam.
0:25 273 þ 0:161 exp 8:45 1 T ext
ð11Þ
where cc (cloud and cover) (Dumortier, 1995) is a nebulosity indicia. 2.1.3. Air flow model HYBCELL1.0s airflow model is based on the flow network approach (nodal method) (Allard,
912
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
1998) where each zone of a building is represented by a pressure node. Boundary nodes are also used to represent the environment outside the building. Those nodes are then interconnected by flow paths, such as crack or windows. Application of mass balance equation on a zone i with j flow paths gives: 0¼
N X
qi Qik
ð12Þ
k¼1
where Qik qi
volumetric flow from zone i to zone k (m3 s1) air density in the direction of the flow (kg m3)
Air flow is attributed to pressure differences across the openings taking into account the air motion due to the wind and the temperature difference across the opening resulting to buoyancy driven flow. The wind pressure at the external node is calculated from Bernoullis equation as follows: Pw ¼
1 C p q v2 2
where Cp v
ð13Þ
pressure coefficient () wind speed at the height of the building (m s1)
The reference pressure on the ground being Po,i and according to hydrostatic gradient, the pressure due to stack effect only at a height z, is: P z;i ¼ P o;i qi g z
ð14Þ
The pressure difference across an exterior opening at a height z is then equal to: DP z;i ¼ P o;i P o;j g ðqi qa Þ z P w
ð15Þ
The air flow through large opening is assumed to be bi-directional. Thus, a level of zero pressure level (neutral level: DPz, i = 0) is located at height ZNL from the ground. The bi-directional flow is then calculated by integrating (15) between bottom or top of the opening and the neutral level: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DP z dz 2 QBelow ¼ C d W q Zb |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Z
Z NL
Air flow below neutral level
ð16Þ
and sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DP z dz 2 QAbove ¼ C d W q Z NL |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Z
Zs
ð17Þ
Air flow above neutral level
Newton–Raphsons method is then used to solve the system obtained by applying Eq. (12) where Cd W Zb Zs ZNL
discharge coefficient (m3 s1 Pa2) opening width (m) bottom of the opening (defined with reference to the floor) (m) top of the opening (defined with reference to the floor) (m) neutral level (m)
2.2. The simulation tool using the solver SPARK The object-oriented solver SPARK is used at LEPTAB laboratory (France) to compute problems with airflow and thermal model equations. SPARK is a general solver, thus the equations for the room, the outdoor and indoor environments, the occupancy pattern, the ventilation system and the control strategies are all implemented in SPARK. Some models that have been written for a previous Annex 35 study (Cron et al., 2002) are also used here. SPARK is robust and able to compute an entire problem with differential and non-linear equations, so airflow and thermal equations are solved simultaneously at each time step (1 min). 2.2.1. Models describing the air within the zone The model developed using SPARK considers a single zone model with an assumed hydrostatic pressure variation. This modeling is based on the ‘‘pure’’ air and the pollutant mass balance equations. For a zone i with n openings: n n X X qmasji qmasij ¼ 0 ð18Þ j¼1
j¼1
and n n X X dq qmesji qmesij þ S es ¼ V es dt j¼1 j¼1
ð19Þ
qmasji (kgas s1) is the pure air mass flow rate from the zone j to the zone i and likewise for the CO2 mass flow rate qmesji (kges s1). Ses is the pollutant source (kges s1) and V the volume (m3).
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
The thermal balance is n X
of accumulation taken into account according to ROUX (Rumianowsky et al., 1989). In that case, we had R ¼ R4tot and C = Ctot, and M was placed in the middle of the partition.
qmasji cpas þ qmesji cpes T tz;j
j¼1
n X
913
qmasij cpas þ qmesij cpes T tz;i
2.2.3. Convective heat transfer The convective heat transfer was described by uconv = h Æ (T Ts,i) where Ts,i is the surface temperature in (C). b For an indoor surface, hint ¼ ajT tint T ts;i j , a and b given by ALLARD (Allard, 1987). For an outdoor surface hext = c + dvn where c = 2.5, d = 3.5, n = 1 and v is the wind velocity in (m s1) according to (Ferries, 1984).
j¼1
þ S es cpes ðT es T tz;i Þ þ Uheating þ UGain þ Uconv @T z;i @q þ cpes V T tz;i es ¼ qas cpas þ qes cpes V @t @t dT z;i dqes t þ cpes V T z;i ¼ qas cpas þ qes cpes V dt dt ð20Þ For closure, the ideal gas is used: P ¼ ðqas ras þ qes res Þ ðT þ 273:15Þ
2.2.4. Radiative heat exchange The long-wave radiation model used for the indoor radiative heat exchange was a mean radiative temperature model. The following relations were used for a unit of surface:
ð21Þ
2.2.2. Conductive heat transfer model For the walls, the conductive heat transfer was described by an electrical 2R-3C model (see Fig. 2) to have a good response to an indoor high frequency excitation (Rumianowsky et al., 1989). We made a comparison between the response of this model and the one given by a finite difference model. The position of M (see Fig. 2) was set to the optimum value of eA that provided the minimum error of EeA. EeA is given by Z 1 ð22Þ EeA ¼ jT A finite difference T A j dt
uiLW
¼ hrmi ðT trm T ts;i Þ t 3 T s;i þ T trm with hrmi ¼ 4r0 ei þ 273:15 2 and
P hrmi S i T ts;i þ Urad þ ni¼1 uiLW int S i T ts;i Pn i¼1 hrmi S i
layer j Outdoor
Ti (t)
TM (t) Tj-1 (t)
Tj (t) TB (t)
e A , nA
CA
eB , nB TM (t)
TA (t) RA
ð24Þ
The direct short-wave radiation that is transmitted through the window was supposed to be entirely incident on the floor surface. One part is absorbed and the other part is supposed to be reflected in a diffuse way. Both the entering diffuse solar radiation and the one reflected on the floor were supposed to
layer i
Ti-1(t)
i¼1
ð23Þ
ð25Þ
The simulated room was surrounded by rooms with the same conditions, so the partitions were adiabatic and considered symmetrical. The electrical model used was a 1R-1C electrical model, to have the effect
TA (t)
Pn
T trm ¼
0
Indoor
int
CM
TB(t) RB
Fig. 2. Electrical model used in SPARK model.
CB
914
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
be distributed all over the surfaces depending on their area ratio and absorbed (for the window, one fraction was transmitted outdoors again). The outdoor long-wave radiation per surface for an outdoor surface i is given by 1 4 1 4 i 4 T þ T T s;i uLW ¼ ei r0 ð26Þ 2 ext 2 v
numerical models. HYBCELL is implemented within a large hall where temperature can be controlled to create an artificial climate around the experimen-
Ts,i is the outdoor surface temperature, Text the outside ambient air temperature and Tv the sky vault temperature. All temperatures are in Kelvin. The short-wave solar radiation absorbed by the outdoor wall surface was calculated given the surface incident solar radiation and the solar properties. 3. Comparison between experimental and numerical results Experimental data provided by the HYBCELL test cell, designed at LASH/ENTPE laboratory (El Mankibi et al., 2001), has been used to adjust the
Fig. 4. Internal view of HYBCELL test cell.
Fig. 3. HYBCELL test cell architecture.
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
915
Fig. 5. Location of sensors and actuators inside HYBCELL test cell.
tal cell. The test cell has only one external wall that is sloped and has six sash windows power-driven by step by step motors. The cell is 5.1 m long, 3.25 m wide and 2.95 m high (see Figs. 3 and 4). The cell is equipped with an electric heater, a fan, window engines, CO2 generation and sensible heat supply devices. Various sensors (temperature, pressure, relative humidity, CO2 concentration, VOC) have been installed in the test cell as well as in the hall and outdoors. In addition, wind velocity, wind direction and solar radiation data are provided by a meteorological station located near the cell. The CO2 generation and sensible heat supply devices simulate occupancy in the experimental cell. The data acquisition system is made up from a DAQ card and is supervised by a computer. The following figure gives the location of sensors and actuators in the test cell (see Fig. 5). Experimental monitoring has been held in for seven days in summer (from 19 to 25 June 2001) and seven days in winter (from 25 to 31 December 2002). Fig. 6 shows the outdoor boundary conditions in terms of air temperature and solar radiation. Results indicate that SPARK and HYBCELL1.0 agree well in modeling indoor air temperature (see Fig. 7). The small differences between the tools are mostly due to the different models used, such as the convec-
tion coefficients and the way short-wave and longwave radiation exchanges have been modeled. In addition, a comparison made between HYBCELL1.0, SPARK and coupled COMIS and TRNSYS models in the framework of HYBVENT project (Heiselberg, 2000) has shown a good agreement. HYBCELL1.0 and SPARK have been chosen to lead this study thanks to the simplicity of control strategies implementation that they provide. 4. Simulated test room A fictive classroom assumed to be located in Copenhagen has been modeled to test various control strategies. This room is a single zone at the middle floor of a three-storey building, is 9 m long, 6 m wide and 3 m high (see Fig. 8) and has a large glazed facade oriented South. The classroom is equipped either with a natural ventilation system (two inlet grilles and an exhaust chimney) or with mechanical ventilation systems (fans with or without heat recovery). The control strategies are based on indoor air temperature and CO2 concentration. The window is a line of 3/12/4 argon-filled lowemissivity double glazing (U-value of 1.5 W m2 K1) along the whole facade and is 1 m above the floor. The overall transmittance and absorptance
916
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Fig. 6. Outdoor conditions.
for each pane is a function of the solar radiation incident angle. The Table 1 summarizes the characteristics of the envelope and materials. The classroom is occupied by 24 children and a teacher from Monday to Friday between 8:00 and 12:00 and from 13:00 to 15:00. On Tuesdays between 10:00 and 11:00 only half of the pupils are
in the room and from 14:00 to 15:00 everybody is out. On Wednesday afternoons everybody is out. Internal loads related to occupants and lighting were taken into account: • occupant: 80 W per person and 0.018 m3 h1 of CO2 per person, • lighting: 10 W per m2 of floor area.
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
917
Fig. 7. Measured and simulated indoor air temperatures.
Outdoor CO2 was assumed to be constant and equal to 400 ppm. 5. Ventilation system and control strategies A 0.2 ach permanent infiltration rate is assumed, independently of the weather conditions and the ventilation system. For each simulation, a solar shading coefficient equal to 0.2 is used in spring and summer when global solar radiation is higher than 200 W m2.
Two reference mechanical systems and one hybrid ventilation system were compared. The classroom is equipped with two inlet grilles and an exhaust chimney with a fan (see Fig. 9). The grilles are positioned under the windows, 0.5 m above the floor level. Each grille is 1.8 m wide and is closed outside normal school hours except when night cooling is on. The exhaust chimney is 4.0 m high and is a simple duct with a fan. The top of the chimney is 10 m above the ground.
918
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
1m
Window
Chimney with fan 3m
1m 1m
4m
Vertical section
Facade
6m
2.5 m
9m
Grilles
3m
Horizontal section Fig. 8. Architecture of Copenhagen test room.
The cp-values of the facade, given in Table 2, are from AIVC (length to width ratio: 2:1 and building surrounded by obstructions equivalent to half the height of the building) (Liddament, 1996). The roof (chimney top) has a cp-value independent of wind direction and equal to 0.6. The airflow rates through the grilles and the chimney are given by a power law relation: Q = C Æ (DP)0.5, with • each inlet grille: C = 0.053 m3 s1 Pa0.5, • the exhaust chimney: C = 0.088 m3 s1 Pa0.5.
3m
Fig. 9. Copenhagen test room with the ventilation system.
15:00). The low fan power consumption is 1 W per m3 s1. The supply air is preheated to 18 C in winter and spring when the external temperature is below 18 C.
5.1. Reference ventilation system 1: mechanical exhaust
5.2. Reference ventilation system 2: balanced mechanical ventilation
A constant supply airflow rate of 0.150 m3 s1 is provided during all normal school hours (8:00 to
The mechanical supply and exhaust air flows are 0.150 m3 s1 with a fan consumption of 2.5 W per
Table 1 Characteristics of Copenhagen test room envelope Material from inside
Thickness (m)
Conductivity (W K1 m1)
Density (kg m3)
Floor and ceiling
PVC Massive concrete
0.005 0.15
0.35 1.7
1300 2300
External wall
Massive concrete Mineral wool Brick
0.1 0.15 0.11
1.7 0.04 0.8
Partition walls
Plaster board Mineral wool Plaster board
0.013 0.05 0.013
0.2 0.04 0.2
Specific heat (J kg1 K1)
Solar absorptance ()
Emissivity ()
960 920
0.5 0.5
0.9 0.9
2300 30 180
920 840 840
0.5 – 0.7
0.9 – 0.9
800 30 800
1100 840 1100
0.5 – 0.5
0.9 – 0.9
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926 Table 2 Pressure coefficients used in the simulations Wind angle
0
45
90
135
180
cp-value
0.25
0.06
0.35
0.60
0.50
m3 s1. A heat recovery with a temperature efficiency of 60% is available and works in spring and winter. The supply air is preheated up to 18 C using a heating coil. The heat recovery is continuously controlled to get a supply air temperature of 18 C and to turn off for outdoor temperatures higher than 18 C. This system is on during all normal school hours (8:00 to 15:00).
919
perature goes below 21 C. The air change rate when the window is open does not depend on outdoor conditions: it is a constant value of 4 ach additional to the ventilation system airflow rate. The temperature is also controlled using a heating system in winter and spring. Heating hours are 7.00–15.00 from Monday to Friday. The heating controller is a proportional controller based on air temperature with two set points: • heating hours: 21 C, • non-heating hours: 18 C. The maximum heating power is 5 kW.
5.3. Hybrid ventilation system
5.6. Night cooling
The specific hybrid system simulated here is a fan-assisted stack exhaust with preheating of supply air at 18 C in winter and spring. The inlet grilles and exhaust chimney are the same as in reference ventilation system 1, but a low-pressure fan was considered.
Night cooling is provided only in summer and is scheduled between 22:00 and 07:00. It turns on when the indoor air temperature is higher than 24 C and when the indoor–outdoor air temperature difference is higher than 2 C. It turns off whenever indoor air temperature is below 18 C. For the reference mechanical systems, night cooling works with grilles opened and fans on. For the specific hybrid ventilation system the night cooling is provided by mechanical ventilation (the grilles open and the fan turned on).
5.4. CO2 control The mechanical systems operate with constant airflow rates during normal school hours and do not depend on the CO2 concentration. Thus there is no CO2 control when simulating the two reference systems. For the hybrid system, the designed control strategy is an on–off control strategy based on CO2 concentration: • the first inlet grille opens if the CO2 is higher than 800 ppm, • the second inlet grille opens if the CO2 is higher than 1000 ppm, • the fan in the exhaust chimney is turned on if the CO2 is higher than 1200 ppm, • the fan turns off if the CO2 is lower than 1100 ppm, • the second inlet grille close if the CO2 is lower than 900 ppm, • the first inlet grille close if the CO2 is lower than 700 ppm. 5.5. Temperature control The window is open when both indoor and outdoor air temperatures are higher than 23 C and 12 C respectively. It is closed when the indoor tem-
6. Results and discussion Three periods of three weeks were simulated: from the 1st to the 21st of January (winter), from the 2nd to the 22nd of April (spring) and from the 21st of May to the 10th of June (summer). Figs. 10 and 11 show outdoor conditions (temperature and solar radiation) for the last week of each period. The three weeks of each season were simulated, but only the last one was used for result discussion. The first two weeks served to eliminate the effects of different assumptions of the starting conditions. 6.1. Winter results For all three ventilation system predictions, both simulation tools give equivalent results in terms of indoor air temperatures and CO2 concentrations. As an example, Fig. 12 shows the indoor air temperature and the CO2 concentration plots for the specific hybrid ventilation system. Both HYBCELL1.0 and SPARK predict about the same
920
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Fig. 10. Copenhagen outdoor air temperature for the last week of each simulation period.
Fig. 11. Copenhagen solar radiation for the last week of each simulation period.
behavior for the classroom in terms of indoor air temperature. However, a difference around 0.5 C can be noticed. This is due to differences in modeling the envelope of the building.
The CO2 concentration results are somehow different during occupancy for the 1st day (see Fig. 12). In fact, the differences are mostly due to the airflow: a variation around 15% of the
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
921
Fig. 12. Air temperature and CO2 concentration for the hybrid ventilation system in winter.
mass air flow rate can lead to a CO2 difference of 100 ppm. This mass air flow rate can vary between the two models because of the way the dynamic wind pressure has been taken into account (the coefficients of correction of wind velocity were different: 0.47 for HYBCELL1.0 and variable coefficient for SPARK) and also because of the air density difference due to differences in temperature.
The influence of the airflow on the CO2 concentration is also shown during the 5th day. Indeed, the dead band of 100 (ppm) for the second inlet grille control seems to be a little bit tight, since HYBCELL1.0 results present some oscillations. The difference between the CO2 decrease at the end of the 1st, the 4th and the 5th days is due to the way the grilles close. With SPARK they close
922
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
at 15:00 while the kids are leaving the classroom, whereas, with HYBCELL1.0, the grilles close right after everybody is out. HYBCELL1.0 predicts lower CO2 concentration and higher maximum indoor air temperatures than SPARK (see Tables 3 and 4). The comparison between results obtained with the reference mechanical ventilation systems and the hybrid ventilation system shows that the indoor air temperature is about the same for the three cases, but hybrid ventilation has lower mean and maximum CO2 concentration when the pupils are in the classroom. Hybrid ventilation appears to provide better indoor air quality in terms of CO2 concentrations. HYBCELL1.0 gives higher air flow rates with the hybrid ventilation system than SPARK. This explains the lower CO2 concentration as well as the higher preheating consumption obtained with HYBCELL1.0. Differences in heating consumption are first due to higher heating needs for HYBCELL1.0 in the morning when the set point temper-
Table 3 Mean and maximum indoor air temperatures for the three ventilation systems in winter Model
Mean air temperatures during occupancy (C)
Maximum air temperatures during occupancy (C)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
SPARK HYBCELL1.0
21.50 21.58
21.50 21.58
21.30 21.35
22.60 23.48
22.60 23.48
22.30 23.66
Table 4 Mean and maximum CO2 concentrations for the three ventilation systems in winter Model
SPARK HYBCELL1.0
Mean CO2 concentrations during occupancy (ppm)
Maximum CO2 concentrations during occupancy (ppm)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
1105 1051
1105 1051
958 902
1193 1137
1193 1137
1062 1048
ature changes to 21 C at 7:00. The differences between radiative heat transfer models explain also the heating need differences. Although there are differences in predicting heating and preheating consumption for the hybrid ventilation system, both HYBCELL1.0 and SPARK give higher room heating and preheating needs with hybrid ventilation than with the reference ones. There is no consumption for the fan, the stack effect is important to ensure enough airflow rates. The best results in terms of energy consumption are the ones obtained with the reference system 2 (see Table 5). 6.2. Spring results As for the winter results, there is still a small difference here for indoor air temperature and for CO2 concentration, see Fig. 13. Solar radiation is more important, so there are more internal gains, and for both tools the indoor air temperature is higher than the set point. HYBCELL1.0 again presents more airflow rates than SPARK and it leads to lower CO2 concentrations. Despite higher airflow rates, the indoor air temperature predicted by HYBCELL1.0 is slightly higher than the one for SPARK and without additional heating. This is due to the differences between the radiative heat transfer models. In spring, Tables 6 and 7 show again that the mean and maximum CO2 concentrations are lower with the hybrid system than with the mechanical ones. Hybrid ventilation is also better in spring in terms of indoor air quality. CO2 concentration is still below 1200 ppm so here again the fan does not need to be turned on. For this season too, the hybrid ventilation system is still the most expensive system in terms of energy consumption. But, unlike the winter period results, here the tools predict more significant differences in heating and preheating consumption (see Table 8). This can be explained by the higher airflow rates for HYBCELL1.0 than for SPARK concerning the hybrid mode.
Table 5 Energy consumption in winter Model
SPARK HYBCELL1.0
Heating (kW h/week)
Pre-heating (kW h/week)
Fan power (kW h/week)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
21.39 32.64
21.39 32.64
21.43 34.22
119.00 128.69
34.85 37.11
125.69 185.00
5.26 5.25
13.16 13.13
0.00 0.00
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
923
Fig. 13. Air temperature and CO2 concentration for the hybrid ventilation system in spring.
Table 6 Mean and maximum indoor air temperatures for the three ventilation systems in spring Model
SPARK HYBCELL1.0
Table 7 Mean and maximum CO2 concentrations for the three ventilation systems in spring
Mean air temperatures during occupancy (C)
Maximum air temperatures during occupancy (C)
Model
Mean CO2 concentrations during occupancy (ppm)
Maximum CO2 concentrations during occupancy (ppm)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
22.30 22.21
22.30 22.21
21.80 21.70
23.00 23.12
23.00 23.12
23.00 22.81
1107 1071
1107 1071
975 884
1194 1154
1194 1154
1148 1053
SPARK HYBCELL1.0
924
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Table 8 Energy consumption in spring Model
SPARK HYBCELL1.0
Heating (kW h/week)
Pre-heating (kW h/week)
Fan power (kW h/week)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
8.57 6.70
8.57 6.70
6.03 8.45
58.21 60.92
8.21 8.59
61.23 94.24
5.26 5.25
13.16 13.13
0.00 0.00
Fig. 14. Air temperature and CO2 concentration for the hybrid ventilation system in summer.
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
925
Table 9 Mean and maximum indoor air temperatures for the three ventilation systems in summer Model
Mean air temperatures during occupancy (C)
Maximum air temperatures during occupancy (C)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
SPARK HYBCELL1.0
24.40 24.37
24.40 24.37
24.90 24.54
26.60 26.69
26.60 26.69
27.00 26.88
Table 10 Mean and maximum CO2 concentrations for the three ventilation systems in summer Model
Mean CO2 concentrations during occupancy (ppm)
Maximum CO2 concentrations during occupancy (ppm)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
SPARK HYBCELL1.0
754 755
754 755
843 811
929 1158
929 1158
925 1081
Thus, for this season, these differences mean quantitative results have to be handled with care before giving any conclusion on the ventilation system performance. 6.3. Summer results Fig. 14 gives the evolutions of indoor air temperature and CO2 concentration for the hybrid system in summer. During occupancy, HYBCELL1.0 and SPARK give again quite similar results except for the night cooling. The difference occurs during the last night when the indoor air temperature given by HYBCELL1.0, unlike the one for SPARK, is not high enough to make the night cooling work. The CO2 concentration raises quickly the first day for the HYBCELL1.0 simulations. In fact, SPARK predicts an indoor temperature higher than 23 C, so the window opens whereas HYBCELL1.0 has still the window closed and thus lower airflow rate and a higher CO2 concentration. For both simulation tools Tables 9 and 10 show that the mean and maximum indoor air temperaTable 11 Energy consumption in summer Model
Fan power (kW h/week) Ref 1
Ref 2
HV
SPARK HYBCELL1.0
9.47 9.30
23.66 23.24
0.92 0.79
tures are now slightly higher for the hybrid system than for the mechanical ones. Here the mean CO2 concentration is higher with hybrid ventilation system than with the mechanical modes, but stays below 1000 ppm. In fact, as the window is open most of the time, the mechanical ventilation works with the fan on and the window open, so there is more airflow rate than with the hybrid ventilation. However, the maximum value of CO2 concentration remains the lowest one and some exposure to high CO2 concentration is avoided. Here again, this specific hybrid ventilation is better in terms of indoor air quality regarding to CO2 concentration. Since there is neither heating nor preheating energy, the results of this period indicate that the hybrid ventilation system is more interesting than the two reference systems (see Table 11). The difference for the fan consumption is simply due to the night cooling that was on more often for the SPARK simulations. 7. Conclusions Through a cross-simulation study, this paper has reported upon both the comparison of three ventilation systems performances and the ability of two numerical models (HYBCELL1.0 and SPARK) to predict similar behavior of the simulated test room for several control strategies. According to the first point of view, the simulation results show that
926
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
• In winter, hybrid ventilation system has higher energy consumption but there is a significantly lower mean and maximum CO2 concentrations compared to the mechanical system ones. There is a better indoor air quality for the hybrid ventilation system with equivalent thermal comfort in terms of indoor air temperature for the three ventilation systems. • In spring, hybrid ventilation system still has higher energy consumption with better indoor air quality and thermal comfort. However, differences in energy consumption between the three ventilation systems are lower than in winter. • In summer, hybrid ventilation energy consumption is lower than reference systems ones but the CO2 concentrations are a little bit higher. On the other hand, tools were in agreement according to CO2 concentration and indoor air temperature and the relative energy consumption performances of the three ventilation systems. The noticed disagreement in energy consumption between HYBCELL1.0 and SPARK is due to differences in modeling long-wave and short-wave radiation and convection heat transfer coefficients. According to this study results and to the noticed agreement between used models, simulation tools can be used to predict the relative performances of ventilation systems. However, particular care must be taken to avoid wrong interpretations linked to energy consumption. Indeed, ON–OFF controller can lead to differences between tools. Thus, advanced control strategies (PID or fuzzy controller) must be tested. Acknowledgements This work has been carried out within the frame of the Annex 35 ‘‘hybrid Ventilation in New and Retrofitted Buildings’’ and was supported by the French environmental and energy agency ADEME (Agence De lEnvironnement et de la Maıˆtrise de lEnergie), by the French electricity company EDF (Electricite´ de France) and by the French Ministry. The co-author thanks also Ashok Gadgil and Dimitri Curtil from the LBNL (Lawrence Berkeley National Laboratory) for their help.
References Aggerholm, A., 2002. Hybrid ventilation principle and control strategies. In: Proceedings of the 4th International Forum on Hybrid Ventilation, Montreal, Canada, 14–15 May 2002. Allard, F., 1987. Contribution a` le´tude des transferts de chaleur dans les cavite´s thermiquement entraıˆne´es a` grand nombre de Rayleigh, The`se dEtat, INSA de Lyon. Allard, F., (Cord), 1998. Natural ventilation in building. James & James Ltd., London, p. 356. Clarcke, J.A., 1985. Energy Simulation in Building Design. Adam Hilger Ltd, Bristol and Boston, p. 388. Cron, F., El Mankibi, M., Inard, C., Michel, P., 2002. Experimental and numerical analysis of a hybrid-ventilated room. In: 8th International Conference on Air Distribution in Rooms, Copenhagen, Denmark, 8–11 September 2002. Dumortier, D., 1995. Mesure, Analyse et Mode´lisation du gisement lumineux. Application a` le´valuation des performances de le´clairage naturel des baˆtiments, Doctorat: Universite´ de Savoie, p. 292. El Mankibi, M., Michel, P., Guarracino, G., 2001. Control Strategies for hybrid ventilation development of an experimental device. In: Proceedings of the 22nd Annual AIVC Conference, Bath, UK, 11–14 September 2001. El Mankibi, M., 2003. De´veloppement et e´valuation nume´rique et expe´rimentale Des strate´gies de re´gulation de la ventilation hybride, The`se de doctorat. INSA Lyon, p. 384. Ferries, B., 1984. Contribution a` le´tude des enveloppes climatiques et aide a` leur conception par micro-ordinateur, The`se Doct.-Ing., INSA de Lyon. Frascatoro, G.V., Perino, M., 2002. Natural vs. mechanical ventilation—a tool to help making a choice. International Journal of Ventilation 1 (2), 101–108. Heiselberg, P. (Ed.), 2000. Principles of hybrid Ventilation, IEA Energy Conservation in Buildings and Community Systems Program Annex 35: Hybrid Ventilation in New and Retrofitted Buildings. Hensen, J.L.M., 1999. A comparison of coupled and de-coupled solution for temperature and air flow in a building. ASHRAE Transactions 105 (2), 962–969. Jeong, Y., Haghighat, F., 2002. Modeling of a hybrid-ventilated building—using ESPr. International Journal of Ventilation 1 (2), 127–139. Li, Y., Delsante, A., 2001. Natural ventilation induced by combined wind and thermal forces. Buildings and Environment 36 (1), 59–71. Liddament, M.W., 1996. A guide to energy efficient ventilation AIVC, Coventry, UK. Pfrommer, P., 1995. Thermal modelling of highly glazed spaces. Ph.D: De Montfort University, Leicester, p. 229. Rumianowsky, P., Brau, J., Roux, J.J., 1989. An adapted model for simulation of interaction between a wall and the building heating system. In: Thermal performance of the exterior envelopes of buildings IV conference, Orlando, USA. Sacadura, J.F., 1980. Initiation aux transferts thermiques, Techniques et Documentation, CAST INSA, Lyon.
Prediction of hybrid ventilation performance using two simulation tools M. El Mankibi
a,*
, F. Cron b, P. Michel a, C. Inard
b
a
b
LASH, ENTPE, Rue Maurice Audin, 69518 Vaulx en Velin, France LEPTAB, University of La Rochelle, Avenue Michel Cre´peau, 17042 La Rochelle Cedex 1, France Received 22 October 2004; received in revised form 3 June 2005; accepted 29 August 2005 Available online 3 October 2005 Communicated by: Associate Editor Matheos Santamouris
Abstract This paper describes the results of a collaboration between ENTPE-LASH and LEPTAB within the framework of the IEA Annex 35 ‘‘Hybrid Ventilation in New and Retrofitted Buildings’’. The aim of the work is to carry out a cross-simulation study and to identify optimal control strategies for ventilation systems in order to provide a comfortable thermal indoor environment and a good indoor air quality with energy efficiency. Two models were developed by ENTPE-LASH and LEPTAB in order to carry out hybrid system simulations taking into account air flows, heat transfers and CO2 concentrations, and numerical results are compared in this study. The models were first adjusted to an experimental cell, HYBCELL, created in the same project. The simulations were carried out using a fictive classroom. The test room was assumed to be in Copenhagen and to be equipped with a natural ventilation system (two inlet grilles and an exhaust chimney) or with mechanical ventilation systems (fans with or without heat recovery). This work also reveals what are the differences in results between the two tools and outlines some conclusions on relative performance of the specific control strategies chosen in this study. 2005 Elsevier Ltd. All rights reserved. Keywords: Hybrid ventilation; Control strategies; Comfort; Energy efficiency
1. Introduction Nowadays, there is a growing interest in taking into account the ventilation at the early design stage of a building. Designers are interested in ventilation features as well as in indoor air quality, thermal comfort and energy consumption (Heiselberg, * Corresponding author. Tel.: +33 04 72 04 77 46; fax: +33 04 72 04 70 41. E-mail address: [email protected] (M. El Mankibi).
2000). One way to reach these targets is to combine both natural and mechanical ventilation advantages into a hybrid system. In addition, reliable design tools are needed to generate guidelines on control strategies. Examples of existing hybrid ventilation and control strategy systems in office and educational buildings (Aggerholm, 2002) show that control strategies have to be carefully designed to take into account the building design, the ventilation system, the occupant behaviors and their indoor climate expectations.
0038-092X/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.solener.2005.08.003
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
909
Nomenclature t Dt V b Si T tint T text T tz;i T ts;i T ti Tv T trm T tair hint
vent
hext qas qes cpas cpes ci ki ai ei cc r0 Fi,j Fvault
time (s) step time (s) zone volume (m3) slop () area of surface i (m2) indoor air temperature at time t (C) outdoor air temperature at time t (C) or (K) indoor air temperature at time t for zone i (C) surface i surface temperature at time t (C) or (K) node i temperature at time t (C) sky vault temperature (K) mean radiative temperature at time t (C) ventilation temperature at time t (C) internal convection coefficient (W m2 K1) external convection coefficient (W m2 K1) air sec density (kg m3) pollutant density (kg m3) specific heat of air sec at constant pressure (J kg1 K1) specific heat of pollutant at constant pressure (J kg1 K1) specific heat of material in layer i (J kg1 K1) thermal conductivity of material in layer i (W m1 K1) absorbance of material in layer i () emissivity () nebulosity indicia () Stefan–Boltzman constant (W m2 K4) view factor from i to j () view factor between fac¸ade and the sky vault ()
This paper describes the work done by both ENTPE/LASH and LEPTAB laboratories within the framework of IEA Annex 35. The main work objective is to identify optimal ventilation systems within a simple school building in Denmark. Two mechanical ventilation systems were compared with a hybrid ventilation system during three seasons. Analytical solutions are available to study natural driving forces (Li and Delsante, 2001) as well as numerical tools to perform a ventilation perfor-
FEnv Dxi W z Zb Zs ZNL Po,i Pz,i DPz,i Cp Cd v qmasji qmesji Ses uiConv uiSW uiLW Qventil QBelow QAbove Qik Uheating Ugains Urad
view factor between fac¸ade and the environment () sub-layers thickness (m) opening width (m) height (m) bottom of the opening (defined with reference to the floor) (m) top of the opening (defined with reference to the floor) (m) neutral level (m) reference pressure on the ground for the zone i (Pa) pressure at height z for the zone i (Pa) pressure difference across an exterior opening at a height z (Pa) pressure coefficient () discharge coefficient (m3 s1 Pa2) wind speed at the height of the building (m2 s1) pure air mass flow rate from the zone j to the zone i (kgas s1) CO2 mass flow rate from the zone j to the zone i (kges s1) pollutant source (kges s1) convective heat for surface i (W m2) short-wave radiative heat for surface i (W m2) long-wave radiative heat for surface i (W m2) ventilation flow (kg s1) volumetric flow below ZNL (m3 s1) volumetric above ZNL (m3 s1) volumetric flow from zone i to zone k (m3 s1) heating power (W) gains power (W) radiative gain power (W)
mance analysis (Frascatoro and Perino, 2002). Even if a hybrid-ventilated school has already been simulated with one of these tools (Jeong and Haghighat, 2002), some concerns remain about • the difficulties encountered to simulate the problem, • the robustness of the tools, • the possibility to generate general guidelines on control strategies to be developed.
910
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Each enclosure (opaque or transparent) has been meshed into insolated and shaded surfaces thanks to a sun patch model (Clarcke, 1985). The surface temperature T ts;i is assumed constant over the whole area of a surface.
Thus two simulation tools were developed for this study using coupled airflow and thermal models: HYBCELL1.0 and one implemented in the object-oriented solver SPARK. We propose here to compare results of simulations performed with the two different tools and to see how reliable these results can be use in order to conclude about the potential of a hybrid ventilation system compared to more traditional ones.
2.1.1. Heat transfer through enclosure structures Explicit finite difference method has been used to calculate the wall surface temperature. Each surface j is considered as one-dimensional transient heat conduction problem and is discretized into sub-layers of equal thickness (see Fig. 1), with temperatures being calculated at each time step. For sub-layers in contact with interior or exterior air, the following equations apply:
2. Description of the tools 2.1. HYBCELL1.0 HYBCELL1.0 (El Mankibi, 2003) has been developed at LASH/ENTPE. Simulations were carried out by coupling a thermal model based on finite differences and a pressure air flow model using the onion approach (Hensen, 1999). This tool has been developed under the Mathworks MATLAB/SIMULINK environment, and several control strategies for hybrid ventilation based on CO2 and temperature were implemented in the model. Schedule and occupation patterns were also taken into account in the model. Indoor air temperature is calculated from various heat transfers phenomena such as heat transfer through the walls and other enclosure structures, air infiltration and ventilation, internal heat gains and auxiliary heating or cooling. The air temperature Tint change in time dt is calculated as: dT int qas V cpas dtffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflffl
T tþdt nþ1 ¼
T tþdt ¼ 1
2 hint Dt T tint 2 k1 Dt T t2 þ q1 Dx1 c1 q1 ðDx1 Þ2 c1 2 hint Dt 2 k1 Dt 1 q1 Dx1 c1 q1 ðDx1 Þ2 c1
þ
2 Dt a ujSW int þ ujLW þ q1 Dx1 c1
Indoor air temperature change
¼
2 hext Dt T text 2 kn Dt T tn þ qn Dxn cn qn Dx2n cn 2 hext Dt 2 kn Dt þ 1 T tnþ1 qn Dxn cn qn Dxn cn 2 Dt a ujSW ext þ ujLW ext þ ð2Þ qn Dxn cn
X
S i hint ðT ts;i T tint Þ Surfaces |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
For interior sub-layer:
þ
T tþ1 ¼ i
Heat transfer through enclosures
UHeating þ Ugains |fflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflffl} Auxiliary heating or cooling and internal gains þ Qventil cpas T tair vent T tint |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl}
ð1Þ
ki Dt 2
Ventilation
ρi , λ i , c i
ρn, λn, cn Δxn
Exterior Tn+1
… Tn
Δxi Ti+1
ρ1, λ1, c1
… Ti
Interior
Δx1 T2
T1 (Tint, hint)
(Text, hext) Fig. 1. Enclosure discretization used in HYBCELL1.0 model.
ð3Þ
ki Dt
ϕ SW_int ϕ LW int
ϕSW_ext ϕ LW ext
T t1
int
qi ðDxi Þ2 ci ! 2 ki Dt 1 T ti 2 qi ðDxi Þ ci
qi ðDxi Þ ci T ti1 þ
T tiþ1 þ
!
ð4Þ
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
For interior surfaces of adjoining constituent wall layer: T itþDt ¼
ki1 Dt Dxi1 ðqi1 ci1 2 þ qi ci Dx2 i Þ Dxi1
þ
T ti1
ki Dt T tiþ1 ðqi1 ci1 Dx2i1 þ qi ci Dx2 i Þ Dxi
1
ki1 Dt Dxi1 ðqi1 ci1 2 þ qi ci Dx2 i Þ Dxi1
! ki Dt þ T ti ðqi1 ci1 Dx2i1 þ qi ci Dx2 i Þ Dxi with h ci ki qi uSW uSL ai Dt T ti Dx
ð5Þ
convection coefficient (hint = 4) for inside and hext = 15 for outside (W m2 K1) specific heat of material in layer i (J kg1 K1) thermal conductivity of material in layer i (W m1 K1) density of material in layer i (kg m3) short-wave radiative heat (W m2) long-wave radiative heat (W m2) absorbance () step time (s) node i temperature at time t (C) sub-layers thickness (m)
The stability of the solution requires a` step time value so that the expressions (6) and (7) remain true. Therefore time interval depends on enclosure construction materials. The time step chosen for this problem is 1 min. Stability conditions: 2
Dt 6
ðDxÞ q c 2k
Dt 6
q ðDxÞ c 2 ðh Dx þ kÞ
for internal sub-layers
911
To compute the internal distribution of the shortwave radiation used in HYBCELL1.0, multiple reflections, direct retransmission and reflection to the outside has been taken into account (Pfrommer, 1995). Reflection has been assumed to be diffuse on the interior surfaces. Internal net long-wave radiation exchanges have been carried out using a radiosity method (Sacadura, 1980) and view factors are based on an area weighting technique as following: 8 S > PN j if j 62 Xi > > Sk > > < k62Xi if j 2 Xi F i;j ¼ 0 > > Xi ¼ set of facets located > > > : on the wall containing S i ð8Þ Radiative heat exchange of the fac¸ade (slop = b) with the outside environment (at the air temperature) and with the sky vault at the temperature Tv is calculated according the following view factors: 1 þ cosðbÞ 1 cosðbÞ and F Env ¼ ð9Þ 2 2 Thus, the following relation has been used in HYBCELL1.0
F vault ¼
uiLW ¼ ei r0 F vault ðT 4v T 4s;i Þ þ F Env ðT 4ext T 4s;i Þ ð10Þ where r0 ei
Stefan–Boltzman constant (W m2 K4) emissivity ()
ð6Þ
Sky vault temperature has been computed using esp-r (clarck) model:
ð7Þ
T v ¼ 9:365574 106 ð1 ccÞ T 6ext 4 þ T ext cc ð1 0:84 ccÞ 0:527 þ 0:84 cc
2
for boundary sub-layers
2.1.2. Radiative heat exchange The short-wave radiation calculation is based on a sun patch model (Clarcke, 1985). This model includes the calculation of shades generated by the surrounding masks at the fac¸ade, the determination of the sunny parts of glazed areas and their projection on the interior surfaces following the direction of the solar beam.
0:25 273 þ 0:161 exp 8:45 1 T ext
ð11Þ
where cc (cloud and cover) (Dumortier, 1995) is a nebulosity indicia. 2.1.3. Air flow model HYBCELL1.0s airflow model is based on the flow network approach (nodal method) (Allard,
912
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
1998) where each zone of a building is represented by a pressure node. Boundary nodes are also used to represent the environment outside the building. Those nodes are then interconnected by flow paths, such as crack or windows. Application of mass balance equation on a zone i with j flow paths gives: 0¼
N X
qi Qik
ð12Þ
k¼1
where Qik qi
volumetric flow from zone i to zone k (m3 s1) air density in the direction of the flow (kg m3)
Air flow is attributed to pressure differences across the openings taking into account the air motion due to the wind and the temperature difference across the opening resulting to buoyancy driven flow. The wind pressure at the external node is calculated from Bernoullis equation as follows: Pw ¼
1 C p q v2 2
where Cp v
ð13Þ
pressure coefficient () wind speed at the height of the building (m s1)
The reference pressure on the ground being Po,i and according to hydrostatic gradient, the pressure due to stack effect only at a height z, is: P z;i ¼ P o;i qi g z
ð14Þ
The pressure difference across an exterior opening at a height z is then equal to: DP z;i ¼ P o;i P o;j g ðqi qa Þ z P w
ð15Þ
The air flow through large opening is assumed to be bi-directional. Thus, a level of zero pressure level (neutral level: DPz, i = 0) is located at height ZNL from the ground. The bi-directional flow is then calculated by integrating (15) between bottom or top of the opening and the neutral level: sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DP z dz 2 QBelow ¼ C d W q Zb |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Z
Z NL
Air flow below neutral level
ð16Þ
and sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi DP z dz 2 QAbove ¼ C d W q Z NL |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Z
Zs
ð17Þ
Air flow above neutral level
Newton–Raphsons method is then used to solve the system obtained by applying Eq. (12) where Cd W Zb Zs ZNL
discharge coefficient (m3 s1 Pa2) opening width (m) bottom of the opening (defined with reference to the floor) (m) top of the opening (defined with reference to the floor) (m) neutral level (m)
2.2. The simulation tool using the solver SPARK The object-oriented solver SPARK is used at LEPTAB laboratory (France) to compute problems with airflow and thermal model equations. SPARK is a general solver, thus the equations for the room, the outdoor and indoor environments, the occupancy pattern, the ventilation system and the control strategies are all implemented in SPARK. Some models that have been written for a previous Annex 35 study (Cron et al., 2002) are also used here. SPARK is robust and able to compute an entire problem with differential and non-linear equations, so airflow and thermal equations are solved simultaneously at each time step (1 min). 2.2.1. Models describing the air within the zone The model developed using SPARK considers a single zone model with an assumed hydrostatic pressure variation. This modeling is based on the ‘‘pure’’ air and the pollutant mass balance equations. For a zone i with n openings: n n X X qmasji qmasij ¼ 0 ð18Þ j¼1
j¼1
and n n X X dq qmesji qmesij þ S es ¼ V es dt j¼1 j¼1
ð19Þ
qmasji (kgas s1) is the pure air mass flow rate from the zone j to the zone i and likewise for the CO2 mass flow rate qmesji (kges s1). Ses is the pollutant source (kges s1) and V the volume (m3).
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
The thermal balance is n X
of accumulation taken into account according to ROUX (Rumianowsky et al., 1989). In that case, we had R ¼ R4tot and C = Ctot, and M was placed in the middle of the partition.
qmasji cpas þ qmesji cpes T tz;j
j¼1
n X
913
qmasij cpas þ qmesij cpes T tz;i
2.2.3. Convective heat transfer The convective heat transfer was described by uconv = h Æ (T Ts,i) where Ts,i is the surface temperature in (C). b For an indoor surface, hint ¼ ajT tint T ts;i j , a and b given by ALLARD (Allard, 1987). For an outdoor surface hext = c + dvn where c = 2.5, d = 3.5, n = 1 and v is the wind velocity in (m s1) according to (Ferries, 1984).
j¼1
þ S es cpes ðT es T tz;i Þ þ Uheating þ UGain þ Uconv @T z;i @q þ cpes V T tz;i es ¼ qas cpas þ qes cpes V @t @t dT z;i dqes t þ cpes V T z;i ¼ qas cpas þ qes cpes V dt dt ð20Þ For closure, the ideal gas is used: P ¼ ðqas ras þ qes res Þ ðT þ 273:15Þ
2.2.4. Radiative heat exchange The long-wave radiation model used for the indoor radiative heat exchange was a mean radiative temperature model. The following relations were used for a unit of surface:
ð21Þ
2.2.2. Conductive heat transfer model For the walls, the conductive heat transfer was described by an electrical 2R-3C model (see Fig. 2) to have a good response to an indoor high frequency excitation (Rumianowsky et al., 1989). We made a comparison between the response of this model and the one given by a finite difference model. The position of M (see Fig. 2) was set to the optimum value of eA that provided the minimum error of EeA. EeA is given by Z 1 ð22Þ EeA ¼ jT A finite difference T A j dt
uiLW
¼ hrmi ðT trm T ts;i Þ t 3 T s;i þ T trm with hrmi ¼ 4r0 ei þ 273:15 2 and
P hrmi S i T ts;i þ Urad þ ni¼1 uiLW int S i T ts;i Pn i¼1 hrmi S i
layer j Outdoor
Ti (t)
TM (t) Tj-1 (t)
Tj (t) TB (t)
e A , nA
CA
eB , nB TM (t)
TA (t) RA
ð24Þ
The direct short-wave radiation that is transmitted through the window was supposed to be entirely incident on the floor surface. One part is absorbed and the other part is supposed to be reflected in a diffuse way. Both the entering diffuse solar radiation and the one reflected on the floor were supposed to
layer i
Ti-1(t)
i¼1
ð23Þ
ð25Þ
The simulated room was surrounded by rooms with the same conditions, so the partitions were adiabatic and considered symmetrical. The electrical model used was a 1R-1C electrical model, to have the effect
TA (t)
Pn
T trm ¼
0
Indoor
int
CM
TB(t) RB
Fig. 2. Electrical model used in SPARK model.
CB
914
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
be distributed all over the surfaces depending on their area ratio and absorbed (for the window, one fraction was transmitted outdoors again). The outdoor long-wave radiation per surface for an outdoor surface i is given by 1 4 1 4 i 4 T þ T T s;i uLW ¼ ei r0 ð26Þ 2 ext 2 v
numerical models. HYBCELL is implemented within a large hall where temperature can be controlled to create an artificial climate around the experimen-
Ts,i is the outdoor surface temperature, Text the outside ambient air temperature and Tv the sky vault temperature. All temperatures are in Kelvin. The short-wave solar radiation absorbed by the outdoor wall surface was calculated given the surface incident solar radiation and the solar properties. 3. Comparison between experimental and numerical results Experimental data provided by the HYBCELL test cell, designed at LASH/ENTPE laboratory (El Mankibi et al., 2001), has been used to adjust the
Fig. 4. Internal view of HYBCELL test cell.
Fig. 3. HYBCELL test cell architecture.
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
915
Fig. 5. Location of sensors and actuators inside HYBCELL test cell.
tal cell. The test cell has only one external wall that is sloped and has six sash windows power-driven by step by step motors. The cell is 5.1 m long, 3.25 m wide and 2.95 m high (see Figs. 3 and 4). The cell is equipped with an electric heater, a fan, window engines, CO2 generation and sensible heat supply devices. Various sensors (temperature, pressure, relative humidity, CO2 concentration, VOC) have been installed in the test cell as well as in the hall and outdoors. In addition, wind velocity, wind direction and solar radiation data are provided by a meteorological station located near the cell. The CO2 generation and sensible heat supply devices simulate occupancy in the experimental cell. The data acquisition system is made up from a DAQ card and is supervised by a computer. The following figure gives the location of sensors and actuators in the test cell (see Fig. 5). Experimental monitoring has been held in for seven days in summer (from 19 to 25 June 2001) and seven days in winter (from 25 to 31 December 2002). Fig. 6 shows the outdoor boundary conditions in terms of air temperature and solar radiation. Results indicate that SPARK and HYBCELL1.0 agree well in modeling indoor air temperature (see Fig. 7). The small differences between the tools are mostly due to the different models used, such as the convec-
tion coefficients and the way short-wave and longwave radiation exchanges have been modeled. In addition, a comparison made between HYBCELL1.0, SPARK and coupled COMIS and TRNSYS models in the framework of HYBVENT project (Heiselberg, 2000) has shown a good agreement. HYBCELL1.0 and SPARK have been chosen to lead this study thanks to the simplicity of control strategies implementation that they provide. 4. Simulated test room A fictive classroom assumed to be located in Copenhagen has been modeled to test various control strategies. This room is a single zone at the middle floor of a three-storey building, is 9 m long, 6 m wide and 3 m high (see Fig. 8) and has a large glazed facade oriented South. The classroom is equipped either with a natural ventilation system (two inlet grilles and an exhaust chimney) or with mechanical ventilation systems (fans with or without heat recovery). The control strategies are based on indoor air temperature and CO2 concentration. The window is a line of 3/12/4 argon-filled lowemissivity double glazing (U-value of 1.5 W m2 K1) along the whole facade and is 1 m above the floor. The overall transmittance and absorptance
916
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Fig. 6. Outdoor conditions.
for each pane is a function of the solar radiation incident angle. The Table 1 summarizes the characteristics of the envelope and materials. The classroom is occupied by 24 children and a teacher from Monday to Friday between 8:00 and 12:00 and from 13:00 to 15:00. On Tuesdays between 10:00 and 11:00 only half of the pupils are
in the room and from 14:00 to 15:00 everybody is out. On Wednesday afternoons everybody is out. Internal loads related to occupants and lighting were taken into account: • occupant: 80 W per person and 0.018 m3 h1 of CO2 per person, • lighting: 10 W per m2 of floor area.
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
917
Fig. 7. Measured and simulated indoor air temperatures.
Outdoor CO2 was assumed to be constant and equal to 400 ppm. 5. Ventilation system and control strategies A 0.2 ach permanent infiltration rate is assumed, independently of the weather conditions and the ventilation system. For each simulation, a solar shading coefficient equal to 0.2 is used in spring and summer when global solar radiation is higher than 200 W m2.
Two reference mechanical systems and one hybrid ventilation system were compared. The classroom is equipped with two inlet grilles and an exhaust chimney with a fan (see Fig. 9). The grilles are positioned under the windows, 0.5 m above the floor level. Each grille is 1.8 m wide and is closed outside normal school hours except when night cooling is on. The exhaust chimney is 4.0 m high and is a simple duct with a fan. The top of the chimney is 10 m above the ground.
918
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
1m
Window
Chimney with fan 3m
1m 1m
4m
Vertical section
Facade
6m
2.5 m
9m
Grilles
3m
Horizontal section Fig. 8. Architecture of Copenhagen test room.
The cp-values of the facade, given in Table 2, are from AIVC (length to width ratio: 2:1 and building surrounded by obstructions equivalent to half the height of the building) (Liddament, 1996). The roof (chimney top) has a cp-value independent of wind direction and equal to 0.6. The airflow rates through the grilles and the chimney are given by a power law relation: Q = C Æ (DP)0.5, with • each inlet grille: C = 0.053 m3 s1 Pa0.5, • the exhaust chimney: C = 0.088 m3 s1 Pa0.5.
3m
Fig. 9. Copenhagen test room with the ventilation system.
15:00). The low fan power consumption is 1 W per m3 s1. The supply air is preheated to 18 C in winter and spring when the external temperature is below 18 C.
5.1. Reference ventilation system 1: mechanical exhaust
5.2. Reference ventilation system 2: balanced mechanical ventilation
A constant supply airflow rate of 0.150 m3 s1 is provided during all normal school hours (8:00 to
The mechanical supply and exhaust air flows are 0.150 m3 s1 with a fan consumption of 2.5 W per
Table 1 Characteristics of Copenhagen test room envelope Material from inside
Thickness (m)
Conductivity (W K1 m1)
Density (kg m3)
Floor and ceiling
PVC Massive concrete
0.005 0.15
0.35 1.7
1300 2300
External wall
Massive concrete Mineral wool Brick
0.1 0.15 0.11
1.7 0.04 0.8
Partition walls
Plaster board Mineral wool Plaster board
0.013 0.05 0.013
0.2 0.04 0.2
Specific heat (J kg1 K1)
Solar absorptance ()
Emissivity ()
960 920
0.5 0.5
0.9 0.9
2300 30 180
920 840 840
0.5 – 0.7
0.9 – 0.9
800 30 800
1100 840 1100
0.5 – 0.5
0.9 – 0.9
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926 Table 2 Pressure coefficients used in the simulations Wind angle
0
45
90
135
180
cp-value
0.25
0.06
0.35
0.60
0.50
m3 s1. A heat recovery with a temperature efficiency of 60% is available and works in spring and winter. The supply air is preheated up to 18 C using a heating coil. The heat recovery is continuously controlled to get a supply air temperature of 18 C and to turn off for outdoor temperatures higher than 18 C. This system is on during all normal school hours (8:00 to 15:00).
919
perature goes below 21 C. The air change rate when the window is open does not depend on outdoor conditions: it is a constant value of 4 ach additional to the ventilation system airflow rate. The temperature is also controlled using a heating system in winter and spring. Heating hours are 7.00–15.00 from Monday to Friday. The heating controller is a proportional controller based on air temperature with two set points: • heating hours: 21 C, • non-heating hours: 18 C. The maximum heating power is 5 kW.
5.3. Hybrid ventilation system
5.6. Night cooling
The specific hybrid system simulated here is a fan-assisted stack exhaust with preheating of supply air at 18 C in winter and spring. The inlet grilles and exhaust chimney are the same as in reference ventilation system 1, but a low-pressure fan was considered.
Night cooling is provided only in summer and is scheduled between 22:00 and 07:00. It turns on when the indoor air temperature is higher than 24 C and when the indoor–outdoor air temperature difference is higher than 2 C. It turns off whenever indoor air temperature is below 18 C. For the reference mechanical systems, night cooling works with grilles opened and fans on. For the specific hybrid ventilation system the night cooling is provided by mechanical ventilation (the grilles open and the fan turned on).
5.4. CO2 control The mechanical systems operate with constant airflow rates during normal school hours and do not depend on the CO2 concentration. Thus there is no CO2 control when simulating the two reference systems. For the hybrid system, the designed control strategy is an on–off control strategy based on CO2 concentration: • the first inlet grille opens if the CO2 is higher than 800 ppm, • the second inlet grille opens if the CO2 is higher than 1000 ppm, • the fan in the exhaust chimney is turned on if the CO2 is higher than 1200 ppm, • the fan turns off if the CO2 is lower than 1100 ppm, • the second inlet grille close if the CO2 is lower than 900 ppm, • the first inlet grille close if the CO2 is lower than 700 ppm. 5.5. Temperature control The window is open when both indoor and outdoor air temperatures are higher than 23 C and 12 C respectively. It is closed when the indoor tem-
6. Results and discussion Three periods of three weeks were simulated: from the 1st to the 21st of January (winter), from the 2nd to the 22nd of April (spring) and from the 21st of May to the 10th of June (summer). Figs. 10 and 11 show outdoor conditions (temperature and solar radiation) for the last week of each period. The three weeks of each season were simulated, but only the last one was used for result discussion. The first two weeks served to eliminate the effects of different assumptions of the starting conditions. 6.1. Winter results For all three ventilation system predictions, both simulation tools give equivalent results in terms of indoor air temperatures and CO2 concentrations. As an example, Fig. 12 shows the indoor air temperature and the CO2 concentration plots for the specific hybrid ventilation system. Both HYBCELL1.0 and SPARK predict about the same
920
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Fig. 10. Copenhagen outdoor air temperature for the last week of each simulation period.
Fig. 11. Copenhagen solar radiation for the last week of each simulation period.
behavior for the classroom in terms of indoor air temperature. However, a difference around 0.5 C can be noticed. This is due to differences in modeling the envelope of the building.
The CO2 concentration results are somehow different during occupancy for the 1st day (see Fig. 12). In fact, the differences are mostly due to the airflow: a variation around 15% of the
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
921
Fig. 12. Air temperature and CO2 concentration for the hybrid ventilation system in winter.
mass air flow rate can lead to a CO2 difference of 100 ppm. This mass air flow rate can vary between the two models because of the way the dynamic wind pressure has been taken into account (the coefficients of correction of wind velocity were different: 0.47 for HYBCELL1.0 and variable coefficient for SPARK) and also because of the air density difference due to differences in temperature.
The influence of the airflow on the CO2 concentration is also shown during the 5th day. Indeed, the dead band of 100 (ppm) for the second inlet grille control seems to be a little bit tight, since HYBCELL1.0 results present some oscillations. The difference between the CO2 decrease at the end of the 1st, the 4th and the 5th days is due to the way the grilles close. With SPARK they close
922
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
at 15:00 while the kids are leaving the classroom, whereas, with HYBCELL1.0, the grilles close right after everybody is out. HYBCELL1.0 predicts lower CO2 concentration and higher maximum indoor air temperatures than SPARK (see Tables 3 and 4). The comparison between results obtained with the reference mechanical ventilation systems and the hybrid ventilation system shows that the indoor air temperature is about the same for the three cases, but hybrid ventilation has lower mean and maximum CO2 concentration when the pupils are in the classroom. Hybrid ventilation appears to provide better indoor air quality in terms of CO2 concentrations. HYBCELL1.0 gives higher air flow rates with the hybrid ventilation system than SPARK. This explains the lower CO2 concentration as well as the higher preheating consumption obtained with HYBCELL1.0. Differences in heating consumption are first due to higher heating needs for HYBCELL1.0 in the morning when the set point temper-
Table 3 Mean and maximum indoor air temperatures for the three ventilation systems in winter Model
Mean air temperatures during occupancy (C)
Maximum air temperatures during occupancy (C)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
SPARK HYBCELL1.0
21.50 21.58
21.50 21.58
21.30 21.35
22.60 23.48
22.60 23.48
22.30 23.66
Table 4 Mean and maximum CO2 concentrations for the three ventilation systems in winter Model
SPARK HYBCELL1.0
Mean CO2 concentrations during occupancy (ppm)
Maximum CO2 concentrations during occupancy (ppm)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
1105 1051
1105 1051
958 902
1193 1137
1193 1137
1062 1048
ature changes to 21 C at 7:00. The differences between radiative heat transfer models explain also the heating need differences. Although there are differences in predicting heating and preheating consumption for the hybrid ventilation system, both HYBCELL1.0 and SPARK give higher room heating and preheating needs with hybrid ventilation than with the reference ones. There is no consumption for the fan, the stack effect is important to ensure enough airflow rates. The best results in terms of energy consumption are the ones obtained with the reference system 2 (see Table 5). 6.2. Spring results As for the winter results, there is still a small difference here for indoor air temperature and for CO2 concentration, see Fig. 13. Solar radiation is more important, so there are more internal gains, and for both tools the indoor air temperature is higher than the set point. HYBCELL1.0 again presents more airflow rates than SPARK and it leads to lower CO2 concentrations. Despite higher airflow rates, the indoor air temperature predicted by HYBCELL1.0 is slightly higher than the one for SPARK and without additional heating. This is due to the differences between the radiative heat transfer models. In spring, Tables 6 and 7 show again that the mean and maximum CO2 concentrations are lower with the hybrid system than with the mechanical ones. Hybrid ventilation is also better in spring in terms of indoor air quality. CO2 concentration is still below 1200 ppm so here again the fan does not need to be turned on. For this season too, the hybrid ventilation system is still the most expensive system in terms of energy consumption. But, unlike the winter period results, here the tools predict more significant differences in heating and preheating consumption (see Table 8). This can be explained by the higher airflow rates for HYBCELL1.0 than for SPARK concerning the hybrid mode.
Table 5 Energy consumption in winter Model
SPARK HYBCELL1.0
Heating (kW h/week)
Pre-heating (kW h/week)
Fan power (kW h/week)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
21.39 32.64
21.39 32.64
21.43 34.22
119.00 128.69
34.85 37.11
125.69 185.00
5.26 5.25
13.16 13.13
0.00 0.00
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
923
Fig. 13. Air temperature and CO2 concentration for the hybrid ventilation system in spring.
Table 6 Mean and maximum indoor air temperatures for the three ventilation systems in spring Model
SPARK HYBCELL1.0
Table 7 Mean and maximum CO2 concentrations for the three ventilation systems in spring
Mean air temperatures during occupancy (C)
Maximum air temperatures during occupancy (C)
Model
Mean CO2 concentrations during occupancy (ppm)
Maximum CO2 concentrations during occupancy (ppm)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
22.30 22.21
22.30 22.21
21.80 21.70
23.00 23.12
23.00 23.12
23.00 22.81
1107 1071
1107 1071
975 884
1194 1154
1194 1154
1148 1053
SPARK HYBCELL1.0
924
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
Table 8 Energy consumption in spring Model
SPARK HYBCELL1.0
Heating (kW h/week)
Pre-heating (kW h/week)
Fan power (kW h/week)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
8.57 6.70
8.57 6.70
6.03 8.45
58.21 60.92
8.21 8.59
61.23 94.24
5.26 5.25
13.16 13.13
0.00 0.00
Fig. 14. Air temperature and CO2 concentration for the hybrid ventilation system in summer.
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
925
Table 9 Mean and maximum indoor air temperatures for the three ventilation systems in summer Model
Mean air temperatures during occupancy (C)
Maximum air temperatures during occupancy (C)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
SPARK HYBCELL1.0
24.40 24.37
24.40 24.37
24.90 24.54
26.60 26.69
26.60 26.69
27.00 26.88
Table 10 Mean and maximum CO2 concentrations for the three ventilation systems in summer Model
Mean CO2 concentrations during occupancy (ppm)
Maximum CO2 concentrations during occupancy (ppm)
Ref 1
Ref 2
HV
Ref 1
Ref 2
HV
SPARK HYBCELL1.0
754 755
754 755
843 811
929 1158
929 1158
925 1081
Thus, for this season, these differences mean quantitative results have to be handled with care before giving any conclusion on the ventilation system performance. 6.3. Summer results Fig. 14 gives the evolutions of indoor air temperature and CO2 concentration for the hybrid system in summer. During occupancy, HYBCELL1.0 and SPARK give again quite similar results except for the night cooling. The difference occurs during the last night when the indoor air temperature given by HYBCELL1.0, unlike the one for SPARK, is not high enough to make the night cooling work. The CO2 concentration raises quickly the first day for the HYBCELL1.0 simulations. In fact, SPARK predicts an indoor temperature higher than 23 C, so the window opens whereas HYBCELL1.0 has still the window closed and thus lower airflow rate and a higher CO2 concentration. For both simulation tools Tables 9 and 10 show that the mean and maximum indoor air temperaTable 11 Energy consumption in summer Model
Fan power (kW h/week) Ref 1
Ref 2
HV
SPARK HYBCELL1.0
9.47 9.30
23.66 23.24
0.92 0.79
tures are now slightly higher for the hybrid system than for the mechanical ones. Here the mean CO2 concentration is higher with hybrid ventilation system than with the mechanical modes, but stays below 1000 ppm. In fact, as the window is open most of the time, the mechanical ventilation works with the fan on and the window open, so there is more airflow rate than with the hybrid ventilation. However, the maximum value of CO2 concentration remains the lowest one and some exposure to high CO2 concentration is avoided. Here again, this specific hybrid ventilation is better in terms of indoor air quality regarding to CO2 concentration. Since there is neither heating nor preheating energy, the results of this period indicate that the hybrid ventilation system is more interesting than the two reference systems (see Table 11). The difference for the fan consumption is simply due to the night cooling that was on more often for the SPARK simulations. 7. Conclusions Through a cross-simulation study, this paper has reported upon both the comparison of three ventilation systems performances and the ability of two numerical models (HYBCELL1.0 and SPARK) to predict similar behavior of the simulated test room for several control strategies. According to the first point of view, the simulation results show that
926
M. El Mankibi et al. / Solar Energy 80 (2006) 908–926
• In winter, hybrid ventilation system has higher energy consumption but there is a significantly lower mean and maximum CO2 concentrations compared to the mechanical system ones. There is a better indoor air quality for the hybrid ventilation system with equivalent thermal comfort in terms of indoor air temperature for the three ventilation systems. • In spring, hybrid ventilation system still has higher energy consumption with better indoor air quality and thermal comfort. However, differences in energy consumption between the three ventilation systems are lower than in winter. • In summer, hybrid ventilation energy consumption is lower than reference systems ones but the CO2 concentrations are a little bit higher. On the other hand, tools were in agreement according to CO2 concentration and indoor air temperature and the relative energy consumption performances of the three ventilation systems. The noticed disagreement in energy consumption between HYBCELL1.0 and SPARK is due to differences in modeling long-wave and short-wave radiation and convection heat transfer coefficients. According to this study results and to the noticed agreement between used models, simulation tools can be used to predict the relative performances of ventilation systems. However, particular care must be taken to avoid wrong interpretations linked to energy consumption. Indeed, ON–OFF controller can lead to differences between tools. Thus, advanced control strategies (PID or fuzzy controller) must be tested. Acknowledgements This work has been carried out within the frame of the Annex 35 ‘‘hybrid Ventilation in New and Retrofitted Buildings’’ and was supported by the French environmental and energy agency ADEME (Agence De lEnvironnement et de la Maıˆtrise de lEnergie), by the French electricity company EDF (Electricite´ de France) and by the French Ministry. The co-author thanks also Ashok Gadgil and Dimitri Curtil from the LBNL (Lawrence Berkeley National Laboratory) for their help.
References Aggerholm, A., 2002. Hybrid ventilation principle and control strategies. In: Proceedings of the 4th International Forum on Hybrid Ventilation, Montreal, Canada, 14–15 May 2002. Allard, F., 1987. Contribution a` le´tude des transferts de chaleur dans les cavite´s thermiquement entraıˆne´es a` grand nombre de Rayleigh, The`se dEtat, INSA de Lyon. Allard, F., (Cord), 1998. Natural ventilation in building. James & James Ltd., London, p. 356. Clarcke, J.A., 1985. Energy Simulation in Building Design. Adam Hilger Ltd, Bristol and Boston, p. 388. Cron, F., El Mankibi, M., Inard, C., Michel, P., 2002. Experimental and numerical analysis of a hybrid-ventilated room. In: 8th International Conference on Air Distribution in Rooms, Copenhagen, Denmark, 8–11 September 2002. Dumortier, D., 1995. Mesure, Analyse et Mode´lisation du gisement lumineux. Application a` le´valuation des performances de le´clairage naturel des baˆtiments, Doctorat: Universite´ de Savoie, p. 292. El Mankibi, M., Michel, P., Guarracino, G., 2001. Control Strategies for hybrid ventilation development of an experimental device. In: Proceedings of the 22nd Annual AIVC Conference, Bath, UK, 11–14 September 2001. El Mankibi, M., 2003. De´veloppement et e´valuation nume´rique et expe´rimentale Des strate´gies de re´gulation de la ventilation hybride, The`se de doctorat. INSA Lyon, p. 384. Ferries, B., 1984. Contribution a` le´tude des enveloppes climatiques et aide a` leur conception par micro-ordinateur, The`se Doct.-Ing., INSA de Lyon. Frascatoro, G.V., Perino, M., 2002. Natural vs. mechanical ventilation—a tool to help making a choice. International Journal of Ventilation 1 (2), 101–108. Heiselberg, P. (Ed.), 2000. Principles of hybrid Ventilation, IEA Energy Conservation in Buildings and Community Systems Program Annex 35: Hybrid Ventilation in New and Retrofitted Buildings. Hensen, J.L.M., 1999. A comparison of coupled and de-coupled solution for temperature and air flow in a building. ASHRAE Transactions 105 (2), 962–969. Jeong, Y., Haghighat, F., 2002. Modeling of a hybrid-ventilated building—using ESPr. International Journal of Ventilation 1 (2), 127–139. Li, Y., Delsante, A., 2001. Natural ventilation induced by combined wind and thermal forces. Buildings and Environment 36 (1), 59–71. Liddament, M.W., 1996. A guide to energy efficient ventilation AIVC, Coventry, UK. Pfrommer, P., 1995. Thermal modelling of highly glazed spaces. Ph.D: De Montfort University, Leicester, p. 229. Rumianowsky, P., Brau, J., Roux, J.J., 1989. An adapted model for simulation of interaction between a wall and the building heating system. In: Thermal performance of the exterior envelopes of buildings IV conference, Orlando, USA. Sacadura, J.F., 1980. Initiation aux transferts thermiques, Techniques et Documentation, CAST INSA, Lyon.